«65:5... 1.12.4. A 55337 :5... r91? 1...... 19.55:??? Scyflllfldlll!‘ l ..k€e.€:._(.t:r.lu£r§-I . irks-lg; 1:;4. ,..;v..frrt¥lrtl.}9¢.vln¢r}§l$§ in t . r rel) . attaiif §§§10rlftillttrv .l. : r. arrck. rte: i. .,. tart! .flllt , :1) 1:52.}. {(51.5}: a gar/0% 'Hii’i "M ' 293 00753 8352 may} Liam} 7/ Michigan State 1 University This is to certify that the dissertation entitled Older Part- and Full-time Employees: Assessing Overall Job Satisfaction and its Relationship with Organizational Commitment and Withdrawal Intentions presented by Scott Alan Cohen has been accepted towards fulfillment of the requirements for Ph.D. degreein Psychology WM Major professor Date March 1, 1990 MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. ll DATE DUE DATE DUE DATE DUE 10%% l :"\ \ OLDER PART— AND FULL—TIME EMPLOYEES: ASSESSING OVERALL JOB SATISFACTION AND ITS RELATIONSHIP WITH ORGANIZATIONAL COMMITMENT AND WITHDRAWAL INTENTIONS 1 By Scott Alan Cohen A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1989 (0050547‘2. ABSTRACT OLDER PART- AND FULL-TIME EMPLOYEES: ASSESSING OVERALL JOB SATISFACTION AND ITS RELATIONSHIP HITH ORGANIZATIONAL COMMITMENT AND HITHDRANAL INTENTIONS By Scott Alan Cohen Declining birthrates, increasing life expectancy, and increasing rates of early retirement may all soon contribute to an increase in the number of older part-time employees in the workforce. However, no previous research has been conducted examining older part-timers’ attitudes toward work. The present study explored whether there were any significant differences in the levels and relationship between several job attitudes for older part- and full- timers employed in retail/sales and grocery stores located in the Midwest and Northeast. The findings suggested that self-reported measures of overall job satisfaction and organizational commitment were more strongly predictive of turnover intentions for older part-timers than older full-timers. While there was no difference in their reported levels of overall job satisfaction, the part—timers expressed significantly more organizational compliance than the full—timers. The best fitting causal order of the satisfaction, commitment, and turnover variables was dependent on the particular measure of commitment used in the analysis. Overall job satisfaction was significantly related to satisfaction with pay, satisfaction with the perceived meaningfulness of work, and satisfaction with the manner in which work fills up one’s time for both older part- and full-timers; satisfaction with supervision, satisfaction with Scott Alan Cohen promotional opportunities, and satisfaction with social interactions at work were not significantly related to overall job satisfaction when all six facets were simultaneously entered in a regression equation. Implications and future research directions are suggested, with an emphasis on the need to reconsider the use of paper-and-pencil measures of organizational commitment in future studies. For Susan You made the frustrations in graduate school a little less painful, and you make the joys of life a lot more meaningful... iv ACKNOWLEDGMENTS This may be one of the most difficult sections of this manuscript to write. I see it as an opportunity to thank those who have helped me during my graduate education, and not just those who helped me complete the process of working on this dissertation. I am grateful that I had many opportunities to work with and learn from Dr. Neal Schmitt. Neal has the knowledge, skills, energy, and initiative needed to take on numerous complex projects simultaneously and get them all completed within impossible time constraints. More importantly, he applies his expertise quite skillfully to everything he attempts. It’s no wonder that he is regarded so highly in the field, and respected by all who have come to know him. Thank you, Neal, for not only serving as Chairperson on my dissertation committee, but for giving me years of attention and knowledge that I will take with me on all future endeavors. Additional thanks are due to Drs. Steve Kozlowski, Dan Ilgen, and Michael Lindell, who have joined Neal in helping me progress through a Masters Thesis, Comprehensive Examination, and finally, this Dissertation. While there were many days when it felt like this time would never arrive, I am pleased to break down old distinctions between professors and graduate student and welcome Steve and Dan as my colleagues. I am grateful to the Lansing Senior Employment Service for allowing me to interview job applicants for this dissertation. The interviews were most useful in helping me form the hypotheses discussed in this manuscript. 1 am also appreciative of the many retail store managers who agreed to distribute my surveys to their employees so I could conduct the study described here. None of my educational achievements could have been possible without the love and support I received from my parents, brothers, and sisters. While the time spent in graduate school (and now work) have placed miles between us, I always look forward to the time we spend together and am always grateful to have each and every one of them in my family. Finally, my wife, Susan, has helped me shape who I am today. It has been very important to her to see that I finish this task, and I thank her for all the sacrifices she has made to help us achieve our goals. Her countless readings of drafts of the manuscript have helped me clarify my thoughts. Since I did not always follow her recommendations, any unclear passages are completely my own responsibility and should bear no reflection on Sue. My fondest memories of graduate school are the times I have shared with Sue, and I am most thankful to have entered Michigan State as a "me" and to be leaving as a "we". The pride I feel in completing this task is meaningful primarily because it is shared with my wife. Thank you for everything, Sue, and everything yet to come! vi TABLE OF CONTENTS LIST OF TABLES .......................... LIST OF FIGURES .......................... INTRODUCTION ........................... Why Older Part-Timers Have Not Been Previously Studied . . Availability of Part-Timers ................. Part-Timers’ Abilities and Skills .............. Reliability of Part-Timers ................. Demographic composition .................. Nature of the tasks .................... Summary .......................... Differentiating Between Older Part- and Full—Time Employees . . Nature of the Tasks ..................... Pay, Fringe Benefits, and Promotional Opportunities ..... Demographic Characteristics ................. Summary .......................... The Conceptual Framework .................... Organizational Commitment ................... Mean differences ..................... Relational differences .................. Intention to Withdraw ..................... Valued Outcomes for Part-Time, Full-Time, and Older Workers . . . Part- and Full-Time Employees ................ Summary .......................... Older and Younger Workers .................. Summary .......................... Work Role ............................ Psychological Importance of the Work Role ........... Self—Identity ........................ Social Interaction ..................... Allocation of Time and Leisure Pursuits ........... Summary .......................... An Exploratory Model ...................... Determinants of Overall Job Satisfaction ........... Organizational Commitment ................... Intention to Withdraw ..................... METHOD .............................. Sample and Sample Size Requirements ............. Hypotheses 1 - 2b ..................... Hypotheses 3a — 4c .................... Hypotheses 5a - 5b .................... Distribution of Questionnaires ............... Common Method Variance/Biased Results Issues ......... Selecting the Measures ................... Construction of the Questionnaire .............. Acquiescence ....................... Social desirability .................... Factor analysis, internal consistency reliability, and discriminant validity ................... Development of Instrument .................. Job satisfaction ..................... Organizational commitment ................. Withdrawal intention ................... Job descriptive information ................ Demographic information .................. Pretesting the Instrument .................. Data Analyses ......................... Reliability and Validity of Measures ............ Examination of Hypotheses .................. Descriptive statistics .................. Discriminant analysis ................... Determinants of overall job satisfaction ......... Mediation test of organizational commitment ........ Differential hypotheses regarding propensity to search for another job or retire ................. RESULTS .............................. Survey Respondents ..................... Factor Analyses and Scale Reliabilities ........... Scale Mean and Demographic Differences Between Part- and Full-Timers ......................... Scale Intercorrelations for Part- and Full-timers ...... Examination of Hypotheses .................. Hypothesis 1 ....................... Hypotheses 2a and 2b ................... Hypotheses 3a, 3b, and 3c ................. Hypothesis 4a ....................... Hypothesis 4b ....................... Hypothesis 4c ....................... Hypothesis 5a ....................... Hypothesis 5b ....................... DISCUSSION ............................ Summary of Results ....................... Determinants of Overall Job Satisfaction .......... Differences Between Part- and Full-timers’ Levels of Organizational Compliance .................. Relationship Between Overall Job Satisfaction, Organizational Commitment, and Turnover Intentions ..... Reconsidering what the OCQ measures ............ viii 86 97 103 110 111 114 114 125 134 145 152 152 Page Organizational compliance, identification, and internalization ...................... 171 Commitment (OCQ) Is a Relevant Construct for Older Part-Timers ......................... 175 Propensity to Search for Another Job or Retire ....... 176 Limitations of the Study .................... 178 Sample ........................... 178 Research Design ....................... 180 Implications and Future Directions ............... 182 APPENDICES ............................ 186 Appendix A ........................... 186 Interview Responses ...................... 186 Appendix B ........................... 190 Survey of Senior Employees .................. 190 General Instructions ................... 190 Section A ........................... 191 Facets of Job Satisfaction ................. 191 Satisfaction with Pay ................... 191 Satisfaction with Social Interactions at Work ....... 192 Satisfaction with the Manner in Which Work Fills Up One’s Time ........................... 192 Satisfaction with Supervision ............... 192 Satisfaction with Promotional Opportunities ........ 193 Satisfaction with the Type of Work I Do .......... 193 Overall Job Satisfaction ................. 194 Section B ........................... 195 Facets of Organizational Commitment ............. 195 Organizational Compliance ................. 195 Organizational Identification ............... 196 Organizational Internalization .............. 197 Organizational Commitment ................. 197 Section C ........................... 199 Withdrawal Intentions .................... 199 Intention to Turnover ................... 200 Intention to Stay Home from Work ............. 200 Propensity to Search for Another Job or Retire ...... 200 Section D ........................... 202 Job Descriptive Information ................. 202 Section E ........................... 207 Background Information ................... 207 BIBLIOGRAPHY ........................... 210 ix Table \0 on \l 03 01 #wNH H t—I H O O O 12. 13. 14. 15. 16. 17. 18. 19. LIST OF TABLES Respondent Demographic and Job Experience Levels ..... Rotated Factor Loadings for Part-Timers ......... Rotated Factor Loadings for Full-Timers ......... Items and Internal Consistency Reliabilities of Final Scales ...................... Means and Standard Deviations of Scales by Work Status (Part- or Full-Time) .............. Demographic and Job Descriptive Differences Between Part- and Full-Timers ................. Job Titles with the Dictionary of Occupational Title’s (1977) Data, People, and Things Codes ......... Scale Internal Consistency Reliabilities and Observed Intercorrelations for Part—Timers ........... Scale Internal Consistency Reliabilities and Observed Intercorrelations for Full—Timers ........... Observed Intercorrelations of All Variables for All Employees ....................... Regressing Overall Job Satisfaction on Control Variables, Facet Satisfaction Indices, and Work Status Moderator .................... Discriminant Function Analysis Predicting Work Status Using Organizational Compliance and Organizational Identification ..................... Discriminant Function Analysis Predicting Work Status Using Organizational Compliance and Organizational Internalization .................... Discriminant Function Analysis Predicting Work Status Using Organizational Compliance and Organizational Commitment (OCQ) .................... Regressing Organizational Compliance on Control Variables, Overall Job Satisfaction, and Work Status Moderator ....................... Regressing Intention to Turnover on Control Variables, Organizational Compliance, and Work Status Moderator . . . Regressing Intention to Turnover on Control Variables, Overall Job Satisfaction, and Work Status Moderator Exploring Whether Organizational Compliance Mediates the . . Relationship Between Overall Job Satisfaction and Intention to Turnover ................. Regressing Organizational Identification on Control Variables, Overall Job Satisfaction, and Work Status Moderator ....................... Page 85 90 95 98 100 102 104 105 106 112 116 118 120 126 128 129 Table 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. Regressing Organizational Internalization on Control Variables, Overall Job Satisfaction, and Work Status Moderator ........................ Regressing Intention to Turnover on Control Variables, Organizational Identification, and Work Status Moderator ........................ Regressing Intention to Turnover on Control Variables, Organizational Internalization, and Work Status Moderator ........................ Exploring Whether Organizational Identification Mediates the Relationship Between Overall Job Satisfaction and Intention to Turnover .................. Exploring Whether Organizational Internalization Mediates the Relationship Between Overall Job Satisfaction and Intention to Turnover .................. Regressing Organizational Commitment (OCQ) on Control Variables, Overall Job Satisfaction, and Work Status Moderator ........................ Regressing Intention to Turnover on Control Variables, Organizational Commitment (OCQ), and Work Status Moderator ........................ Exploring Whether Organizational Commitment (OCQ) Mediates the Relationship Between Overall Job Satisfaction and Intention to Turnover .......... Regressing Intention to Turnover on Control Variables, Organizational Compliance, and Propensity to Search for Another Job or Retire Moderator for Part—Timers Regressing Intention to Turnover on Control Variables, Organizational Identification, and Propensity to Search for Another Job or Retire Moderator for Full-Timers ....................... Regressing Intention to Turnover on Control Variables, Organizational Internalization, and Propensity to Search for Another Job or Retire Moderator for Full-Timers ....................... Regressing Intention to Turnover on Control Variables, Organizational Commitment (OCQ), and Propensity to Search for Another Job or Retire Moderator for Full-Timers ....................... xi Page 136 137 138 141 143 146 147 150 154 Figure 4an LIST OF FIGURES Components of Part-Time Employment (Adapted from Plewes, 1986) .................... General Conceptual Framework ............. The Exploratory Model ................. Regression Lines of Overall Job Satisfaction on Expected Turnover Intentions for Part- and Full-Timers ..................... Regression Lines of Organizational Identification on Expected Turnover Intentions for Part- and Full—Timers ..................... Regression Lines of Organizational Internalization on Expected Turnover Intentions for Part- and Full-Timers ..................... Regression Lines of the Organizational Commitment Questionnaire Score on Expected Turnover Intentions for Part- and Full-Timers ........ Graphical Summary of Results Obtained from Hypotheses 4a, 4b, and 4c .............. xii Page 6 17 46 130 139 140 153 INTRODUCTION Much organizational research has examined the conceptual relationships between job satisfaction, organizational commitment, and withdrawal intentions (e.g., Mobley, Horner, & Hollingsworth, 1978; Steers & Rhodes, 1978; Wanous, 1978). The theoretical frameworks underlying this research are based on studies of the needs, values, attitudes, and behaviors of regular full-time employees; part-time workers are often ignored (Rotchford & Roberts, 1982). Those who have conducted research on part— time employees (e.g., Gannon & Nothern, 1971; Hom, 1979; Logan, O’Reilly, & Roberts, 1973; Miller & Terborg, 1979; Wakefield, Curry, Mueller, & Price, 1987) have failed to make a distinction between the needs and values of different types of part-timers. This failure has often resulted from atheoretical research conducted on convenience samples. Needless to say, results are often uninterpretable and conflicting. A cursory examination of the needs and values of different types of part- time workers suggests that they should not be lumped into one group. For instance, teenagers, working mothers, moonlighters, and older individuals may all be working on a part-time basis for different reasons. The teenager may be interested in earning some pocket money to be spent on material goods and/or social activities which s/he desires. The working mother may enter the workforce on a part-time basis after her children are grown and the amount of time she needs to devote to household duties has subsided. Or, she may reduce her full-time working hours so she can be home when her children are not in school (Kahne, 1985; Leon & 2 Bednarzik, 1978). She may need to continue working at least part—time in order to supplement the family income or provide the whole income if her earnings are the primary source of family livelihood. The moonlighter may find that a second “night job" is necessary in order to meet his/her expenses, or is the only time available for exploring other career interests while still holding on to a primary job (Stinson, 1986). Finally, many older individuals may have chosen retirement or been forced into retirement, only to find that the reduction in income and/or social contacts, abandonment of the work role, or any combination of the above was not desirable (e.g., Friedmann & Havighurst, 1954; Jondrow, Brechling, & Marcus, 1983; Kahne, 1985). While it is not my intention to compare and contrast the needs of these different groups of part-timers in the present study, I would like to focus attention on one of these groups; the older part-time worker. An examination of all of these groups could be interesting, yet a study focusing on even just one type of part-time worker would be sufficient to determine the extent to which our current state of knowledge concerning conceptual relationships among full-timers’ job satisfaction, organizational commitment, and withdrawal intentions generalizes to a sample of part-time employees. A conceptual model of these relationships based on theoretical developments derived from previous studies of full- time employees will be presented. The strength of these relationships will be examined for older part- and full-time workers. These are individuals who are at least 55 years of age and work in the same type of occupation and organization (i.e., retail/sales) but vary on the average number of hours employed per week. The definition of part- and full- time employment adopted by the Bureau of Labor Statistics will be used to 3 distinguish between each group. A person who works equal to or less than 35 hours per week is a part-time employee; all others are full-time employees (Nardone, 1986). A comparison group of full-time workers will be selected because many of the theoretical frameworks and much of the research examining job attitudes such as satisfaction, commitment, and withdrawal intentions has focused exclusively on full-time employees. However, an appropriate comparison group to older part-timers is older full-timers, not just a group of full-timers of all ages. Older and younger employees may have different work attitudes largely due to psychosocial aging effects (Rhodes, 1983). Rhodes defines psychosocial aging as "systematic changes in personality, needs, expectations, and behavior as well as performance in a sequence of socially prescribed roles and accumulation of experiences" (p. 329). Super (1980) explains that these socially prescribed roles change as people become older (e.g., student, spouse, parent, etc.) and affect changes in peoples’ needs, expectations, and behaviors. Utilizing older full—timers as a comparison group to older part-timers would rule out these psychosocial aging effects as alternative explanations for any significant differences which may arise in older part- and full-timers’ job attitudes. While it would also be interesting to focus on differences between older and younger part-timers’ job attitudes, the primary goal of this study is to generalize theoretical frameworks and research findings derived from previous studies of full-timers to part-timers. Hence, data will only be collected from older part-time and older full-time employees in order to investigate four research issues: 1) whether there are differential determinants of overall job satisfaction for older part- and 4 full-time employees; 2) whether there are mean differences in the depth of older part- and full-time employees’ organizational commitment; 3) whether the depth of organizational commitment expressed by older part- and full-time employees differentially mediates the relationship between overall job satisfaction and withdrawal intentions; and 4) whether organizational commitment is differentially related to an older employee’s turnover intentions (i.e., based on whether s/he intends to turnover to find another job or whether s/he intends to turnover in order to completely retire). In the next section, the importance of studying older part-timers is discussed. This is followed by a discussion of the development of the older part- and full-time employee constructs utilized in the present study. Emphasis will be placed on distinguishing between the nature of part- and full-time jobs and the individuals who hold these jobs. Why Older Part-Timers Have Not Been Previously Studied Werther (1975) offered three generally accepted beliefs why part-time workers may be overlooked and undervalued in the labor market: 1) There are not many part-timers in the labor force. 2) Part-timers do not have the necessary abilities and skills to be competent workers. 3) Part-timers are unreliable employees. While Werther refers to these beliefs as "myths", he did not provide enough empirical evidence to show that they are not true. Some 1 literature suggests that these beliefs may be inaccurate. Although he addressed each of these beliefs in terms of part-time workers in general, the following discussion addresses these beliefs in terms of older part- timers. Availability of Part—Timers The Current Population Survey conducted by the Bureau of Labor Statistics surveys approximately 60,000 households nationwide. Information is provided on the employment status and related characteristics of the civilian population 16 years of age and older for the particular survey week. The official boundary between full- and part-time employment status has stood at 35 hours a week since 1947 (Nardone, 1986). Figure 1 describes how the Bureau of Labor Statistics classifies these part—time workers. First, they are divided into two groups based on their usual work status (i.e., full— or part— time). They are then subdivided into groups based on their reasons for working part-time during the particular survey week (i.e., involuntary— economic or voluntary—noneconomic). While involuntary part-time workers generally remain a low percentage of the total labor force of full—time and part—time workers, the proportion of all employees who work part-time has increased from roughly 1 in 6 workers in the late 19505 to I in 5 in 1977. Although, in absolute numbers, greater growth took place among full—time workers, the relative rate of growth was much higher for part-time than for full—time workers (140% vs. 50%) (Deutermann & Brown, 1978; Plewes, 1984). Much of this increase has been associated with women entering the labor force and baby—boomer students searching out part-time jobs during their teenage years. However, trends toward early full-time retirement (the proportion of Social Security beneficiaries who were early retirees rose from 16.3% in December, 1961 to 58.9% in September, 1977) together with higher Social Security earning ceilings have also contributed to increases in part-time work among older Americans. The Social Security E0: mama? .mwsfld Cowaoo Hmcomooc o:tocoom Ice: Lo xcoac:_o> .xoo; c_ Umeocttcow pom Lo Uofiofiwnofi Eoc .mtoaoc >EoEzoocc Lo Ecol .moootoEm _ocowoflu.xcoi x006 Hmcomooc 07c0cooo Lo speaczro>cw .mCOmomL meyO Lo .Etomg .mcsoL mm Love: osty 23L .xco; octw :zaacoa Ho: moon Uwcomooc o¢t0cooococ Lo >+ot::o> 3L0; octw tog bco Dc: ano icos 3005 umcomooc 07t0coom Lo Fayc20>c cos x 03m “Lou xLoE n< _ 7 program allows beneficiaries to retain a limited amount of their earnings without losing benefits. Therefore, part-time work is an economically feasible manner of employment for beneficiaries. It is not surprising then that among those age 65 or older, the proportion working part-time voluntarily increased from 38% in May, 1968 to 49% in May, 1977. In 1985, over 19 million Americans were employed in part-time work, with nearlyz/3 of the male part-timers between the ages 16-24 and over 65 (Nardone, 1986). The available supply of older part-time workers is expected to increase in the future. Declining birthrates, increasing life expectancy, and increasing rates of early retirement may all lead to an increased supply. This supply may also be in much demand as fewer young people enter the workforce and the current growth of service-sector jobs continues. Personick (1983) has suggested that I in 3 new jobs created between 1982 and 1995 will be in direct service areas such as medical care, personal services, and business and professional services. These are all areas that are expected to use a high percentage of part—time labor. Finally, as proportionately fewer young persons enter the job market, proportionately more people will be drawing pensions from the Social Security system than paying into it. It is my belief that this could threaten the future solvency of the system. While some older people may receive private pensions, it is my understanding that many of these pensions are not indexed for inflation. Therefore, they may not prove to be a substantial means of support after retirement. Hence, inadequate pensions, limited retained earnings from Social Security, and increased life expectancy may all be factors which force an increased portion of older people into work. Part-Timers’ Abilities and Skills The incentive to train part—timers may be quite low. Part—timers do not work as many hours as full-timers, so the relative cost of providing similar training to part-timers is higher. Further, the high supervisory, coordination, and communication costs of dividing complex, high—paying jobs into many more part-time positions can be prohibitive. For these reasons, part—timers are often relegated to lower, simpler jobs where they do not have much opportunity for training, upward mobility, or job security (Barrett, 1984; Owen, 1978; Plewes, 1984). Not all part-timers have to be in such a bleak situation. While teenagers may have only limited work experience, older part-time workers may have held one or more previous jobs where numerous skills were required and learned. Perhaps some of these skills could be applied to a part—time job. Furthermore, any additional training costs can be spread over the length of employment. While many teenagers may only be willing to work part—time on a job for a few years before they leave home to attend college, my impression is that many older individuals may be happy to continue working as long as their health allows. If some people who retire early from full-time jobs opt for part—time employment, they may have the potential of working fifteen or twenty additional years. Therefore, hiring and training some older part-timers may prove to be cost effective for employers. Reliability of Part—Timers According to Werther (1975), the thought of part-time workers often conjures up visions of unreliable teenagers. Often unskilled, these part-timers are relegated to menial jobs and equally menial wages. Since 9 their tasks and pay offer few attractions, since they are often responsible only to themselves (i.e., they are usually single and do not have children), and since their jobs are essentially dead end positions, unreliable behavior such as "calling in sick" too frequently can only be penalized by the loss of a meaningless job. Werther states that the "myth" of unreliability might stem from two problems. The first problem relates to the incorrect assumptions made about the demographic composition of the part-time labor force. The second problem relates to inappropriate attributions made about part- timers’ work-related attitudes and behaviors without recognizing the nature of the tasks embedded in many part—time jobs. Demographic Composition. While many part—time workers are teenagers, an increasing number of older people have been working part-time in recent years (Nardone, 1986). Between 1977 and 1985, the proportion of all part—time workers younger than age 25 decreased from 39% to 37% while the proportion of all part—time workers older than age 49 increased from 21% to 24% (Nardone, 1986; Owen, 1978). Although the changes in these proportions over time do not appear large, it is my belief that they may represent the beginning of a change in the demographic composition of the part-time labor force. This change may signal a need to alter our perceptions of part—time workers. Unlike teenagers, many older part— timers have work histories which provide evidence of their reliable behavior and acceptance of responsibility (Werther, 1975). Since many older part-timers may also depend on the income they earn through part— time employment for survival (e.g., Deutermann & Brown, 1978), it is my belief that many will not tend to exhibit unreliable behavior (e.g., unexplained absences). 10 Naturgiof the Tasks. It was previously stated that most part-timers hold mundane, noninvolving jobs (Deutermann & Brown, 1978). However, many older part-timers may actually be quite satisfied with their jobs despite the fact that the tasks they conduct tend to be routine and noninvolving. If older individuals who choose to work part-time do not desire complex jobs with many responsibilities, they may not place as much value on task complexity as older full-timers. Therefore, research may suggest that older part-timers can be quite satisfied with jobs comprised mostly of routine tasks. If, however, part-timers do value complex jobs then unreliable behavior exhibited by some part-timers may be a function of their reaction to their routine, noninvolving jobs. As Werther suggests, even mature full-time workers in mundane jobs are likely to give less than optimal performance because the job limitations are frustrating. Such frustration is dissipated through unexplained absences, horseplay, or other irresponsible behavior (Werther, 1975). In order to conclude that older part-timers have a tendency to exhibit more unreliable behavior than older full-timers, it is my understanding that the differential nature of the tasks they conduct must be recognized and controlled statistically. This will be discussed in the next section. Summa_y. The changing demographic structure of the labor force suggests that older part-timers may become a growing segment of the work force (e.g., Nardone, 1986). Unfortunately, what is known about the attitudes and behaviors of older part-time workers is not based on scientific study but on popular mythology (Werther, 1975). The present study will more rigorously examine and compare the relationships between job attitudes and withdrawal intentions for older part- and full-time 11 employees. Before these job attitudes are addressed, however, it is important to differentiate between older part- and full-time employees and the nature of the work they conduct. It is my understanding that this would be crucial in order to conclude that differences between older part- and full—timers’ job attitudes can be attributed to differences in the type of people attracted to part- and full-time positions and not just to the type of things they do on the job. The older part— and full- time employee constructs utilized in the present study are discussed and operationalized in the next section. Differentiating Between Older Part- and Full-Time Employees The Bureau of Labor Statistics makes a distinction between part- and full-time employment solely on the basis of the average number of hours an employee works per week. A person who is employed equal to or less than 35 hours per week is a part—time employee; all others are full-time employees (Nardone, 1986). This criterion will be used to distinguish between part— and full-time employees in the present study. Additionally, an older employee will be defined as an employed individual who is at least 55 years old. Fifty five has been selected since it seems reasonable to me that individuals may begin to contemplate early retirement from their full-time careers at this age and begin to consider seeking employment in part-time positions. Pension plans often mature or employers have "30 and out" plans for which many are eligible. This study will compare the nature of different job attitudes of people 55 years of age or older who are employed less than or equal to 35 hours per week with those employed more than 35 hours per week across several retail/sales organizations. This type of organization was selected because personal contacts with personnel managers have suggested 12 that many older part- and full-time employees can be found in this type of work setting. It is important that both types of employees be in the same type of organization so direct comparisons can be made between both groups. However, it is also important to recognize that the nature of the tasks they typically conduct, the pay, fringe benefits, and promotional opportunities they receive, and the demographic composition of both groups of employees often differ (Rotchford & Roberts, 1982). Each of these differences is described below. Nature of the Tasks Frease and Zawachi (1979) state that several organizations have instituted job sharing programs. Job sharing refers to a situation in which two or more equally qualified part-time employees divide the hours, responsibilities, and benefits of a full-time job by performing complementary tasks. The skills and education of job sharers are equal to those of full-time workers. While job sharing does not change the basic nature of the work, it does allow an organization to tap previously unavailable labor markets (Steers, 1984). Evidence of job sharing suggests that many full-time tasks can be conducted by part-timers, provided that they have adequate training. However, Deutermann and Brown (1978) reported ten years ago that the percentage of part-time workers was much higher in lower level than higher level job categories. Forty-one percent of people who normally work part-time were employed in sales and clerical positions, while only 3% were employed in more involving and complex managerial positions. If these results have not changed drastically over the past ten years, it is reasonable to assume that job sharing has not been applied to most complex full-time jobs. Part-timers typically fill the less involving 13 and more mundane jobs in many organizations. Therefore, in an assessment of the differences between part- and full—timers’ job attitudes, one also needs to assess differences in the nature of the job tasks they conduct, even if they are employed in the same type of occupation. Controlling the variance attributable to the nature of the job tasks will ensure that any attitudinal differences between both groups of employees are not confounded with differences in the nature of the work they do. Pay. Fringe Benefits. and Promotional Opportunities Although it is logical to expect part-timers to have a lower net income than full-timers since part-timers work fewer hours, Deutermann and Brown (1978) have also reported that, on average, part—time workers earn less per hour than their full—time counterparts ($2.87 compared with $5.04 in May, 1977). Owen (1978) suggests that part—timers typically receive lower earnings primarily because they are concentrated in lower paying (generally low skill, low level, and fairly routine) jobs. Part-timers also receive fewer fringe benefits (i.e., vacation days, sick leave, life insurance, health insurance, pension plan opportunities, and profit sharing plans), have few promotional opportunities, and receive little or no training (Nollen & Martin, 1978). In cases in which part-time employees do receive the same pay and benefits (prorated) as full-time employees, exclusion from training and promotional opportunities effectively lowers their future earning potential (Rotchford & Roberts, 1982). This is not meant to suggest that part- timers are always treated worse than full-timers. However, since satisfaction with pay, fringe benefits, and promotional opportunities would be contingent on the amount received (e.g., Locke, 1976), one cannot compare part- and full—timers’ attitudes with these specific I4 facets of work unless the variability attributable to different amounts of pay, fringe benefits, and promotional opportunities is controlled. Demographic Characteristics Comparative samples of part- and full-timers may differ with respect to demographic characteristics such as sex composition, job tenure, and age. For instance, Deutermann and Brown (1978) reported that in May, 1977, only three percent of all working males age 55 to 64 years were working part-time compared with 19% of all working females. The proportions of males and females working part-time increased for older people (+65 years) to 39% of all working males and 54% of all working females. The proportions of older full-time workers were just the opposite: 61% of all working males and 46% of all working females. Therefore, a greater proportion of females may be found in any sample of older part-time employees than in a comparative sample of older full- time employees. Tenure and age may also differ between the part- and full—time samples. To my knowledge, older full-timers are likely to have much longer tenure since many middle-aged people are likely to stay in the same full-time job until retirement. However, it is my belief that older people may be more likely to search for and acquire a part—time job, especially after retiring from a full—time position. Therefore, a study focusing on older part- and full-time workers may result in a sample of full-timers who have much longer tenure than a corresponding sample of part-timers. Furthermore, while both groups of employees will be comprised of people at least 55 years old, the possibility also arises that the part-time workers could generally be older than the full-time workers. This may result if most individuals retire from their full- 15 time jobs between the ages 62-65 (when they can begin to receive Social Security benefits) and then go on to work part-time to a later age. Controlling for demographic differences in sex, job tenure, and age will ensure that differences in the job attitudes held by part- and full- timers cannot be explained by different job attitudes held by males and females, experienced and newly hired employees, or older and younger workers. Summa y. The present study will focus on comparing the nature of different job attitudes for older part- and full-timers in retail/sales organizations. It is necessary to statistically control several confounding variables which may differentiate these two groups of employees aside from the number of hours they typically work each week. Controlling these variables will enable one to attribute different job attitudes expressed by these two groups to the different work-related values and desires of people attracted to these jobs, and not to other alternative explanations such as the different nature of the work conducted or demographic correlates of the type of people attracted to these jobs. The confounding variables which will need to be statistically controlled include: task complexity (nature of the work), pay, fringe benefits, promotional opportunities, sex, tenure, and age. Three constructs (job satisfaction, organizational commitment, and withdrawal intentions) comprise the focus of the present study in which the job-related attitudes of older part- and full-time employees are examined. These three constructs were chosen for two reasons. First, a great deal of literature is available which has examined the relationship between these constructs among full-time employees. This literature is reviewed below. Second, a comparison between older part- and full- 16 timers’ job-related attitudes and withdrawal intentions could provide some evidence to support Werther’s (1975) claim that not all part—time employees are unreliable employees. For instance, perhaps older part- timers who are satisfied with their jobs and committed to their respective work organizations are no more likely to withdraw from work than their full—time counterparts. The results of the present study will suggest whether there are differences between the levels of and relationships among job attitudes, organizational attitudes, and withdrawal intentions for older part- and full-time employees. The conceptual model which specifies the relationships between job satisfaction, organizational commitment, and withdrawal intentions is introduced in the next section. THE CONCEPTUAL FRAMEWORK A general conceptual framework often used to study the relationships among job satisfaction, organizational commitment, and withdrawal intentions is presented in Figure 2. Research addressing each of the concepts in Figure 2 and their interrelationships will be presented in order to provide a theoretical basis for the specific hypotheses developed later in this chapter. Literature relevant to the organizational commitment construct is presented below, followed by research addressing withdrawal intentions. A discussion focusing on the nature and causes of older part- and full-timers’ job satisfaction will be presented later in this chapter. Organizational Commitment Organizational commitment has been shown to mediate the relationship between job satisfaction and withdrawal tendencies (Hammer, Landau, & Stern, 1981; Steers & Rhodes, 1978; Wanous, 1978; Williams & Hazer, ~- . - ___._....—.'-- ~~.~- "' 17 .xcozoEo: EJEoocoo .9960 "N 830$ HZUZtEEOo M .._0 18 1986). Organizational commitment includes a strong belief in and acceptance of organizational goals and values, a willingness to exert considerable effort on behalf of the organization, and a strong desire to maintain membership in the organization (Mowday, Porter, & Steers, 1982). The construct of organizational commitment links attitudes (e.g., job satisfaction) and behavioral intentions (e.g., withdrawal intentions) in Figure 2. Mowday et al. (1982) have suggested that commitment is composed of both attitudinal and behavioral components. Attitudinal commitment reflects the individual’s identification with organizational goals and his/her willingness to work towards them. Behavioral commitment, on the other hand, reflects an individual’s binding to an organization through extraneous rewards and interests (e.g., pensions, seniority) rather than favorable affect toward the organization (Becker, 1960). Unfortunately, a widely accepted measure of commitment which incorporates both these attitudinal and behavioral components (Porter, Steers, Mowday, & Boulian, 1974) also contains items which measure withdrawal intentions. One would expect a high relationship between two measures containing similar items (Nunnally, 1978). Therefore, it is my opinion that any study which only uses this measure to demonstrate that organizational commitment mediates the relationship between job satisfaction and withdrawal intentions should be called into question. Another measure of organizational commitment developed by O’Reilly and Chatman (1986) which does not appear to include items measuring withdrawal intentions will also be used in the present study. This measure is described below. As an attitude, commitment is distinguished from job satisfaction in that the former is an affective response to the whole organization, 19 whereas the latter represents an affective response to specific aspects of the job (Williams & Hazer, 1986). Variables in turnover research that are conceptually similar to organizational commitment include job attachment and job commitment. Job attachment (Koch & Steers, 1978) describes an attitudinal response to a job, characterized by a congruence between one’s real and ideal jobs, an identification with one’s chosen occupation, and a reluctance to seek alternative employment. Alternatively, job commitment (Farrell & Rusbult, 1981; Rusbult & Farrell, 1983) refers to the extent to which an employee perceives that s/he is connected to a job and involves feelings of psychological attachment, independent of affect. Recent research further examining the construct of organizational commitment has focused on its underlying dimension of psychological attachment to the organization (O’Reilly & Chatman, 1986). Based on research conducted by Kelman (1958), O’Reilly and Chatman noted that individuals can accept organizational influence in three conceptually distinct ways: 1) compliance or exchange, 2) identification or affiliation, and 3) internalization or value congruence. Compliance occurs when attitudes and behaviors are adopted not because of shared beliefs but simply to gain specific rewards. Identification, in Kelman’s terms, occurs when an individual accepts influence to establish or maintain a satisfying relationship; that is, an individual may feel proud to be part of a group, respecting its values and accomplishments without adopting them as his/her own. Internalization occurs when influence is accepted because the induced attitudes and behaviors are congruent with one’s own values; that is, the values of an individual and the group or organization are the same. 20 O’Reilly and Chatman have suggested that new employees appear to base their attachment on compliance, exchanging behavior for extrinsic rewards. Over time, as one comes to understand and appreciate the goals and values represented by the organization, identification or pride in affiliation may develop. Internalization may involve the more psychodynamic process suggested by Bowlby (1982), Sanford (1955) and others in which a person imitates a model or adopts characteristics and values of the model. O’Reilly and Chatman’s review of previous research on commitment suggests that this process might occur as a result of a combination of clear role models, self—selection processes, social pressures, justification, and retrospective rationalization (e.g., Brown, 1969; O’Reilly, 1983; O’Reilly & Caldwell, 1981; Pfeffer & Lawler, 1980; Salancik, 1977). Bateman and Strasser (1984) challenged the notion that satisfaction is a cause of commitment. They provided empirical evidence that satisfaction is not a cause of commitment but rather a result of it. This evidence supports Salancik and Pfeffer’s (1978) assumptions that commitment to a course of action may determine subsequent attitudes. Commitment initiates a rationalizing process through which individuals "make sense" of their current situation by developing attitudes that are consistent with their commitment. However, other empirical findings provide evidence that satisfaction is a cause, not a result, of commitment. For example, Williams and Hazer (1986) noted that few studies relevant to establishing the relation between job satisfaction and organizational commitment have been conducted. Porter, Steers, Mowday, and Boulian (1974) proposed that commitment represents a global evaluative link between the employee and 21 the organization, with job satisfaction among commitment’s specific components. They further speculated that satisfaction would be associated with aspects of the work environment and thus would develop more quickly than commitment, which would require a worker to make a more global assessment of his/her relationship to the organization. The instability and rapid cause of satisfaction would suggest it as a cause of commitment, rather than vice versa. Other researchers supported the notion that satisfaction may be an antecedent of commitment, although many did not specifically address this relationship. Steers (1977) developed an exchange model of commitment, in which he proposed that individuals would first use their skills in the work environment in order to satisfy their needs and desires. Through this process, they would begin to develop attachment and commitment to the organization. Similarly, Stevens, Beyer, and Trice (1978) suggested that a manager evaluates perceived benefits and costs. Attachment and commitment subsequently develop. Although job satisfaction was not included in their analysis, they suggested that it might be an important predictor of commitment. Williams and Hazer (1986) recently tested a latent variable model to examine the causal links between job satisfaction, organizational commitment, and withdrawal intentions. They obtained support for only one model: job satisfaction ----> organizational commitment —---> withdrawal intentions. They conclude that "through a process of the evaluation of costs and benefits, individual needs and desires are satisfied, and the resulting affective state becomes associated with the organization, which has provided the job and its associated characteristics and environment. Commitment results from this 22 association". (Williams & Hazer, 1986, p. 230). The relevance this model has to older part- and full-time employees will be examined in the present study. A worker’s evaluation of costs and benefits is clearly associated with the discrepancy notion of satisfaction described earlier. Employees’ job satisfaction is a function of the degree to which they perceive that their important desires are met on the job. The extent to which job satisfaction becomes associated with the larger organization and leads to different degrees of organizational commitment for older part- and full-timers will be examined. The causal ordering of the variables which fits the data collected in the present study will also be explored (James & Brett, 1984), although it is my expectation that commitment will mediate the relationship between job satisfaction and withdrawal intentions. Hypotheses will be derived suggesting that there are mean differences in the depth of older part- and full-timers’ organizational commitment as well as differential relationships between the manner in which the depth of commitment experienced by older part— and full-timers mediates the relationship between overall job satisfaction and withdrawal intentions. Both of these types of hypotheses are described briefly below. Mean differences. Whether or not part-time employees would become less committed than full-timers to an organization even when differences between their levels of organizational tenure are statistically controlled is unclear. While all employees are only partially included in an organization, it is my belief that part—timers are not included as much as full-timers. The concept of partial inclusion was developed by Allport (1933) and further refined by Katz and Kahn (1978) to suggest that individuals are involved in a particular social system with only 23 part of themselves, since they play other roles in addition to that of organizational membership. Therefore, people might behave less as members of any given organization and more in terms of some compromise of their many segmental commitments in life at a particular time. While Katz and Kahn did not address the partial inclusion of part- and full- timers, their thesis would suggest that part-time employees would typically be less involved in a work role than full-time employees since part-timers do not spend as much time at work and have more time to pursue other non-work interests and activities. Hence, part—timers would not typically develop as strong an organizational identity as full- timers. O’Reilly and Chatman’s (1986) conceptualization of organizational commitment would seem to suggest that simply questioning whether or not older part-timers are committed toward their work organizations may not be appropriate since commitment is not a dichotomous construct. Organizational commitment is a matter of degree ranging from superficial compliance to deeper identification to the strongest feeling of commitment, internalization. Since it is my belief that part-timers would not typically develop a strong organizational identity, one can hypothesize that older full-timers would express significantly higher levels of organizational identification and internalization than older part-timers. In my opinion, organizational commitment expressed by older part-timers will be significantly more shallow and superficial than commitment expressed by older full-timers. Therefore, older part-timers would exhibit significantly higher levels of organizational compliance (i.e., exchanging behavior for extrinsic rewards) than older full- timers. Since Porter, et al.’s (1974) measure of organizational 24 commitment seems to me to contain similar items as O’Reilly and Chatman’s (1986) measures of internalization and identification (see Section D in Appendix B), it is also expected that older full-timers will exhibit significantly higher levels of organizational commitment than older part- timers using Porter, et al.’s measure. Relational differences. While it is expected that deeper feelings of commitment such as identification and internalization will be significantly lower among older part-timers than older full-timers, it is also my impression that these deeper feelings of commitment will be relatively invariant among older part—timers. Therefore, identification and internalization will be unable to differentiate feelings of organizational commitment expressed by older part—time workers. One can then hypothesize that the effect of satisfaction on withdrawal intentions will only be mediated by organizational compliance and not by identification or internalization for older part-time workers. The deeper feelings of organizational commitment (i.e., identification, internalization) are expected to mediate the relationship between satisfaction and withdrawal intentions for older full-time workers. Since Porter, et al.’s (1974) measure seems to contain similar items as O’Reilly and Chatman’s (1986) measures of internalization and identification, Porter, et al.’s measure is also expected to mediate the relationship between satisfaction and withdrawal intentions for older full-time workers. Whether or not it will mediate this relationship for older part-timers is unclear and will be explored. While Porter, et al.’s measure does not seem to include organizational compliance items, it does contain some items which appear to measure withdrawal intentions. Therefore, there might be enough variance for Porter, et al.’s measure to 9F significantly mediate the relationship between overall job satisfaction and withdrawal intentions for older part—timers, primarily because Porter, et al.’s measure does not seem to measure only organizational commitment. Intention to Withdraw Operationalization and prediction of actual withdrawal from the workplace is more meaningful than predicting withdrawal intentions. After all, actual absenteeism and turnover certainly have a larger economic impact on employers than measurements of one’s intentions to be absent or intention to turnover. However, conceptual frameworks which explore the relationship between variables such as job satisfaction, organizational commitment, and actual withdrawal behaviors (i.e., incidence of absenteeism and turnover) suggest that a simple relationship between these variables does not exist. A turnover model developed by Mobley, Horner, and Hollingsworth (1978) suggests that job satisfaction is related to turnover through several intermediate steps, including thinking of quitting, intention to search, and intention to quit. Moreover, the perceived probability of finding an acceptable alternative may influence one’s intention to search. Therefore, this literature suggests that intention to turnover would comprise 1) an intention to search for another job and 2) an intention to quit one’s present job. The measure of turnover intent developed for this study is comprised of items which incorporate both of these intentions. It is my belief that withdrawal from a job is not just represented by turnover, but by absenteeism for non—health reasons as well. Steers and Rhodes (1978) developed a conceptual model which explores major influences on employees’ attendance (or absence from work). This model 26 suggests that work attendance is a function of one’s motivation and ability to attend. Ability to attend is affected by health, transportation, and other factors normally considered beyond one’s control. Attendance motivation, in turn, is affected by an employee’s satisfaction with the job situation and a number of pressures to attend. Pressures to attend include such factors as organizational commitment, economic and labor market conditions, and incentive and reward systems. Other researchers have noted additional factors such as climate for attendance (Johns & Nicholson, 1982) and alternative valued leisure pursuits (Youngblood, 1984). Given all of these factors which may influence an individual’s withdrawal from the workplace, attitudes toward work are likely to have a stronger relationship with withdrawal intentions than with actual withdrawal behavior (Ajzen & Fishbein, 1980). Of course, method variance could also explain a high relationship between attitudes toward work and withdrawal intentions assessed from the same administration of a self-report questionnaire (e.g., Spector, 1987). The manner in which the effects of method variance can be minimized in questionnaire development and evaluated in one’s data will be discussed in the Method section of this proposal. In order to apply the conceptual model pictured in Figure 2 and discussed above to a population of older part- and full-time workers, it is necessary to first specify the work outcomes which they value the most. These valued work outcomes will most likely be the major determinants of their overall job satisfaction. A review of literature examining the different determinants of job satisfaction for full-time versus part-time and older versus younger employees may provide some preliminary information relevant to the types of work outcomes highly 27 valued by older part- and full-time workers. Research findings from these areas of literature will be presented in the next section. VALUED OUTCOMES FOR PART-TIME. FULL-TIME. AND OLDER WORKERS Part- and Full—Time Employees Several determinants of overall job satisfaction have been investigated in the voluminous published research examining job satisfaction of full—time employees (e.g., Locke, 1976). Perhaps one of the most often used instruments to measure job satisfaction has been the Job Descriptive Index (JDI) (Smith, Kendall, & Hulin, 1965). This instrument measures job satisfaction in the areas of pay, promotion, supervision, type of work, and people on the job. A review of the literature examining and comparing the determinants of part— and full— timers’ job satisfaction may suggest whether these five facets measured by the 001 are also the aspects of the job most highly valued by part— timers. One can then conclude whether the JDI (or a similar measure) could be used in a comparative study of part— and full—time employees. Unfortunately, my impression of most of the studies described below with few exceptions (e.g., Wakefield, Curry, Mueller, & Price, 1987) is that they were poorly designed. Many have confounded differences between the nature of the work conducted, benefits received, and work-related values of part- and full-timers so it is impossible to attribute empirical difference to any particular source(s). Earlier in this chapter the importance of statistical controls in research of this kind was discussed. While I recognize the limitations of these studies, they are the sole empirical source of hypotheses regarding those aspects of work which may be differentially valued by part- and full—timers. Additional data containing information about the differential nature of 28 the tasks conducted, pay, benefits, and promotional opportunities received, as well as the demographic composition of both the part- and full-time samples will be collected in the present study so different aspects of work valued by each group of employees can be more directly assessed. Hall and Gordon (1973) provided some evidence that different types of individuals may be attracted to part- and full-time employment. After an examination of role conflict and job satisfaction of females working full-time, females working part-time, and full-time homemakers, they stated that female part-time workers’ satisfaction may come from multiple involvements whereas full-time employees and homemakers seek deeper involvements and achievements in a more limited number of activities. Allen, Keaveny, and Jackson (1979) also revealed significant differences in preference rankings of various job characteristics (i.e., high income, job security, short hours, advancement opportunities, and feelings of accomplishment) among male and female full— and part-timers employed in managerial/professional and blue-collar jobs. Most notably, female part- timers did not place much importance on high income. If high income was important to them, Allen, et al. suggest that they probably would have pursued full-time jobs. Male part—timers in managerial/professional jobs placed a significantly higher value on job security than full-timers in these jobs. Since over half of these part-timers were married and over 70% of the wives were full-time homemakers, it is reasonable to assume that part-time employment was not voluntary in this case. The relatively higher ranking of job security could indicate that, in the face of economic pressures, a part-time job is better than no job at all. Finally, full-time male blue-collar workers tended to rank the importance 29 of advancement higher than their counterparts employed part-time. This finding could reflect a realization by male blue-collar workers employed part-time that most organizations restrict advancement opportunities to full-time employees. Given that advancement may have been an unrealistic expectation for part-timers, those who attach great importance to advancement either become full-time members of the labor force, or alter the importance of advancement to reduce dissonance (Allen, et al., 1979). Others have reached similar conclusions that many part- and full- timers may value different aspects of their jobs. Logan, O’Reilly, and Roberts (1973) determined that while satisfaction for work itself was viewed largely in terms of satisfaction with coworkers, supervision, and promotional opportunities for full-time hospital personnel, only satisfaction with coworkers was relevant for part-timers working in the same organization. Wakefield, Curry, Mueller, and Price (1987) more recently discovered that hospital employees who work fewer hours per week attach significantly less importance to pay and fringe benefits. These results were even obtained after income differences between full- and part-timers were held constant statistically. Miller and Terborg (1979) found that part-timers in retail/sales organizations were as satisfied with pay but less satisfied with benefits, work, and the job in general than full-timers in the same organizations. Miller and Terborg explained that while the hourly pay rate for part- and full-timers did not differ in these organizations, interviews with unit managers disclosed that part-timers received fewer fringe benefits and less desirable tasks, perhaps providing a reason for their lower levels of satisfaction with benefits, work, and the job in general. Unfortunately, no objective data was collected and recorded pertaining to fringe benefits or task 3O complexity which would allow the researchers to reanalyze their data, controlling for these confounding variables. umm r . Many facets of work which may be important to full-timers (e.g., promotional opportunities, high income, etc.) seem not to be as important to part-timers (e.g., Allen, et al., 1979). Perhaps this is true because many of these outcomes are virtually nonexistent to most part-timers (e.g., Deutermann & Brown, 1978; Mullen & Martin, 1978), although most of the studies described above (with the exception of Wakefield, et al., 1987) failed to obtain measures of these outcomes. It is my impression that if a part-timer valued any of these outcomes and had the option to choose between a full- and part-time job, s/he would probably pursue a full-time job. This is not meant to suggest that many part-time jobs could not be improved by offering more fringe benefits, more varied task responsibilities, and so on. However, the findings reviewed above suggest that many part-timers seem to be content with their jobs if the following specific, valued outcomes are present: 1) flexibility of work hours which allows one to pursue valued non-work activities (e.g., Hall & Gordon, 1973) and 2) interactions with coworkers (e.g., Logan, et al., 1973), suggesting that social interactions in the workplace may be highly valued by many part-time workers. The JDI does not measure satisfaction with the manner in which work fills one’s time and allows one to pursue valued non-work activities. Since the number and distribution of hours full—timers are scheduled to work may be highly invariant, satisfaction with this facet may not be meaningful in studies examining just full-time workers. To my knowledge, the JDI and other satisfaction instruments frequently cited in the literature have most often been used with full-time workers, perhaps explaining why this facet 31 would not have been included. However, it is my belief that a comparative study of part- and full-timers’ job satisfaction should include this facet. Satisfaction with coworkers is already one of five facets measured by the JOI and other satisfaction instruments (e.g., Lofquist & Oawis, 1969; Smith, et al., 1969) and seems to be an important facet to both full- and part-timers. It is unclear which of these outcomes (or additional outcomes) may be strongly valued by older part- and full-time workers. Some research has explored the determinants of job satisfaction for older and younger full- time workers. Unfortunately, like many of the studies reviewed above, statistical controls of other sources of variance (e.g., different nature of tasks conducted by younger and older workers; different pay and fringe benefits received by younger and older workers) are rarely introduced (see Gould, 1979 for an exception). Nevertheless, this research can help provide a framework for understanding the changing needs and desires of employees as they age in the workplace and society. Research addressing the aspects of work valued by older and younger workers is presented in the next section. The focus in this section will be on those aspects of work most likely to be valued by the older employees under examination in the present study. Older and Younger Employees Different aspects of work valued by older and younger employees develop largely as a function of psychosocial aging. Unlike physiological aging, psychosocial aging represents changes in personality, needs, expectations, and behavior due to the accumulation of experiences and changes in socially prescribed roles (Hall & Mansfield, 1975; Rhodes, 1983). These roles carry with them societal expectations 32 for behavior and have an influence on an individual’s needs and desires. Career theories allude to the changing roles and expectations associated with aging. According to these theories, while a younger employee is exploring a career and seeking growth and advancement, an older employee tends to be approaching a state of career maintenance and decline (e.g., Super & Bohn, 1970). Younger employees are often learning their jobs and expanding their skills while older employees may be expected to serve as mentors before they retire, passing on the knowledge they have acquired to younger workers. Social policy has reinforced at least partial retirement to everyone age 65 by specifying a limit to the amount of earnings Social Security recipients can retain (Morrison, 1982). While senior citizens may continue to work, it is not usually economically desirable for them to work on a full—time basis. Therefore, the workplace as well as society as a whole have prescribed different roles and expectations for older employees. Given this and the fact that older and younger employees are at different stages in the career and life cycles, it seems reasonable to suggest that they place different values on their work activities. In my opinion, younger workers may seek to maximize career opportunities largely because of upcoming financial responsibilities (e.g., paying children’s college expenses, making house payments, etc.). Most older workers’ children have grown and begun careers of their own. Many older people are more likely than younger people to own homes or have paid off most of their mortgages. Therefore, their financial debts and long—term commitments are usually much smaller. An older worker is no longer watching the job market in an attempt to maximize his/her potential earnings. This notion was recently supported by Pond and Geyer 33 (1987) who demonstrated that perceived work alternatives were a significantly stronger predictor of job satisfaction for younger workers than for older (i.e., middle-aged) workers. As these researchers suggested, older workers may not place as much value on outside job opportunities because they are in a later life and career stage. Similarly, Wright and Hamilton (1978), using data from the 1972-1973 Quality of Employment Survey, discovered that younger workers attached greater importance to opportunities for promotion, while the importance of extrinsic factors including security, fringe benefits, and working hours has tended to increase with age (Hall & Mansfield, 1975; Porter, 1963). ngmepy. The previous literature seems to suggest that older and younger employees value different aspects of their jobs, largely because of the different roles they play in both the workplace and society and because of the different amounts of work experience they have acquired. These roles and amounts of experience influence their needs and expectations, which suggests that older and younger employees value different components of work. A younger employee may evaluate his/her job relative to other job options and promotional opportunities so s/he can grow at work and meet financial obligations at home (e.g., Wright & Hamilton, 1978). By contrast, many older employees are not searching for growth in their careers but are instead planning for eventual retirement (e.g., Pond & Geyer, 1987; Super & Bohn, 1970). They may place more value on job security as well as preferences for fringe benefits and flexible working hours (e.g., Hall & Mansfield, 1975; Porter, 1963). One can suggest a few hypotheses regarding the work outcomes valued most by older part- and full-time workers by combining the two areas of 34 literature just presented. Many older employees seek a diminished work role, either because they want to retire soon or because retirement is a socially acceptable role for them to pursue in their stage of life (Morrison, 1982; Super & Bohn, 1970). Therefore, promotional opportunities may not be a particularly important aspect of either older part- or full-timers’ job satisfaction. However, retirement may mean a loss of valued social interactions if one does not partake in non-work activities which allow him/her to maintain social contacts. Part-time work can provide a solution to this dilemma by providing a valuable source of social interaction while also providing the flexibility and time needed to pursue other valued interests outside of work. Hence, satisfaction with the manner in which work fills one’s time and allows one to pursue valued non-work activities may account for significantly more variance in overall job satisfaction for older part-timers than for older full-timers. Satisfaction with social interactions is expected to be an important component of overall job satisfaction for all older workers; no differential hypothesis for older part- and full-time employees is offered. Personal interviews have also provided evidence that the opportunities to interact with others and work a limited number of hours per week are highly valued by older people searching for employment. The interviewed job applicants were clients of a state agency that serves as a job placement service for any interested citizen over the age of 55 in the local metropolitan area. A report on these interview findings is presented in Appendix A. While a few of this agency’s clients may have been fired from previous full—time positions, many of them retired early only to find that they 35 are unhappy with retirement and want to return to work at least on a part—time basis. Many do not want full-time positions because they do not have the energy or the desire to undertake major job responsibilities during this stage of their lives. Instead, they are looking for part- time jobs primarily because they feel bored and want something to fill their time. Several clients claim that family and friends have moved away and a job would provide a way to meet new people and provide a valuable source for social contacts. For some other clients, the illness and death of a spouse or loved one has depleted retirement savings, requiring a need to find a job in order to survive. Most clients in this latter situation would prefer to find a full-time job over a part-time job since they desperately need the extra income. However, others may have some savings but not as much as they would like in order to maintain a desired standard of living. For these individuals, a part-time job is more desirable since they can earn enough income and still have time to pursue non-work activities. Social interactions, the manner in which work fills one’s’time and allows one to pursue valued non-work activities, the nature of the job tasks, and income may be major facets of a job that older part-timers consider when they rate their level of overall job satisfaction. These four facets seem to represent important reasons why older people may desire to work part-time. The interviews suggested that they are important dimensions that are considered by older job applicants before they accept a new job. The limited amount of literature that has been reviewed and interviews that have been conducted do not provide enough evidence to conclude that these are typically the most important or only job facets relevant to 36 this population. Additional literature from sociology and industrial gerontology was reviewed in order to help provide an explanation of why some older people may decide to continue working at an older age or return to work after a period of retirement. This literature describes the psychological importance of the work role in society and adjustment to abandonment of the work role upon retirement. Examining the nature of the valued outcomes many people receive through their work roles and later miss after retirement may not only suggest why an older person continues to work full-time or secures a part-time job, but may also provide a basis on which s/he later evaluates his/her level of satisfaction with that particular job. Research relevant to these issues is reviewed in the next section. WORK ROLE Psyehological Importance of the Wprk Role Many sociologists and psychologists (e.g., Atchley, 1976; Friedmann & Orbach, 1970; Herzberg, 1966; Sofer, 1970) have identified several functions which appear to be served by work. These functions provide intrinsic and extrinsic meaning for the individual. The intrinsic elements relate to the interaction between the worker and the job. They include work as an important source of obtaining approved types of societal rewards and achievements, a significant component of one’s identity and self-concept, a creative outlet, and a locus for friendship relations and/or the gratification of other individual values. The extrinsic elements include work as an income-producing activity instrumental to the attainment of many life goals, a source of status and prestige, a structuring of time, and even a refuge from the stresses of participation in other areas of life activity. Deutscher, a 37 psychoanalytically trained psychologist, expressed the importance of the work role this way: "... Work has social reality; the work a man does gives him a contributing place in his community, and defines his sense of status and prestige. Through his occupation, man makes a bridge between his family and the outside world....” (Clark, 1972). All jobs will not provide every intrinsic and extrinsic element listed above. For instance, to my knowledge it is unlikely that an unskilled or semiskilled laborer positioned on an assembly line perceives work as a creative outlet. However, not all of these elements need to be present for an employee to perceive work as meaningful. It was noted earlier that the attainment of an employee’s values (or the closely related concept of "need fulfillment") has been consistently associated with job satisfaction (e.g., Greenhaus, Seidel, & Marinis, 1983; Locke, 1976; Lofquist & Oawis, 1969; Vroom, 1964). Therefore, as long as work provides an employee with valued intrinsic and extrinsic elements, it should serve a meaningful role in his/her life. Whether an occupation permits self-expression and self-fulfillment, such as many managerial and professional positions, or whether the work is routine and intrinsically uninteresting, both types of work organize the day’s activity and involve certain expectations on the part of others. Therefore, e11 work constitutes social roles (Super & Bohn, 1970). Clearly, then, people do not only lose an income when they abandon a work role through retirement. Many lose valued extrinsic and intrinsic elements associated with work, such as a sense of self-identity and an important source of social contacts. All retirees must also learn how to adapt to fewer responsibilities and larger amounts of discretionary time. 38 Self-Identity Sociologists have suggested that a person’s self-concept is the sum total of other’s perceptions of him/her combined with his/her own perceptions of self (Cooley, 1922; Head, 1930). Schein (1980) believes that the need to maintain and develop one’s self-concept and self-esteem is important for human adults. People do things which are consistent with how they see themselves and believe others see them, and avoid things which are inconsistent with how they see themselves or believe others see them. In general, Schein believes people strive to feel good about themselves and avoid situations which make them feel bad about themselves. When people take on new social roles they experience a new cycle in the socialization process and find themselves coping to develop new integrations of the self (Schein, 1978; Van Maanen & Schein, 1977). Such times can be constructive and growth producing or constrictive and limiting, depending upon the person’s ability to cope and upon the environment’s ability to provide growth opportunities (Nicholson, 1984). For many older people, a significant part of adapting to and coping with the retirement role appears to be associated with maintaining a sense of continuity, especially in one’s value system (Clark & Anderson, 1967; Lowenthal, 1971; Lowenthal & Chiriboga, 1973). These researchers suggest that maintaining continuity in values and in the perceived goodness of fit between values and goals (or past achievements) is a critical adaptive process. However, many people may not know where to turn in the environment to reclaim the self-identity that may have been lost when they retired from their jobs. While leisure activities can fill the time formerly occupied by work, they may not be capable of 39 giving an individual the kind of self-respect and identity s/he received from a job (Atchley, 1971). Self-identity derived through work is conceptually related to job involvement. Lodahl and Kejner (1965) described job involvement as "the degree to which a person is identified psychologically through his work, or the importance of work in his total self-image" (p. 24). Lawler and Hall (1970) focused on job involvement as referring to "psychological identification with one’s work", as well as "the degree to which the job situation is central to the person and his identity” (pp. 310-311). Wanous (1974) later speculated that the consequence of such a value orientation toward work is that an individual (when asked) is likely to state strong desires for certain job characteristics, such as autonomy, variety, challenge, feedback, and task identity. Hackman and Oldham’s (1975, 1976) Job Characteristics Model suggests that the presence of three related task characteristics (task variety, significance, and identity) would lead individuals with high growth need strength to derive feelings of meaningfulness out of their work. High growth need strength represents a strong desire to move towards self-actualization, which Maslow (1970) has defined as "becoming everything that one is capable of becoming" or "becoming more and more what one is." Clearly, in order to fulfill this potential one must become aware of and accept his/her self- identity. If this self—identity is derived largely from a job, the task characteristics which affect the psychological meaningfulness one derives from work would also affect the self-identity and self-respect s/he derives from work. Therefore, those who place a strong desire on deriving and maintaining a sense of self through work would also place high value on the task characteristics which affect the psychological 4O meaningfulness one derives from work. Since part-timers may not typically develop as strong an organizational identity as full—timers largely because work is a relatively smaller segment of a part-timer’s life, the importance of one’s self-identity derived through work may be less important for part-timers than for full-timers. Therefore, one can hypothesize that satisfaction with the perceived meaningfulness of work (i.e., task variety, significance, and identity) will account for significantly less variance in overall job satisfaction for older part— timers than for full—timers, even after differences in task complexity are statistically controlled. Social Interaction To my knowledge, work is an important source of one’s social relationships. Friendships often develop with one’s associates at work. While these friendships can still remain after retirement, it takes more effort to maintain face-to-face interactions. Once an individual retires, the workplace is no longer a convenient place in which one can meet his/her peers on a daily basis. Further, if many of one’s friends are still involved in their work roles, they are preoccupied during the day and have a limited amount of leisure time. Perhaps, then, an individual is more likely to become socially isolated upon assuming a retirement role. Loether (1964) conducted a survey of retirees where, among other things, they were asked what they missed most about not working. Sixty— eight percent of the respondents answered either "nothing at all" or "my work associates". Unfortunately, Loether did not report the proportion of respondents who gave each of these responses so it is premature to suggest that missed social interactions may be a large reason why some 41 people miss work on the basis of this one study. However, Quinn and Shepard (1974) also asked employees what they thought they might miss most if they stopped working. Twenty—nine percent claimed "nothing" while 37% gave reasons involving coworker relations, providing some additional evidence of the importance of social interactions in the workplace. Some of the other reasons expressed in both studies included keeping busy, staying healthy, justifying one’s existence, and giving one a feeling of self-respect. Allocation of Time and Leisure Pgreuite An historical examination of the relationship between time and all aspects of life reveals that time has only recently had a profound effect on the structuring of work and non—work activities. Atchley (1976) suggests that preindustrial man did not have to be concerned with time allocation; the rhythm of life was marked by the seasons, months, days, as well as the demands of family, friends, community, work, and leisure. These were interrelated demands and the time assigned among these could be varied by the individual to fit his/her perception of their requirements. The emerging factory system of the Industrial Revolution, however, depended upon the linking of the tasks of large numbers of workers into a single production process. As clocks and watches were developed in the early stages of western industrialization, work time became the measure of work effectiveness, and work time was the unit for which the employee was paid. Time has become the unit the worker sells; it has also become a measure of value for non-work activities as well. However, as one passes into a retirement role, s/he is faced with an abundance of unstructured free time after living a life of carefully allocated time. This role carries with it the increased decision-making 42 responsibilities associated with how to reallocate all this free time among non—work activities (Atchley, 1976). Many retirees may be pleased with the idea of having all this discretionary time. Those who are dissatisfied with their jobs may look forward to this time and even retire early if their financial situations allow. However, this does not necessarily mean they will be satisfied with retirement. Atchley (1976) notes that many employees enter a honeymoon phase after retirement, where individuals wallow in their newfound freedom of time and space. If finances, health, and family situation allow, this honeymoon can be a busy time filled with the pursuit of travel and involvement in numerous hobbies and activities. However, Atchley discovered that most individuals eventually settle into some sort of routine. Some individuals may not comfortably adjust to this routine, finding that they have too much free time on their hands. For instance, Moore (1963) suggests that many semiskilled and unskilled workers do not have the education to prepare them for a life of leisure. Their leisure participation has been predominantly concentrated in family affairs, rather than in community and/or voluntary associations. The demand for leisure occurs primarily among those who have found interesting ways of using it and who are deeply enmeshed in networks of social life (Moore, 1963). Eisdorfer (1972) believes that having money helps one enjoy retirement since leisure activities must usually be purchased. Membership dues for clubs, church/synagogue contributions, and travel costs may impose limitations on participation because of the marginal or submarginal income of the retiree. Even visits to the family may involve economic difficulties for the aged individual with a low income, since 43 transportation costs and gifts for grandchildren are among the hidden expenses which make such visits relatively costly (Eisdorfer, 1972). Further, illness in later years can deplete retirement savings quickly, forcing one or both spouses to return to work. Therefore, if one desires to pursue a life of leisure, retirement should be a valued role. Whether one can afford to etey retired may depend on other factors, such as health, inflation, and wise investments of retirement savings. Segmepy. The literature presented in this section suggests that the work role involves an important source of income (Eisdorfer, 1972), self- identity (e.g., Atchley, 1971; Lodahl & Kejner, 1965), social contacts (Quinn & Shepard, 1974), and structuring of one’s time (Atchley, 1971). Although sociologists have not always determined the importance of these dimensions through empirical analyses, it is interesting to note that many of these dimensions coincide with job facets typically measured by psychologists in examinations of job satisfaction (e.g., Locke, 1976). For instance, it was previously mentioned that the Job Descriptive Index (Smith, Kendall, & Hulin, 1965) measures job satisfaction in the areas of pay, promotions, supervision, type of work, and people on the job. The facets of pay and people on the job correspond with the dimensions of income and social contacts described above. The type of work facet is comprised of items which measure some task characteristics (i.e., task variety, significance, and identity) which would create psychological feelings of meaningfulness of work for an individual with high growth need strength (Hackman & Oldham, 1975, 1976) and would provide an important source of this individual’s self-identity (e.g., Lawler & Hall, 1970, Lodahl & Kejner, 1965). Perhaps the only dimension mentioned above that is not typically measured by psychologists is the structuring of 44 one’s time. However, to restate a point made earlier, it is my belief that a measure of satisfaction with the manner in which work fills one’s time may be meaningless for full-time workers since the number of hours they work and distribution of working hours are highly invariant. Such a measure is relevant for part-time workers since it reflects their ability to balance and structure higher amounts of discretionary time with a work schedule. We can summarize the psychological and sociological literature by drawing the following conclusions. Satisfaction with income, perceived meaningfulness of work, and social interactions at work should be major determinants of older part- and full-time employees’ overall job satisfaction. It was previously stated that satisfaction with the manner in which work fills time so one can pursue valued non-work activities should also be a major determinant of older part-timers’ overall job satisfaction. Whether or not this is also true for older full-timers is unclear and will be explored. However, it is expected to account for significantly more variance in overall job satisfaction for older part- timers than for older full-timers. Further, it was hypothesized earlier that satisfaction with perceived meaningfulness of work will account for significantly more variance in overall job satisfaction among older full- timers than part-timers. The JDI also measures satisfaction with promotional opportunities and supervision, since much literature has stated that these are important facets of job satisfaction for full-time employees (Locke, 1976). It was previously hypothesized that satisfaction with promotional opportunities should not account for a significant amount of variance in overall job satisfaction for any older employees, even if enough power is present to 45 detect a significant finding. Whether or not satisfaction with supervision is an important consideration for glee; full- or part-time employees will be explored. Now that the determinants of an older part- and full-timers’ job satisfaction have been hypothesized, a tentative model of the relationship between the facets of job satisfaction, overall job satisfaction, organizational commitment, and withdrawal intentions is presented. This model is discussed in the next section. AN T Figure 3 provides an integration of all the literature that has been presented regarding the relationships between job satisfaction, organizational commitment, and withdrawal intentions among older part— and full-time employees. The relationships (shown with arrows) between many of the components of the model are based largely on the conceptual framework often used to explore similar attitudes among full—timers (see Figure 2). Figure 3 is discussed as a means of specifying several hypotheses to be tested in the present study. The description of the components of this model and their relationships are divided into three sections: determinants of overall job satisfaction, organizational commitment, and intention to withdraw. Formal statements of all hypotheses will be presented at the end of each section. Determinants of Overall Job Satisfection Income and social interactions are believed to be highly valued by both older part- and full-time employees. Perceived meaningfulness of work is believed to be highly valued by older full-time workers, while the manner in which work fills time so one can pursue valued non-work activities is thought to be highly valued by older part-time workers. 46 ._oUoE 30.65.98 of. "m 6.59... 32.7.33... mo METEEV 325 vac; L fl magic . .mmmzoauozzfix 0w»._mommm _ mm>ozm3p 321. MT: 1:; mi: mmzo m3: Oh . m. pZMmm< _ zo:o<...m:.o r 2255.9? .e .._mmm3m Cgmzwmomn. I.:>> ZD:U.93) to detect a medium effect size (d = .5) in the difference between part and full-timers’ respective means on the commitment measures (Hypotheses 3a - 3c) (Cohen, 1977). It is important to note that the interaction between work status and propensity to search for another job or retire was not included in the prediction of turnover intentions in Hypotheses 4a - 4c. Therefore, it was not included in the power analyses for these hypotheses. The power analysis for this interaction is described below. Hypptheeis §e - 5b. Since these hypotheses were analyzed separately for older part- and full-time employees, the power analyses was also conducted separately for these groups. Therefore, the power analyses were not based on the total number of respondents, but on the number of respondents within each work status category (i.e., part-time/full-time). One hundred and fifty six respondents from each group (312 total respondents) would elicit an F value large enough to detect a minimum 2 percent increase in the total accountable variance of turnover intentions at p 5 .05. Therefore, although most of the hypotheses could be tested with fewer than 312 respondents, a minimum of 156 respondents from each group were sought (194 part-timers and 124 full-timers 1‘I’hree hmdred and ten total respondents (155 older part-timers and 155 older full-timers) would be recpired if both the job satisfaction - work status and cotunitment - work status interactions were to be siaultaneously tested. 58 were obtained) so that adequate power exists for examining all the hypotheses. Dietribption of Questionnaires Boxes of questionnaires were mailed or, whenever possible, hand delivered to personnel managers who agreed to participate in the study. They distributed questionnaires, cover letters explaining the purpose of the study, and postage paid envelopes to their older part- and full-time employees who were willing to participate. Each respondent used the envelope to return his/her completed survey to the researcher. All the participating employers were informed that they could not have access to individual responses so that confidentiality could be maintained. The researcher presented summarized results to the personnel manager of each participating organization at the conclusion of the study when there was interest. The University Human Subjects Committee approved the project prior to data collection. Before the content of the questionnaire is described, the issue of common method variance mentioned earlier will be addressed here since this problem partially determined the way the questionnaire was constructed and analyzed (e.g., Mitchell, 1985). A discussion of the steps taken to evaluate and minimize biased responses associated with common method variance is presented next. ngmon Methgd Varieneelfiiased Resulte Issues Common method variance was previously defined as the artifactual inflation of substantive relationships between measured constructs when the same research method and/or source is used to furnish all of the data (e.g., Glick, Jenkins, & Gupta, 1986; Mitchell, 1985). Fiske notes that "...each array of measurements from a construct-method unit contains variance associated with the method. Any obtained relationship between two such units can be due to method variance shared by both" (1982,p. 82). 59 Since it is my belief that the older part- and full-time employees under examination are the most relevant sources for obtaining all the data, and since these data were collected from a single administration of a questionnaire, conclusions regarding the interrelationships of the variables in this study are limited (e.g., Mitchell, 1985). However, several steps were taken in the selection of measures and the construction of the questionnaire to minimize (but certainly not eliminate) the method variance problem. These steps are addressed below. Selecting the Measures Campbell and Fiske (1959) suggested that researchers should use previous measures of constructs that have demonstrated reliability and construct validity. Attempts were made to do this with most of the measures utilized in the present study, although O’Reilly and Chatman’s (1986) measures of compliance, identification, and internalization were only used in the one previous investigation conducted by the instruments’ authors. Therefore, a more frequently used measure of organizational commitment (Porter, Steers, Mowday, & Boulian, 1974) was also examined in the present study. Reichers (1986) noted that many researchers have used the Porter, et al. measure of commitment, which Ferris and Aranya (1983) found to have greater predictive validity with respect to turnover than another measure of commitment developed by Hrebiniak and Alutto (1972). Unfortunately, Porter, et al.’s measure contains some items which appear to me to measure turnover intentions. The item overlap between these two constructs could, of course, account for its high predictive validity with respect to turnover (Nunnally, 1978). Nevertheless, the use of Porter, et al.’s measure in the present study provided the opportunity to examine and test comparative hypotheses between 60 O’Reilly and Chatman’s measures and Porter, et al.’s measure based on the item content of each of these instruments. A more "objective" measure of absenteeism was also added to the perceptual measure of "intention to be absent from work". Respondents were asked how many times they had been absent from work for non-health reasons during the previous three months. Provided there was some variation in their responses, it was hoped that this measure would increase the internal consistency reliability of the "intention to be absent" scale. As we shall see below, high internal consistency of each scale is very important in the examination of discriminant validity of all the measures. Since internal consistency of a scale is a function of the way in which individuals perceive and respond to scale items (Nunnally, 1978), it is important to understand how the construction of an attitude survey can affect individuals’ responses to the items. This issue is addressed in the next section. Cpnstrgetion pf the Questionnaire The wording and format of items are important concerns since they can affect individuals’ responses to these items (e.g., Bradburn & Sudman, 1979; Cronbach, 1960; Edwards, 1970). Two response styles, acquiescence and social desirability, have received a great deal of attention in previous research (Spector, 1987). Each of these response styles are discussed below. Acguiescence. An acquiescence response set is the tendency to agree with attitude statements regardless of content. This threatens the validity of attitude ratings because it is a source of "correlated errors" that can bias scale scores and produce misleading conclusions about group differences in subsequent analyses (Gove & Geerken, 1977). Winkler, Kanouse, and Ware (1982) have also suggested that it disguises true relationships between attitude items by falsely heightening the correlations among items that are 61 worded in the same direction at the expense of items that are conceptually related but worded in opposite directions. It was hoped that participants’ tendency to fall into an acquiescent response set could be minimized in the present study by grouping conceptually related items together in the questionnaire. This should help subjects notice conceptual similarities between items measuring a particular construct, yet also notice differences between items measuring different constructs (Ounham & Smith, 1979). This should reduce ambiguity across items which could help to maximize internal consistency reliability for each scale as well as help provide evidence for the discriminant validity of the measures. Principal components factor analysis with varimax rotation was conducted to evaluate the extent to which acquiescence is a problem in the data. If acquiescence is a problem, factors based on the positive or negative wording of the items, and not on the constructs the items are supposed to measure, would be derived. ' l ' b' t . Social desirability relates to a tendency to describe oneself in a socially desirable manner on self-inventories. Edwards (1970) found strong evidence that this occurs on personality inventories. While it may make some sense to partial social desirability from self— inventories, it is obviously essential to first develop a reliable measure of this construct. The concept of social desirability, however, has not been clearly defined by researchers. Measures of social desirability often lump together items from different constructs (Nunnally, 1978), resulting in instruments which are not always internally consistent. For example, Bachman, et al. (1967) reported low inter-item correlations for the Crowne and Marlowe (1963) Social Oesirability Scale. Out of a total of 465 inter-item correlations, only 15 exceeded .20 for a random sample of 778 boys. It is my 62 belief that using this instrument would be inadequate, since it is unclear what the instrument really measures. While social desirability was undoubtedly present in the responses in this study, I have attempted to minimize its effects. For instance, respondents were told that their answers will be kept confidential and not shared with employers. They were reminded that in most cases (with the exception of the pretest sample) even the researcher would not know their identities. Respondents were also asked in the instructions accompanying the questionnaire to avoid the tendency to give socially desirable responses and to be as honest in their answers as possible. They were told that the researcher had no interest in learning whether or not they were generally satisfied with their jobs, committed to their organizations, or that they intended to withdraw from work. Therefore, the respondents’ desire to present themselves in a socially desirable manner should have been minimized (though not eliminated) through these procedures. To evaluate the extent to which method variance did bias the data, several analytical procedures were conducted before any of the hypotheses were examined. These procedures are described below. F r n is internal nsistenc anal s and examinations f disepiminant validity. Previously it was mentioned that a factor analysis was conducted on all the items in the instrument. Separate factor analyses were conducted for the item responses of older part-timers and older full-timers. Items should load most highly on the factors representing the constructs the items were supposed to measure. The varimax factor loadings of each item for both part— and full-timers are presented in the Results section. Scale means, standard deviations, intercorrelations, and internal consistencies were examined and reported separately for older part- and full- 63 timers. Scale intercorrelations should be less than their respective internal consistency reliabilities (coefficient alpha). If intercorrelations are higher than or equal to the internal consistencies, then scales would be perfectly intercorrelated once they are adjusted for their lack of perfect reliability (Nunnally, 1978). Therefore, high coefficient alpha for each of the scales is very important. It is also important that scale intercorrelations be low. Multiple items have been carefully selected for each construct and clustered together in an attempt to maximize the internal consistency of each scale and to minimize scale intercorrelations. Pretesting the instrument with a small number of respondents (i.e., 5 — 10) also provided some evidence that the questionnaire instructions and meaning of items was unambiguous and clear (Seashore, 1986). The next section presents each of the items that have been selected to measure all of the constructs in the instrument. The items are presented and discussed in separate sections corresponding to the manner in which they are clustered in the instrument. A copy of the instrument can be found in Appendix 8. Development pf Ipstrument gob Satisfaction. Twenty-five items measuring satisfaction with different job facets can be found in Section A of the instrument. Many of these items have been adapted from the Job Descriptive Index (Smith, Kendall, & Hulin, 1969), Minnesota Satisfaction Questionnaire (Lofquist & Oawis, 1969), and the Quality of Employment Survey (Quinn & Staines, 1979). These questions ' were designed to measure satisfaction with salary (Al, A2, A32), social interactions at work (A42,.A5, A6, A7), manner in which work fills up one’s time (A82, A9, A10), supervision (A11, A12, A13), promotional opportunities ZThis item is reverse scored. 64 (A14, A152, A162), and perceived meaningfulness of work. This latter facet is measured by three task characteristics (Hackman & Oldham, 1975, 1976): variety (A172, A22, A232), identity (A20 & A24), and significance (A18, A19, A21). A six point agree-disagree response format was used for these items. Four additional questions (A25, A26, A27, A28) which measure overall job satisfaction can also be found in Section A of the questionnaire. These items were developed by Quinn and Shepard (1974). A coefficient alpha reliability estimate of .90 was recently obtained for a similar yet slightly longer scale used by Pond and Geyer (1987). Organizational Commitment. Thirty items in Section B of the instrument measured different forms of organizational commitment. Eleven items were adapted from O’Reilly and Chatman (1986), where a factor analysis confirmed that three items measured degree of compliance (81, 82, 83), three measured degree of identification (B8, B9, BIO), and five measured degree of internalization (811, 812, 813, 814, 815). Estimates of coefficient alpha have been computed by this researcher from a factor loading matrix presented in their article. While the identification (alpha = .85) and internalization (alpha = .89) scales are internally consistent, the compliance (alpha = .33) scale is unacceptable. Four items have been developed by this researcher (84, 85, 86, B7) and added to this scale in an effort to improve its internal consistency reliability. A six point agree-disagree response format was used for these items. The remaining fifteen items (816 - 830) in Section B comprised Porter, et al.’s (1974) Organizational Commitment Questionnaire (OCQ). The psychometric properties of this scale have been thoroughly investigated (Mowday, Steers, & Porter, 1979). Bateman and Strasser (1984) demonstrated an internal consistency reliability of this scale equal to .90. The response 65 format of these items has been slightly modified to be consistent with the six point agree-disagree format used for many of the other items in the questionnaire. Withdrawel Intention. The eight questions in Section C of the instrument were developed for this study. Three questions (C1, C22, C32) measured intention to turnover and were adapted from research conducted by Kraut (1975) and O’Reilly and Caldwell (1981). Similar items used by O’Reilly and Chatman (1986) were sufficiently reliable (alpha = .81). An additional two items (C7 & C82) asked subjects whether they thought it was most likely that they would leave their present jobs within five years in order to search for another job (with the added assumption that they would not have much difficulty finding new jobs) or leave their present jobs in order to retire. After item C8 was reverse scored, the mean scale score of items C7 and C8 was used to test the moderated relationship proposed in Hypotheses 5a and 5b. Three items (C42, C5, C62) measured intention to be absent from work and were developed specifically for this study. The remaining two items (C9 & C10) asked for descriptive information pertaining to the number of days absent from work over the last three months and the reasons for these absences. gob Descriptive Information. Section D of the instrument included questions pertaining to the respondent’s job tenure, average number of hours worked per week, hourly pay, fringe benefits (number of available sick days per year; number of vacation days per year; ability to get a group life insurance or group health insurance plan through work; involvement in a pension plan), perceived extent of promotional opportunities, and job title. Perceived promotional opportunities were assessed by asking respondents how many months it typically would take before a "good employee" in a similar job 2This item is reverse scored. 66 position and working approximately the same number of hours per week would get a promotion. The greater the perceived nUmber of months it takes is indicative of fewer perceived promotional opportunities. A respondent’s job title was assessed by providing relevant retail/sales job titles and brief job descriptions obtained from the Dictionary of Occupational Titles (1977). Each respondent was asked to choose the most appropriate job title to describe the type of work s/he performed. Corresponding codes (ranging from 1 to 8) from each of the Data, People, and Things scales were used to assess the task complexity of the respondents’ jobs. These codes were partialled from all the data before the hypotheses were tested in order to control for any variance attributed to the nature of the work conducted. All the questions described in this section were included to provide descriptive information and also allow statistical control over many of the confounding variables described earlier (tenure, hourly pay, fringe benefits, promotional opportunities, and nature of the work conducted). Qempgraphic Information. Section E of the instrument included questions pertaining to the respondent’s age, sex, race, marital status, educational level, and previous work experience. These variables were included to provide descriptive information about the participants as well as to allow the statistical control of confounding variables described earlier (age and sex). Other variables which served as statistical controls are found in Section D of the instrument and were described above. 67 P in h n tr m n The researcher surveyed a total of seven senior employees (four part- time and three full—time) on two occasions at a local department store. The employees were asked to complete the survey in the presence of the researcher and to ask for clarification whenever necessary. All seven employees completed the questionnaire in 25 to 30 minutes. At the end of this period, the researcher asked them to discuss their reactions to the survey and to brainstorm on ways any of the questionnaire items might be made less ambiguous and/or confusing. A couple of the employees in the first group had trouble completing the questionnaire, primarily because the instructions telling them how to complete a Likert-scale survey needed to be more explicit. These employees also had some questions about the meaning of a few of the satisfaction items. Some changes were made to the survey instructions and wording of these items without altering the intended meaning of the scales. This "new" survey was administered to different employees on the second occasion, and they appeared to have no problems understanding or completing the items. Copies of this survey were then mailed or hand delivered to store managers who had agreed to participate in the study. The surveys were distributed by personnel managers and supervisors to senior employees in each of these stores. A copy of the survey used in the present study can be found in Appendix 8. Oeta Analyses iab'l' nd V i it of Me 5 res In addition to the factor analyses discussed above, internal consistency calculations were conducted for each of the scales measured in the study after all the data had been collected. These calculations were used to determine whether the constructs were measured with satisfactory reliability and whether 68 any items should be eliminated before scale scores were computed. Scale scores were the mean rating assigned to the remaining items comprising each of the following thirteen measures (item numbers corresponding to the instrument presented in Appendix 8 follow in parentheses): Satisfaction with Income (Al-A3) Satisfaction with Social Interactions at Work (A4-A7) Satisfaction with Manner in Which Work Fills Up One’s Time (AB-A10) Satisfaction with Supervision (All-A13) Satisfaction with Promotional Opportunities (A14-A16) Satisfaction with Perceived Meaningfulness of Work (A17-A24) Overall Job Satisfaction (A25-A28) Organizational Compliance (BI-87) Organizational Identification (88-810) Organizational Internalization (811-815) Organizational Commitment (Porter, et al., 1974) (816-830) Intention to Turnover (Cl-C3) Intention to be Absent from Work (C4-C6) Propensity to Search for Another Job or Retire (C7-C8) The intercorrelation of the scales was examined after adjusting for the unreliability of each scale. There is no universally accepted rule-of-thumb that one can apply to decide when scales are sufficiently discriminable, but I proceeded to test hypotheses when corrected intercorrelations were less than .80. Of course, high multicollinearity of the unadjusted scale correlations for scales entering hierarchical multiple regression in the same step leads to large standard errors for the computational estimates of beta weights involved (Cohen & Cohen, 1983). I examined the size of the coefficients in the correlation matrix as well as the tolerance of the independent variables (Belsley, Kuh, & Welsch, 1980) in an effort to detect possible multicollinearity problems. As the findings indicate in the Results section, the hypotheses did not usually require any highly correlated scales to be entered into the same regression equations. Therefore, multicollinearity was not a significant problem in conducting the tests for the hypotheses. 69 mi on of t The remainder of the Method section describes the empirical analyses needed to support or refute each of the hypotheses. Before the hypotheses were examined, however, demographic differences between the older part- and full-time respondents were assessed. The manner in which this assessment was conducted is described first. ri tiv ta ' i . Univariate statistics examining the differences between older part- and full-time employees were conducted utilizing the personal background and job descriptive variables in Sections 0 and E of the instrument. The data obtained from these analyses provided some initial results describing the similarities and differences between both populations of employees. Most of these variables are the control variables (e.g., task complexity, job tenure, sex, etc.) described earlier. Data obtained from these variables were statistically partialled from the satisfaction, commitment, and withdrawal intention data in the remainder of the analyses described below. Discriminant analysis. To test Hypotheses 3a, 3b, and 3c, a hierarchical discriminant analysis (Klecka, 1975) was conducted to examine mean differences in the scale scores for older part- and older full-timers. First, the seven control variables found in Sections 0 and E of the instrument (task complexity, pay, fringe benefits, promotional opportunities, sex, tenure, and age) were entered as a single set. Second, the seven scales representing all the variables associated with job satisfaction in Figure 3 were entered as a single set. Third, the four measures of organizational commitment were entered simultaneously. Finally, intention to turnover and intention to be absent as well as the propensity to search for another job or retire were entered as one set. Work status (part- or full-time) was the 70 dependent variable. Responses to item 03 (see Appendix 8) were dummy-coded (Cohen & Cohen, 1983) to create the dichotomous work status variable. Individuals who worked an average of 35 or less hours per week were coded as part-timers; all others were coded as full-timers. The following discriminant equation was examined in this analysis:3 Work Status (Part or Full-Time) = Step 1 -- [Task complexity, Pay, Fringe benefits, Promotional opportunities, Sex, Tenure, Age] + Step 2 -- [Satisfaction with the manner in which work fills time, Satisfaction with the perceived meaningfulness of work, Satisfaction with promotional opportunities, Satisfaction with income, Satisfaction with supervision, Satisfaction with social interactions, Overall job satisfaction] 4. Step 3 -— [Compliance, Identification, Internalization, OCQ (Porter, et al. measure)] + Step 4 -- [Intention to be absent from work, Intention to turnover, Propensity to retire or search for another job] While the tendency for any of these scales to significantly discriminate between older part- and full-timers was examined, Hypotheses 3a, 3b, and 3c suggested that the measures of organizational commitment should discriminate between the two groups. Specifically, Hypothesis 3a was supported if the mean scale score of organizational compliance was significantly higher for older part-timers than older full-timers. 'Hypothesis 3b was supported if the mean scale scores of organizational identification and internalization were significantly higher for older full-timers than older part-timers. Hypothesis BC was supported if the mean scale score on Porter, et al.’s Organization 3To save space, beta weights and error terms are not shown with any of the equations presented in the Method section. 71 Commitment Questionnaire was also significantly higher for older full—timers than older part-timers. The analysis described above provided a conservative test of these hypotheses since significant differences between older part— and full-timers’ levels of commitment must be obtained ever and ebove any differences in part- and full-timers’ levels of job satisfaction in order to conclude that the hypotheses were supported. Determinants of overall jop satisfaction. Tests of Hypotheses 1, 2a, and 2b were conducted by using moderated multiple regression (Zedeck, 1971). After the seven control variables were entered into the equation, overall job satisfaction was simultaneously regressed onto the six facets of job satisfaction and the work status variable. The work status variable was the response to item 03. Like the discriminant analysis described above, this item was dummy-coded so part-timers = 0 and full-timers = 1. After these variables were entered into the regression equation as a set, both the satisfaction with the manner in which work fills one’s time X work status and satisfaction with the perceived meaningfulness of work X work status interaction terms were entered into the regression equation as a set. The following regression equation describes this analysis: Overall Job Satisfaction a Step 1 -- [Task complexity, Pay, Fringe benefits, Promotional opportunities, Sex, Tenure, Age] Step 2 -- [Satisfaction with the banner in which work fills one’s time, Satisfaction with the perceived meaningfulness of work, Satisfaction with promotional opportunities, Satisfaction with income, Satisfaction with supervision, Satisfaction with social interactions, Work status] + Step 3 -- [Satisfaction with the manner in which work fills one’s time Positive and statistically significant beta weights corresponding to satisfaction with income and satisfaction with social interactions at work 72 were taken as support for Hypothesis 1. Hypothesis 2a was supported if the beta weight for satisfaction with the perceived meaningfulness of work X work status interaction term was statistically significant and the plot of the separate regression lines for part-timers and full-timers showed a higher slope for the full-timer regression line. Hypothesis 2b was supported if the beta weight corresponding to satisfaction with the manner in which work fills one’s time X work status interaction term was statistically significant and the plot of the separate regression lines for part-timers and full-timers showed a higher slope for the part-timer regression line. Mediation test of organizational commitment. Hierarchical multiple regression (Cohen & Cohen, 1983) was used to examine Hypotheses 4a, 4b, and 4c. Each of these hypotheses suggested that different measures of organizational commitment should differentially mediate the job satisfaction withdrawal intentions relationships for older part-timers and older full- timers. Hierarchical multiple regression was used to determine if there are any differences between the satisfaction - commitment, commitment - withdrawal intentions, and satisfaction - withdrawal intentions relationships for part- and full-timers using each of O’Reilly and Chatman’s three measures of organizational commitment (compliance, internalization, identification) as well as Porter, et al.’s measure of organizational commitment (OCQ). Five regression equations were used to test the overall job satisfaction, organizational commitment, and withdrawal intentions interrelationships for each commitment measure. The first three equations examined sections of the overall model in Figure 3 to determine if work status significantly moderated the satisfaction — commitment, commitment - withdrawal, and/or satisfaction -withdrawal relationships. In these three equations, the variables were entered in sets, with the control variables 73 entering first, the main effects entering second, and the related interaction terms entering third (Cohen & Cohen, 1983). The first equation regressed commitment on satisfaction, with the work status X satisfaction interaction entered after the main effect for satisfaction was entered. The equation is presented below, where ’Commitment’ means one of the four measures of commitment: Equation 1: Commitment - Step 1 -- [Task complexity, Pay, Fringe benefits, Promotional opportunities, Sex, Tenure, Age] Step 2 -- [Overall job satisfaction, Work status] Step 3 -- [Overall job satisfaZtion X Work status] In the second equation, withdrawal was regressed on commitment, with the work status X commitment interaction entered after the main effect for commitment was entered. This equation is presented below, where ’Commitment’ means one of the four measures of commitment and ’Withdrawal’ means one of the two measures of withdrawal: Equation 2: Withdrawal = Step 1 -— [Task complexity, Pay, Fringe benefits, Promotional opportunities, Sex, Tenure, Age] Step 2 -- [Commitment, Work statds] Step 3 -— [Commitment X Work siatus] The third equation regressed withdrawal directly on satisfaction, with the work status X satisfaction interaction entered after the main effect for satisfaction is entered into the regression equation. This equation is presented below, where ’Withdrawal’ means one of the two measures of withdrawal: 74 Equation 3: Withdrawal = Step 1 —- [Task complexity, Pay, Fringe benefits, Promotional opportunities, Sex, Tenure, Age] Step 2 -- [Commitment, Work status] Step 3 -- [Commitment X Work status] Two additional regression equations once again examined the relationships between commitment and withdrawal as well as satisfaction and withdrawal, but they included the work status moderator wherever previous analyses determined it was statistically significant (Step 3 of Equations 1 - 3) in order to examine whether the variables have been specified in the causal order which fits the data (James & Brett, 1984). Equation 4 estimated the direct relationship between job satisfaction and withdrawal intentions (and the satisfaction X work status interaction) after the variance attributed to organizational commitment (and the commitment X work status interaction) had been removed from withdrawal intentions. Equation 5 estimated the direct relationship between organizational commitment and withdrawal intentions (and the commitment X work status interaction) after the variance attributed to job satisfaction (and the satisfaction X work status interaction) had been removed from withdrawal intentions. If the model specified in Figure 3 is appropriate, satisfaction should not have had a statistically significant direct effect with withdrawal intentions once the variance attributed to commitment had been removed from withdrawal intentions. This provides evidence that the relationship between satisfaction and withdrawal intentions is due to the direct relationship between satisfaction and commitment and commitment’s direct relationship with withdrawal intentions. This demonstrates that commitment mediates the relationship between satisfaction and withdrawal intentions. 75 These two regression equations are presented below. It is important for the reader to recognize that ’Commitment’ means one of the four measures of commitment and ’Withdrawal’ means one of the two measures of withdrawal intentions. It is also important to recognize that each hypothesis was tested separately for each of the two measures of withdrawal intentions. For instance, Equations 2 and 3 presented below were used once for intention to be absent and once for intention to turnover in the analytical tests of Hypotheses 4a, 4b, and 4c. Finally, whether work status was included as a main effect in Step 2 and as an interaction term in Steps 3 and 5 of Equations 4 and 5 was dependent on results obtained from Equations 1 - 3. Equations 4 and 5 were used exactly as they appear below only if work status significantly moderated all three of the satisfaction - commitment, commitment - withdrawal, and satisfaction - withdrawal relationships. If only one or two of the moderated terms was statistically significant, only these significant terms were used in Equations 4 and 5. Since propensity to retire or search for another job is not considered in Hypotheses 4a, 4b, and 4c, it was not introduced in the regression equations below. This variable was examined in the analytical tests required for examining Hypotheses 5a and 5b, which will be described in a later section: Equation 4: Withdrawal = Step 1 -- [Task complexity, Pay, Fringe benefits, Promotional opportunities, Sex, Tenure, Age] Step 2 -- [Commitment, Work Status] + Step 3 -- [Commitment X Work status] + Step 4 -- [Overall job satisfaction] 4. 76 Step 5 -- [Overall job satisfaction X Work status] Equation 5: Withdrawal = Step 1 -- [Task complexity, Pay, Fringe benefits, Promotional opportunities, Sex, Tenure, Age] + [Overall job satisfaction, Work Status] Step 2 -- 4. Step 3 -— [Overall job satisfaction X Work Status] + Step 4 -— [Commitment] 4. Step 5 -- [Commitment X Work status] If none of the interaction terms in Equations 1 — 3 (i.e, the satisfaction X work status term in Equation 1, commitment X work status term in Equation 2, and satisfaction X work status term in Equation 3) were statistically significant, then there was no support for Hypotheses 4a and 4b. For instance, if all the interaction terms were not significant when compliance was used as the measure of commitment, Hypothesis 4a was immediately rejected. Similarly, if all these interaction terms were not significant when internalization and identification were used as the measures of commitment, Hypothesis 4b was immediately rejected.‘ If at least one of the interaction terms in either of these regression equations was statistically significant, one or more of the relationships between satisfaction, commitment, and withdrawal intentions was significantly moderated by work status. In order to obtain support for Hypothesis 4a, four conditions had to be met. First, if the beta weight for the satisfaction X work status interaction ‘Ue shall see below that this simple decision rule does not apply to Hypothesis 4c, where we expected the OCQ to mediate the relationship between satisfaction and withdrawal intentions for full-timers. Since no comparative hypothesis had been offered predicting the form of this relationship for part-timers, one or more of the interaction tenms did not have to be statistically significant to obtain support for the hypothesis. In other words, the OCQ may also mediate the relationship for part-timers, but this had no impact on whether Hypothesis 4c was accepted. 77 term in Equation 1 was statistically significant, the plot of the separate regression lines for part-timers and full-timers (regressing compliance on satisfaction) would display a higher slope for the part-timer regression line. If this interaction term was not statistically significant, the beta weight for the main effect of satisfaction would be positive in sign and statistically significant. Second, if the beta weight for the compliance X work status interaction term in Equation 2 was statistically significant, the plot of the separate regression lines for part-timers and full-timers (regressing withdrawal on compliance) would display a higher slope for the part—timer regression line. If this interaction term was not statistically significant, the beta weight for the main effect of compliance would be negative in sign and statistically significant. Third, if the beta weight for the satisfaction X work status interaction term in Equation 3 was statistically significant, the plot of the separate regression lines for part- timers and full—timers (regressing withdrawal directly on satisfaction) would display a higher slope for the full-timer regression line. In other words, the direct effect of satisfaction on withdrawal would be stronger for full— timers than part-timers if the interaction term suggested that there was a differential direct effect of satisfaction on withdrawal for full— and part- timers. Finally, Steps 4 and 5 of Equation 4 would not add any significant increase in overall R2, while either or both Steps 4 and S of Equation 5 would add a significant increase in overall R2. This provides evidence that satisfaction —---> commitment —-—-> withdrawal is the correct specification of the model, since this analytical test demonstrated that satisfaction did not add any significant amount of explained variance in withdrawal not already accounted for by commitment. Furthermore, the test demonstrated that commitment gig add a significant amount of explained variance in withdrawal 78 even after the variance attributed to satisfaction had been partialled from withdrawal. It is important to conduct this final test, since some research has suggested that the proper model is commitment ----> satisfaction ----> withdrawal (e.g., Bateman & Strasser, 1984). At least one of the interaction terms in Equations 1 - 3 had to be statistically significant in order to test the presence of a significant moderator in Equations 4 and 5. At least one of these interaction terms had to be statistically significant in Equation 4 in order to provide preliminary support for Hypothesis 4a; failure to find any statistically significant interaction terms suggested that compliance mediated the satisfaction - withdrawal intentions relationship equally well for older part- and full- timers, which is not what Hypothesis 4a suggests. If the four conditions described above could only be met for one of the measures of withdrawal, then Hypothesis 4a was only partially supported. However, if all four conditions could not be met for either measure of withdrawal, this was taken as failure to obtain any support for this hypothesis. Similar decision rules were applied to Hypothesis 4b, where the major difference is that the relative slopes of the separate part- and full-time regression lines_were expected to be opposite to those reported above when compliance was used as the measure of commitment. These rules are outlined below using identification as the measure of commitment, but it is important to note that they were also true using internalization as the measure of commitment in order to obtain complete support for Hypothesis 4b. The conditions which had to be met include the following: 1) If the beta weight for the satisfaction X work status interaction term in Equation 1 was statistically significant, the plot of the separate regression lines for part- timers and full-timers (regression of identification on satisfaction) had to show a higher slope for the full-timer regression line. If this interaction was not statistically significant, the beta weight 79 for the main effect of satisfaction had to be positive in sign and statistically significant. 2) If the beta weight for the identification X work status interaction term in Equation 2 was statistically significant, the plot of the separate regression lines for part-timers and full- timers (regressing withdrawal on identification) had to display a higher slope for the full-timer regression line. If this interaction term was not statistically significant, the beta weight for the main effect of identification had to be negative in sign and statistically significant. 3) If the beta weight for the satisfaction X work status interaction term in Equation 3 was statistically significant, the plot of the separate regression lines for part-timers and full-timers (regressing withdrawal directly on satisfaction) had to display a higher slope for the part-timer regression line. 4) Steps 4 and 5 of Equation 4 should not have added any significant increase in overall Rf, while either or both Steps 4 and 5 of Equation 5 should have added a significant increase in overall R2. Once again, complete support for the stated hypothesis was contingent on finding at least one significant interaction for each measure of commitment utilizing each measure of withdrawal intentions as the dependent variable. If these four conditions were met for one of the measures of withdrawal intentions (intention to be absent or intention to turnover) or one of the measures of commitment (identification or internalization), only partial support was obtained for Hypothesis 4b. Hypothesis 4c is less stringent than the two preceding hypotheses since differences between part- and full-timers were not proposed.5 The previous requirement that at least one statistically significant interaction term had to be obtained in order to reject the null hypothesis was relaxed. Hence, if satisfaction was positively and significantly related to the OCQ, the OCQ was negatively and significantly related to withdrawal, and if the fourth condition (no significant increase in sithile the OCQ items appear to be conceptually similar to O'Reilly and Chatman's measures of identification and internalization, one might expect the same results as proposed in Hypothesis 4b. However, since it is my belief that the 0C0 also includes a few items which appear to measure turnover intentions (e.g., 824 t 826), the OCQ may also mediate the satisfaction - turnover relationship for older part-timers. 80 R? using Equation 4 and a significant increase in R? using Equation 5) listed above was met, Hypothesis 4c would be accepted. However, it was important to examine the interaction terms (see conditions 1, 2, and 3 on pages 78-79) to see if any of them were statistically significant. If any of these terms were statistically significant, it was not immediately apparent whether the OCQ mediated the satisfaction - withdrawal intentions relationship for older part- timers or older full-timers. Since Hypothesis 4c states that the OCQ should mediate the relationship for full-timers, separate plots of the regression of OCQ on satisfaction and withdrawal intentions on OCQ needed to show higher slopes for full-timers than part-timers it the corresponding beta weights were statistically significant. The separate plots of the regression of withdrawal intentions directly on satisfaction had to show a higher slope for part- timers than full-timers if the corresponding beta weight was statistically significant. The word ’if’ is important here because OCQ may have mediated the relationship for both part- and full—timers. Therefore, failure to find any significant interaction terms between work status and satisfaction and/or OCQ did not imply an immediate rejection of the hypothesis as it did with the tests for Hypothesis 4a and 4b. However, if any of the interaction terms were statistically significant, evidence must exist that demonstrates that the OCQ mediated the relationship for full-timers, and not for part-timers. This required an examination of conditions 1, 2 and/or 3 (see pages 78-79), whichever was relevant to the statistically significant beta weight(s). To summarize, if satisfaction was significantly and positively related to the OCQ, the OCQ was significantly and negatively related to withdrawal, and if condition 4 was met, Hypothesis 4c was accepted. If these conditions were met but only for one of the two measures of withdrawal intentions, Hypothesis 4c was partly supported. Of course, if these conditions were met for each of the 81 measures of withdrawal intentions, we completely accepted Hypothesis 4c. However, if conditions 1, 2, or 3 provided evidence that the OCQ mediated the relationship for part-timers but not for full-timers, this was taken as a failure to obtain support for the hypothesis. Dif er n ‘ l h oth r in re ensi h f r ther ’ b er turnover. Preliminary examinations of the data obtained separately from part-timers and full-timers was conducted using discriminant analysis (Klecka, 1975) to see if there were any mean differences between satisfaction with the job facets, overall job satisfaction, organizational commitment, and withdrawal intentions between those respondents who, if they left their current jobs within the next five years, think they might search for another job or think they might retire. These two groups were formed based on the respondents’ mean scale score (items E7 and E8 in Appendix 8). These items were not used in the computation of any of the withdrawal intentions measures. The test for Hypothesis 5a and 5b required the regression of intention to turnover on a particular measure of organizational commitment moderated by the propensity to search for another job or retire. This regression was conducted separately for older part- and full-timers. Hypothesis 5a states that organizational compliance will account for significantly more variance in intention to turnover among older part-timers who think they would search for other jobs if they left their present jobs within five years than for those older part-timers who would completely retire if they left their present jobs within five years. The following regression equation conducted on the data obtained from part-timers examined this hypothesis: Intention to turnover = Step 1 -- [Task complexity, Pay, Fringe benefits, Promotional opportunities, Sex, Tenure, Age] + Step 2 —- [Organizational compliance, Propensity to search for another 82 job or retire] + Step 3 -- [Organizational compliance X Propensity to search for another job or retire] If the interaction term in Step 3 was statistically significant and the plot of the separate regression lines for older—part timers who would search for other jobs and older part-timers who would retire showed a higher slope for those who would search for other jobs, there was support for Hypothesis 5. Hypothesis 5b states that organizational identification, internalization, and Porter, et al.’s OCQ measure of commitment should each account for significantly more variance in intention to turnover among older full-timers who think they would search for another job if they left their present job within five years than for older full-timers who think they would retire if they left their present job within five years. The following regression equation conducted on the data obtained from full-timers was used to test this hypothesis, where the equation was evaluated three times (substituting for ’commitment’ the measures of internalization, identification, and OCQ, respectively): Intention to turnover = Step 1 -- [Task complexity, Pay, Fringe benefits, Promotional opportunities, Sex, Tenure, Age] + Step 2 -- [Commitment, Propensity to search for another job or retire] Step 3 -- [Commitment X Propensigy to search for another job or retire] Hypothesis 5b was considered completely supported if the three interaction terms in Step 3 derived from running the regression equation three times, once for each measure of commitment (i.e., identification x work status, internalization x work status, OCQ x work status), were each statistically significant and the plots comparing the three pairs of regression lines for 83 older—full timers who would search for other jobs and older full-timers who would retire demonstrated higher slopes for those who would search for other jobs. RESULTS Survey Respondents A total of 932 surveys were distributed to 13 companies in the retail/sales.and grocery trades. Some of these companies had employees at multiple store locations who participated in the study. In total, surveys were sent to employees in 81 separate stores in the Midwest (Michigan and Ohio) and the Northeast (Massachusetts and New Hampshire). Three hundred and eighteen surveys were mailed back to the researcher, constituting a 34% rate of response. Rates of response from each corporation ranged from 22 to 100%, with the latter corresponding to smaller one-site companies with less than 10 senior employees. Some demographic and job experience data regarding the respondents is presented in Table 1. As this table demonstrates, 49.1% of the sample were male, 92.8% were white, and 75.8% were married. Eighty percent of the sample had never received a college education. This high percentage may have resulted from the types of non-professional jobs which these people held, but may also be typical of many people in this age category (2 55) across many occupations. The respondents ranged in age from 55 to 83 and had worked for their present employers for a period ranging from one month to 50 years. One hundred and ninety four (61%) of these respondents worked equal to or less than 35 hours per week, and were therefore treated as part-time employees in all the analyses which are described below. 84 85 Table 1 Respondent Demographic and Job Experience Levels Variable N‘ Percent Sex Male 156 49.1% Female 162 50.9% Race White . 295 95.5% Non-white 14 4.4% Marital Status Married 241 75.8% Never been married 6 1.9% Divorced 21 6.6% Legally separated 3 .9% Widowed 47 14.8% Education 1-4 years of high school 253 79.6% 1-3 years of college 44 13.8% College degree 12 3.8% Some postgraduate study 2 .6% Age 55-59 131 41.2% 60—64 80 25.2% 65-69 57 17.9% 70-74 35 11.0% 75-79 11 3.4% 80+ 4 1.3% Company Tenure _ Less than 1 year 44 13.8% 3 1 year and < 5 years 99 31.1% 2 5 years and < 10 years 62 19.5% 2 10 years and < 20 years 54 17.0% 2 20 years 58 18.2% Average Work Hours Per Week 5 20 hours 90 28.3% > 20 hours and g 35 hours 104 32.7% > 35 hours and g 45 hours 116 36.5% > 45 hours 7 2.2% ‘ Numbers do not always total 318 because of failure of respondent to answer the item. 86 Factor Analyses and Scale Reliabilities Tables 2 and 3 display the rotated factor loadings of the Likert- scale survey item responses for part- and full-timers, respectively. Principal components analysis with varimax rotation was used to obtain these loadings. Thirteen factors were expected to result from this analysis since the items were developed to measure thirteen different attitudinal constructs. If the data had been severely plagued with method variance, only one or two factors might have accounted for most of the variance in the subjects’ responses. While this did not occur, inspection of the eigenvalues suggests a fairly large general factor. It is not possible to determine what proportion of this factor may be "true" covariation and what proportion may be response bias. However, since seventeen factors had eigenvalues greater than one in both the part- and full-time analyses, subjects did not respond in the same manner across all the survey items. This seems to suggest that while method bias may be present in the data, there is evidence that subjects are responding differentially to item content. The first thirteen factors explained 65 percent of the variance in the part-timers’ responses and 67 percent in the full-timers’ responses. Closer inspection of the item loadings demonstrates that items developed to measure the same construct often load on the same factor. Items developed to measure different constructs generally did not load on the same factor, providing evidence that subjects were able to discriminate between items comprising different proposed scales. Three measures of commitment (organizational identification, internalization, and the Organizational Commitment Questionnaire) load on the same factor in both the part- and full-timer factor analyses. This Ite- §ati§E_Eitb_£az A1 .22 A2 .19 A3 .27 Satis. with Social Interactions at Work A4 A5 A6 A7 .01 .14 .24 .12 £37 Table 2 Rotated Factor Loadings for Part-Timers‘ Satis. with Manner in Which work Fills UE_QDSL§_IimE A8 .18 A9 .23 A10 .19 satis. with Sppervision A11 .23 A12 .24 A13 .34 Satis. with Promotional Dpppr. A14 .21 A15 .10 A16 .02 §2tisa_si£b_tbe Type of work I 00 A17 A18 A19 A20 A21 A22 A23 A24 .18 .15 .20 .14 .21 .18 .24 .28 Factor 2 3 4 5 6 7 8 9 1o 11 .06 .00 .11 .12 -.14 .13 .79: .08 .05 -.1o .11 .19 .04 .16 -.15 .07 .76. .09 -.12 -.04 -.08 .14 .08 .23 -.12 .03 .34 .08 .30 -.06 .10 .10 .14 .21 -.30 .37. -.10 .24 -.01 .15 .06 .12 .25 .09 .08 .67. -.02 .14 .18 -.2o .01 .15 .13 .00 -.05 .75_ .20 .04 -.05 .06 .09 .05 .10 .11 -.04 .82 .05 .03 -.04 .08 .06 .06 .06 .10 -.12 .07 .01 .61: .19 .06 -.04 .05 .26 .03 .07 .06 .16 .75. .07 -.13 .26 .16 .09 -.15 .24 .05 .00 .62 -.16 .18 .04 .22 .80. .08 4.07 .26 .05 .04 .10 -.03 .11 .17 .80. .10 .00 .22 .08 .00 .04 -.05 .24 .01 .70 .09 .01 .09 .03 .05 -.21 .oo .18 .16 .19 .55: .10 .17 .26 .22 -.06 -.01 .12 .00 .15 .73. -.04 .05 .11 .12 .03 -.09 .oo .13 .04 .78 -.09 .14 .14 .03 -.05 .07 .04 .15. .06 .57' -.04 .04 -.09 .29 .27 -.16 .29 .57. .13 -.03 -.1o .25 .32 .06 .24 .26 .11 .61. .17 -.09 .09 .13 .25 .02 .29 .29 .09 .69. .31 -.08 .oo .08 .14 .13 .09 . .08 .24 .68. .16 .09 -.06 .03 -.11 .03 -.01 -.01 -.07 .61. .08 .17 .05 .09 -.13 .04 -.06 -.1o .15 .47. .02 .40 -.12 .11 -.09 .26 -.07 -.22 -.05 .60 .05 .12 .10 .23 .08 .09 .11 -.2o 12 .20 -.01 °.25 -.05 -.13 .05 .12 -.O9 -.O9 -.04 -.OS -.02 -.03 °.O4 .18 .05 -.O4 .14 -.25 .06 .12 .24 13 .05 .07 .12 .01 .03 -.O1 .16 -.08 .16 -.O3 .13 .10 -.O3 1 Item Overall Job Satis. A25 .16 A26 .31 A27 .36 A28 .36 Org. eggl i ance 81 -.O4 82 -.29 83 .17 84 -.16 85 -.18 86 -.14 B7 .22 Org. Identification 88 .63. 89 .75. 810 .67 Org. Internalization 811 .48. 812 .62. 813 .63_ 814 .63_ 815 .77 g C 816 .56. 817 .74 818 .37. 819 .60. 820 .76. 821 .75 822 .26. 823 .60 824 .23_ 825 .64. 826 .41. 827 .35. 828 .75. 829 .74 830 .40 .27 .08 .11 .02 .08 .24 .43: .38 .41 .33 .20 .03 .14 .08 .18 .31 .09 .08 .16 .11 .07 .16 .09 .30 .10 .01 .07 .12 .15 .29 .18 .26 .11 .03 °.14 .15 .10 .04 .03 .11 .09 .13 .02 .28 .09 88 Table 2 (Continued) .29 .24 .10 .08 .06. .14 .13 -.1o .07 .14 .14 .18 .11 .OO .03 .26 .01 .15 .01 .17 .07 .01 .31 .10 .12 .02 .24 .12 .13 .11 .21 -.O4 .17 .09 Factor 6 7 .17 -.03 .07 -.15 .03 -.03 .16 -.O7 .28. .07 .53 -.10 .10_ .01 .76. -.O7 .72 .01 .28. -.12 .62 .12 .03 .23 .08 .22 .15 .20 .07 -.02 .03 .00 .21 .11 .OS -.06 .OO .01 .07 .11 .01 .14 .01 .17 .02 -.02 .10 .08 .11 .17 .07 -.01 .12 .00 .24 .03 .13 -.O9 .08 .08 .01 .04 .26 .11 .05 -.O3 .12 -.05 .19 .32 .26 .11 .10 .13 .13 -.19 .23 .20 .30 .15 .15 .19 .23 .25 .31 .26 .05 .10 .06 .02 .03 .01 .18 .03 .10 .10 .02 .06 .17 .19 .22 .12 .10 .05 .07 .05 .00 .07 .33 11 12 .20 .02 .19 -.05 .07 .01 .15 -.09 .36 .06 .24_ -.12 .64 .06 .04 -.06 .16 -.06. .17 -.44 .09 -.05 .02 .03 .08 .01 .04 -.05 .28 .14 .39 .08 .08 -.15 .24 .07 .11 .10 .17 -.05 .04 .03 .07 .07 .12 .23 .08 .08 .oo .05 .15 -.07 .14 .01 .02 -.02 .01 .09 .07 .41 .03 .31 .09 -.05 .03 -.01 .05 .03 13 .12 .13 .01 .18 -.O4 -.O3 .07 .03 .03 -.O4 .06 .14 -.35 -.08 .03 -.02 1 Item [ntention to Turnover Cl -.13 C2 -.15 C3 -.20 tntentigg to soummaiun £25k . C4 -.09 C5 .20 C6 -.01 Eigenvalue 17.25 Percent Variance Explained 27.00 Cumulative Percent Variance Explained 27.00 2 3 -A4: -J5 -A6_ ~17 -22 ~12 .03 ~JO ~30 .13 .07 J3 335 305 520 480 32.20 37.00 Interpretation of Factor Content Factor 1: Factor 2: Factor 3: Factor 4: Factor 5: Factor 6: Factor 7: Factor 8: Factor 9: 89 Table 2 (Continued) Factor 4 5 6 7 -.05 -.06 .04 -.08 -.08 .00 -.07 .01 -008 a“ -008 -009 .00 -003 -0“ -001 -.04 -.05 .19 .03 -.15 .04 .00 -.20 2.84 2.28 2.13 1.85 4.40 3.60 3.30 2.90 41.40 44.90 48.30 51.20 : This factor analysis is based on 148 part-timers. Indicates the highest factor loading for a given variable. I“ .03 -.02 .02 1.75 2.70 53.90 -.01 -.01 -.05 '.10 -.16 -.08 1.66 2.60 56.50 Org. Identification, Org. Internalization, Org. Conant. Cues. Intention to Turnover, Overall Job Satisfaction Satis. with Type of Work I 00, Overall Job Satisfaction Satis. with Supervision Satis. with Promotional Opportunities Org. Compliance Satis. with Social Interactions at Work Satis. with Pay Satis. with the Manner in Which Work Fills Up One's Time 10 11 .08 .08 ~54 .12 ~40 .11 ~8f' ~08 ~04 ~01 .04 .15 L55 L51 230 240 58.90 61.30 12 13 ~04 ~06 ~10 .07 1~24 .06 .05 ~o§ .07. .47 .66 ~05 L38 L29 220 200 63A0 65A0 Ital §atis. with Pay A1 .30 A2 .27 A3 .13 Satis. with Socigl [nteractions at Work A4 .03 A5 -.01 A6 .06 A7 .12 Satis. with Manner in Which Work Fills Dp One's Time A8 .13 A9 .22 A10 .09 Satis. with sppervision A11 .18 A12 .26 A13 .38 Sptis. with Promotional Dpppr. A14 .18 A15 .02 A16 .04 Satis. with the Type of Work I 00 A17 A18 A19 A20 A21 A22 A23 A24 -.14 .04 .12 .15 .22 .14 .23 .09 .02 .02 .02 .20 .01 .45 .20 .05 .18 .29 .13 .22. .53 .35 9(1 10 .04 -.03 .01 -.22 .09 -.15 .07 -.10 .10 .31 .00 -.13 -.15 .24 .01 .00 .11 .12 .12 .17 -.04 -.13 11 .06 -.o3. -.14 -.05 -.07 .14 .28 -.11 -.10 -.27 .11 .05 .02 -.17 .30 .14 -.02 -.07 -.03 -.15 Table 3 Rotated Factor Loadings for Full-Timers‘ Factor 3 4 5 6 7 8 9 .11 .21 .19 -.03 -.05 .67: .02 .06 .15 .21 .07 -.13 .72 .03 -.01 .05 -.14 -.01 -.03 .06 -.02 .14 .11 -.02 .44. .05 .14 .11 .04 .08 .05 .75. -.01 -.02 .28 .06 .41 .04 .65. .04 -.03 .04 .10 .11 -.02 .75 .07 -.04 .00 .06 .23 .54' .06 .36 -.08 -.04 -.07 .18 .38 -.04 .12 -.14 -.11 .06 .11 .21 .06 .20 .06 -.29 .08 .80: .04 -.03 -.04 .16 .02 .20 .81. .13 -.01 .01 -.02 -.03 .24 .42 .30 .04 .13 .11 .09 .28 .17 .48 -.03 -.16 .30 .05 -.08 .01 .79 -.01 - 13 .17 -.07 .09 .10 .78 -.02 -.24 .09 -.01 .34 .10. .48 .24 .04 .16 .20 -.1o .60. -.08 .25 .06 .19 .06 -.03 .52 —.15 .24 .18_ .01 .28 .26 .20 -.09 .15 .4o_ -.07 .15 .00 .18 -.09 .29 .67 -.09_ -.02 .09 .18 .13 .29 .23 -.4o -.17 .05 .09 .28 .00 .23. .10 .19 .32 .34 .02 -.01 .39 -.03 .11 .21 -.05 12 .11 .03 .01 .20 .01 .15 .29 .06 .35 .18 .27 13 -.09 -.11 .53 -.02 -.02 .18 -.25 .13 -.02 -.02 -.04 .08 .03 -.13 .26 .14 -.06 -.11 -.17 .21 .11 1 Item Dyerall Job Satis. A25 .34 A26 .35 A27 .42 A28 .45 Org. Cfll iance 81 .12 82 -.03 83 .24 84 -.1O 85 -.13 86 -.28 87 -.O4 Org. Identification 88 . . 89 .83. 810 .78 Org. Internalization 811 .44. 812 .80. 813 .64. 814 .81. 815 .90 fl . 816 .43 817 .87‘ 818 .46. 819 .63_ 820 .87. 821 .78. 822 .36_ 823 .72 824 .11 825 .48. 826 .61. 827 .47 828 .30. 829 .79 830 .44 .07 .17 .03 .03 .13 .32 .08 .28 50' J5 J3 J0 J9 .31 -.50 .05 -.28 -.21 .00 .05 .26 J3 .01 .01 .02 J» .20 .08 .35 J4 J9 23 J4 Jm .4f .22 -.03 .07 .11 -.20 -.01 -.01 -.03 .04 .05 .10 .05 -.04 -.26 .20 -.01 °.02 .12 .00 -.14 .00 .06 91 Table 3 (Continued) .13 .10 -.01 -.07 -.19 -.13 .01 -.11 .10 .14 -.01 .03 .07 -.05 .01 -.10 .19 -.01 .12 -.28 -.27 -.04 -.01 .06 .02 .06 .03 -.02 .34 .01 -.03 .14 .10 .03 .00 .30 -.11 -.09 .10 Factor .00 .16 .04 .24 .08 -.13 .01 .03 .04 .05 .00 .20 .05 .07 .05 .02 -.03 53' .02 -J3 .03 .03 .0g .53 .24 .16 .11 °.18 -.14 -.04 .05 .03 .02 .20 -.17 .04 -.1O .09 .00 .00 -.13 .09 .15 .24 -.12 .14 .18 .00 -.06 .06 .15 -.10 -.15 .21 .02 -.11 .20 -.06 .19 .22 -.05 .14 .05 .28 .05 .26 .13 .36 10 .00 .07 .10 .21 .02 -.10 .01 .08 .07 .05 .11 -.10 -.03 .00 ~03. ~59 .21 -.30 .09 -.05 11 -.03 .06 .15 .13 -.14 .28 .02 .00 -.08 -.03 .10 .01 .01 .22 -.12 .13 .13 .09 .17 .06 -.08 12 -.02 -.07 -.13 -.13 -J9 ~03 .03 ~04 ~01_ .33 ~06 -.04 -.05 .11 .11 -.01 .35 .03 .02 .36 -.01 -.04 -.02 ~01 .05 .05 ~02 -J4. .43 ~01 J1 13 .01 -.15 .02 .09 .17 .00 .12 .08 .00 .01 .02 .22 .09 .05 .06 .02 .09 .02 .06 .25 .11 59' .41 .04 .31 .08 .18 .02 .12 .42 .06 .02 .08 532 Table 3 (Continued) Factor 1 2 3 4 5 6 7 8 9 Item Intention to T rnover c1 -.20 -.09 -.22 -.12 -.01 -.2o -.19. -.21 -.65' C2 .07 .04 -.02 .08 .09 .00 -.73 -.01 -.20 c3 -.15 -.05 -.02 .01 .05 .08 -.64‘ .06 .14 n tion to stey Home from Work ' C4 -.08 -.13 .12 -.06 .02 .00 ..21 -.05 -.O9_ CS -.06 -.11 -.O9 -.10 .03 -.20 -.O3 .15 -.70 C6 .05 .13 -.05 .09 -.O4 .02 .08 .03 .08 Eigenvalue 15.53 4.58 3.69 3.07 2.36 2.20 2.04 1.90 1.69 Percent Variance Explained 24.30 7.20 5.80 4.80 3.70 3.40 3.20 3.00 2.60 Cumulative Percent Variance Explained 10 -.14 .18 .35 £6. .14 -.01 1.59 2.50 24.30 31.40 37.20 42.00 45.70 49.10 52.30 55.30 57.90 60.40 Interpretation of Factor Contept Factor 1: Factor 2: Factor 3: Factor 4: Factor 5: Factor 6: Factor 7: Factor 8: Org. Identification, Org. Internalization, Org. Commit. Cues. Overall Job Satisfaction Org. Compliance Satis. with Supervision, Satis. with Task Significance (i.e., Type of Work I Do) Satis. with Promotional Opportunities Satis. with Social Interactions at Work Intention to Turnover, Satis. with Task Identity (i.e., Type of Work I Do) Satis. with Pay : This factor analysis is based on 102 full~timers. Indicates the highest factor loading for a given variable. 11 -.08 -.07 .05 -.05 °.01 1.55 2.40 62.80 12 .14 .09 -.% -CM. -.62 13 -.22 .01 ~.12 .05 -.02 .01 65.00 67.20 93 was partly expected since all three scales are measuring the same underlying construct of organizational commitment. Furthermore, some item overlap across scales developed by different researchers (e.g., O’Reilly & Chatman’s B8 and Porter, et al.’s 821) would suggest that items from these different instruments should load highly on the same factor. Items developed to comprise different scales which load highly on the same factor will necessarily mean that these computed scale scores will be highly correlated. This may suggest that high multicollinearity could be a problem if these scales are entered into the same regression analysis since there could be a large degree of error in estimating the size of their beta weights. Since most of the hypotheses presented above require each measure of commitment to be examined separately and not in the same regresSion analysis, this should not be a problem. However, a high correlation between these commitment measures would suggest that the separate regressions are yielding the same information and tests of statistical significance are not independent. This concern will be addressed in greater detail when these results are presented. Results from the factor analyses as well as reliability analyses helped determine which items should be retained to construct the scales used in testing and examining each of the hypotheses. While it is important to maximize the internal consistency reliability of each scale, it is also important to try to keep previously developed items and scales intact so results presented here can be related to results obtained by others who have used these measures. Therefore, I decided to eliminate items only from scales I had originally developed or from scales which I had constructed by pulling together items from different instruments. Intact scales such as the OCQ were not changed. Items were removed from 94 scales only if doing so resulted in what I believe is a large (at least .05) increase in the internal consistency reliability of the scale for either part- or full-timers. However, if elimination of an item resulted in at least a .05 increase in the scale’s coefficient alpha for one group of subjects (e.g., part-timers) but any decrease in coefficient alpha for the other group of subjects (e.g., full-timers), the item was retained. Naturally, the decision to retain or eliminate items was not made simply on the basis of empirical rules. The content and conceptual meaning of each item was reexamined. If I felt I could not justify why an item did not appear to be consistent with other items developed to measure the same construct, I decided to keep the item even though its elimination would result in an increase in the internal consistency of a scale. Eight items were eliminated from additional analyses from the original 66 Likert-scale items presented to the subjects. One of these items referred to satisfaction with pay (A3), one more referred to satisfaction with social interactions at work (A4), two items I had developed (86 & B7) and two items developed by O’Reilly and Chatman (Bl & 83) were eliminated from the organizational compliance scale, one item was developed to measure intention to turnover (C1), and the last item dropped was developed to measure intention to stay home from work (C6). The items comprising each of the final scales and coefficient alpha for each of these scales for part- and full-timers is presented in Table 4. The intercorrelations of these scales are presented and discussed in a subsequent section. All scales appear to be internally consistent except for Intention to Stay Home from Work. Results from the factor analysis had also demonstrated that the three items in this scale load on different Items and Internal Consistency Reliabilities of Final Scales Scale Satis. with Pay Satis. with Social Interactions at Work Satis. with the Wanner in Which Work Fills Up One's Time Satis. with Supervision Satis. with Promotional Opportunities Satis. with the Type of Work I Do Overall Job Satis. Org. Compliance Org. Identification Org. Internalization Org. Commitment (OCQ) Intention to Turnover Intention to Stay Home From Work Propensity to Search for Another Job or Retire Items A1,A2 A5,A6,A7 A31,A9,Aio A11,A12,A13 A14,A15’,A161 A171,A18,A19,A20, A21,A22,A231,A24 A25,A26,A27,A28 32.34.35 33,39,310 311,312,313,314,31s 316,317,313‘,319, 320,321,32211323, 324 ,BZS,BZ6 .3271, 323,329,3301 c2‘,c31 c4‘,c5 c7,c31 1 This item is reverse scored. 955 Table 4 Coefficient Alpha Part-Tilers .87 .75 .91 .79 Full-Tilers .66 96 factors. This scale was eliminated from all the analyses which are presented below. Therefore, only one measure of withdrawal (intention to turnover) has been retained in this study. All hypotheses examining intention to be absent from work (see Figure 3) cannot be tested. Undoubtedly, the low reliability of this scale resulted from extremely low variance across item responses. While the research design attempted to minimize the effects of social desirability (see page 62), the content of some of the items may still have prompted respondents to give socially desirable responses. This may explain why there was not much variance in the items of this scale, since they appear to ask for the socially desirable response, "I intend to come to work everyday and never take a day off." For instance, 95 percent of all respondents who answered item C4 used response 6, 70 percent used response 1 for item C5, and 90 percent used response 6 for item C6. In retrospect, it is not possible to conclude whether most employees who were surveyed actually felt the need to provide socially desirable responses or are in fact desirable employees who would never consider taking a day off from work for reasons other than sickness. A lower coefficient alpha coefficient for satisfaction with the manner in which work fills up one’s time for full-timers than for part- timers was expected. I suggested earlier that the items comprising this scale may not be very meaningful to full-timers, since their hours do not tend to be nearly as flexible as part—timers’ hours nor provide them with as much time to pursue other activities. Given this expectation and the low reliability, the statistically significant relationship between this measure and overall job satisfaction for full-timers as well as part-timers in the analyses presented below is somewhat surprising. 97 Scale scores were computed for each subject by calculating his/her mean response to the items comprising each scale presented in Table 4. In order for a scale score to be computed, the subject had to have responded to at least two thirds of the items comprising a given scale. A mean was then computed based on the items to which s/he did respond. Subjects had to respond to both items for those scales comprised of only two items. If these requirements were not met for a given subject, his/her score for that scale was regarded as missing for all the analyses described below. The number of subjects added to the analyses by following this procedure instead of treating a scale score as missing if any of the items was missing was very small. At a maximum, only two percent (six out of 318 subjects) of the scale scores were added by following this procedure. This occurred with the satisfaction with the type of work, organizational internalization, and OCQ scales. Scale Mean and Demographic Differences Between Part— and Full—Timers Before multivariate techniques were used to test each of the hypotheses, univariate analysis was conducted to examine differences between part- and full—time employees. Since much of the previous research examining part- and full—time employees has not statistically partialled out control variables, it was thought that these univariate results would provide a preliminary basis of comparison similar to the type of results often reported in the published literature. Scale means and standard deviations for part- and full—timers are presented in Table 5. Part—timers were significantly more satisfied with the manner in which work fills up their time (p g .01), more satisfied with supervision (p 5 .01), more satisfied with promotional opportunities (2 g .05), and expressed more organizational compliance 98 Table 5 Means and Standard Deviations of Scales by Work Status (Part- or Full-Time)‘ Part-Tiler Full-Tiler Mean SD Mean SD Satis. with Pay 3.32 1.73 3.50 1.64 Satis. with Social Interactions at Work 5.06 .98 4.92 .94 Satis. with the Manner" in Which Work Fills Up 4.91 1.05 3.96 1.22 One's Time Satis. with supervision" 4.46 1.54 3.44 1.57 Satis. with Promotional. Opportunities 3.71 1.39 3.32 1.62 Satis. with the Type of Work I 00 4.70 .90 4.75 .82 Overall Job Satis. 4.46 1.25 4.32 1.42 Org. Conpliance' 2.42 1.44 2.06 1.32 Org. Identification 4.61 1.28 4.40 1.57 Org. Internalization 4.43 1.04 4.26 1.29 Org. Commitment (0C0) 4.41 .99 4.39 1.04 Intention to Turnover 2.35 1.58 2.05 1.50 Intention to Stay Home from Work 1.51 .84 1.47 .88 Propensity to Search for Another Job 3.32 1.87 3.27 1.83 Rather Than Retire ‘ Part-timer W ranged from 194 to 185. Full-timer W ranged from 124 to 122. ' The mean rating of partztimers was significantly higher than that of full-timers, p g .05 ( 2 5 .01). 99 (p g .05) than full-timers. The first two statistically significant differences were quite large (.85 and .66 standard deviation units, respectively) in comparison to the latter two differences (.26 standard deviation units). Since the sample size was large, a difference of one quarter of a standard deviation was statistically significant. Mean differences on the demographic and job descriptive variables are presented in Table 6. The part-timers who responded to the survey were significantly older than the full-timers (g 5 .01), had less company and job tenure (g_5 .01), and by definition worked fewer hours per week (g g .01) than the full-timer respondents. Part-timers also earned significantly less per hour than full-timers and received fewer fringe benefits (p 5 .01). All of these statistically significant differences are quite large. The smallest difference is on the number of sick days received, which is .89 standard deviation units. All the other statistically significant mean differences between part— and full— timers are greater than one standard deviation unit. Interestingly, those companies that offer a profit sharing plan to full—timers also offer one to part-timers; this was the only fringe benefit that did not exhibit a statistically significant difference between both groups. There were no differences in the type of work conducted as measured by the DOT codes, even though 30 different job classifications were sampled. A list of these job titles with their corresponding Data, People, and Things codes is presented in Table 7. There was also no difference in the sex composition of the part- and full-time workers, which may have resulted from the type of nonprofessional occupations which were sampled. Finally, there were no differences between promotional opportunities available to part- and 1130 Table 6 Demographic and Job Descriptive Differences Between Part‘- and Full-Timersb Part-Tinr Full-Ti-er Range Medi an Mean SD Range Medi an Mean SD Age“ 55.03-32.97 yrs. 65.07 yrs. 65.12 6.53 55.01-70.79 yrs. 59.26 yrs. 59.63 3.32 Sex (Female) 0-1 1 51.55% .50 0-1 .50 50.00% .50 Race (White) 0-1 1 95.77% .20 0-1 1 95.00% .22 cummimma' .m4mWym.3J1wm an 720 menowo1suyn.w32 w57 ubhmn" mruszwm LQyn.4J1 a” .n4swym.w26woian 9m Hourly salary" s3.35-s12.33 $5.55 $5.39 1.45 s4.25-s15.02 s10.55 $10.01 2.35 s of Hours Work/Week" 3-35 hours 22 hours 22.25 5.97 36-70 hours 40 hours 41.07 3.95 a of Sick 0ays/Year“ 0-14 0 1.77 2.92 0-14 3 4.57 3.45 s of Vacation 03ys/Vear" 0-35 7 7.93 7.17 2-42 20 20.47 3.34 Presence of a Grow .. Health Insurance Plan 0-1 0 25.26% .44 0-1 1 92.74% .26 Presence of a Pension (Retirement) Plan 0-1 0 20.62% .41 0-1 1 79.34% .40 Presence of a Grow" Life Insurance Plan 0-1 0 18.04% .39 0-1 1 71.77% .45 Presence of a Profit Sharing Plan 0-1 0 32.47% .47 0-1 0 35.43% .43 143x COMPLEXITY Data 1-6 4 4.01 1.35 1-6 4 4.02 1 74 People 1-3 6 6.40 .94 1-3 6 6 24 1.62 Things 1-7 7 5.90 1.95 2-7 7 6 10 1.70 101 Table 6 (Continued) Part-Tiler Full-Tiler Range Median Mean SD Range Median Mean SD NUMBER OF DAYS MISSED IN THE LAST THREE MONTHS 0-30 0 1.25 3.90 0-40 0 1.23 4.56 Surgery/Hospitalization 0-11 0 .36 1.66 0-40 0 .65 4.07 Actual sickness 0-26 0 .43 2.09 0-20 0 .49 1.99 Sick days 0-4 0 .03 .30 0 .00 .00 Transportation problem 0-1 0 .02 .14 0 0 .00 .00 Personal days 0-3 0 05 .28 0-2 0 .09 .34 Didn't feel like going 0-1 0 01 .10 0 0 .00 .00 Doctor appointment 0-6 0 14 .66 0 0 .00 .00 Had to assist family member/neighbor 0-30 0 21 2.20 0 O .00 .00 MOST PEOPLE DO NOT RECEIVE PROMOTIONS 0-1 1 90.72% .29 0-1 1 91.13% .29 ' N of part-timers ranged from 194 to 187. b N of full-timers ranged from 124 to 118. ' There is a statistically significant difference_between the part- and full-timers' means on this variable, 2 5 .05 ( Q g .01). 102 Table 7 Job Titles with the Dictionary of Occupational Title’s (1977) Data, People, and Things Codes Job Title Data People Things Advertising Manager Assistant Manager -- Merchandise Bakery Clerk Bottle Returns Clerk Cashier -- Checker Cashier -- Courtesy Booth Cashier -- Gift Wrapper Cheese/Pizza Clerk Deli Clerk Delivery Driver (warehouse to store) Demo Person (cook & pass out food in aisles) Department Manager Director of Consumer Affairs Fruit, Vegetable, and Flower Buyer Hair Stylist Laundromat Attendant Lottery Sales Clerk Maintenance/Janitor Meat Cutter/Meat Wrapper Merchandise Deliverer Merchandise Orderer Office Bookkeeper Payroll Clerk Produce Clerk Receiving Clerk Salad Bar Clerk Sales Attendant Salesperson Security Director Stock Clerk UdUOJ‘Ut‘3‘b—l#Oot‘ONddfilN0003‘5-FMU-b-b o-ounxisioowosoxo-Nooo-o-oowm-numo-ooooo-o-oooooox-e VVNNVNVNNNNJ‘I‘N#dNNNbUbt‘NNNNVNN 103 full—timers, largely because most respondents claimed that very few people (part— gr full-time) ever receive promotions in this type of industry. The lack of variance on this item suggests that it would not be useful to partial it out of the attitudinal scale scores. Therefore, promotional opportunities was not used as a control variable. Scale Intercorrelations for Part— and Full-Timers Scale intercorrelations as well as internal consistency reliabilities (coefficient alpha) are presented for part—timers in Table 8 and full-timers in Table 9. Intercorrelations between all the scales, control, and demographic variables for all the employees are presented in Table 10. As the factor analyses results had suggested, most of the scale intercorrelations are moderate in size. Multicollinearity may be a concern with the measures of organizational identification, internalization, and the Organizational Commitment Questionnaire, as correlations between these measures corrected for unreliability are often much higher than .80 (e.g., the highest correlation is between organizational identification and the OCQ for full-timers; unadjusted it is .85 and adjusted it is .96). However, it is important to remember that most hypotheses do not utilize all the measures of commitment simultaneously. Only evaluation of Hypotheses 3a, 3b, and 3c (hierarchical discriminant function analysis) required the use of all these measures in a single regression equation. Adjustments have been made to this analysis since the high correlations between the commitment measures suggest they should not be entered into the same regression equation. The modified analysis will be described shortly when the results for the tests for Hypotheses 3a, 3b, and 3c are presented. 104 Table 8 Scale Internal Consistency Reliabilities and Observed Intercorrelations for Part-Timers‘ 1 2 3 4 S 6 7 8 9 10 11 12 13 1. Satis. with Pay (.87)b 2. Satis. with .. Social Inter. .20 (.75) 3. Satis. with Manner in Which . Work Fills Time .20 .16 (.65) 4. Satis. with Supervision .27 .41 .24 (.88) VI 0 Satis. with ' Promo. Oppor. .33 .28 .16 .36 (.72) 6. Satis. with the Type of Work .31 .25 .37 .39 .39 (.79) 7. werall J» .0 C. Q. Q. Q. 0. satisfaction .41 .19 .44 .42 .36 .61 (.35) 3. Compliance -.34" -.10 .02 -.14' -.16' -.10 -.21" (.31) Q. I. .I O. O. 'I I. 9. Identification .28 .38 .39 .48 .32 .48 .54 -.16' (.84) Q. Q. Q. I. O. O. C. O 10. Internalization .25 .23 .29 .39 .29 .47 .52 -.12 .66“ (.80) I. C. O. O. D. Q. .- I. O. 11. Org. Commit Oues. .41 .28 .39 .50 .36 .55 .62 -.30 .79 .71'. (.91) 12. Intention to I. O I I. O. C. C. .. Turnover -.11 -.07 -.19 -.14 -.12 -.35 -.35 .08 v.29 -.23 -.46 (.79) 13. Propensity to . _. ._ Search for -.07 .02 -.03 .02 .03 .01 .04 -.01 .15 .17 .10 -.21 (.66) Another Job Rather than Retire ' N ranged from 194 to 180. b Internal consistency reliabilities (coefficient alpha) are printed in parentheses on the diagonal. :_One-tailed p g .05 One-tailed p g .01 105 Table 9 Scale Internal Consistency Reliabilities and Observed Intercorrelations for Full-Timers‘ 1 2 3 4 5 6 7 8 9 10 11 12 13 1. Satis. with Pay (.88)b 2. Satis. with Social Inter. .10 (.66) 3. Satis. with Manner in Which .. ._ Work Fills Time .21 .20 (.49) 4. Satis. with Supervision .40 .27 .35 (.82) 5. Satis. with . Promo. Oppor. .36 .05 .34 .26 (.80) 6. Satis. with the Type of Work .24" .47" .31" .43" .26" (.71) 7. merall Jab .. .. .. .. .. .. satisfaction .31 .29 .34 .39 .34 .53 (.92) a. compliance -.22" -.23" -.12 -.33" -.17‘ -.36" -.19‘ (.76) 9. Identification .35 .21 .30 .39 .18. .37 .63 -.20' (.89) .. . .. .. . .. .. . 10. Internalization .26 .19 .23 .44 .18 .35 .55 -.11 .77 . (.86) .. 11. Org. commit oues. .36 .26 .34 .44 .27 .43 .66 -.32" .35 .73" (.39) 12. Intention to Turnover .16' -.09 -.23" -.07 .15' -.21" -.13 .07 -.16' -.10 -.22" (.64) 13. Propensity to " Search for -.07 -.12 .14 .05 .03 .02 -.04 .09 .04 .14 .07 -.27 (.66) Another Job Rather than Retire ‘ N ranged from 124 to 121. Internal consistency reliabilities (coefficient alpha) are printed in parentheses on the diagonal. :.One-tailed p g .05 One-tailed p g .01 10. 11. 12. 13. 14. 15. 16. I N 1 Males are coded as 1 and females are coded as 2. Caucasian = 1, Black I 2, Asian = 3, Hispanic = 4, and Other Satis. Pay Satis. Social Satis. Manner Observed Intercorrelations of All Variables for All Employees‘ with with Inter. with in Which 1.00 Work Fills Time .15 .16 Satis. with Supervision .31 .39 Satis. with Promo. Oppor. .33 .16 Satis. with the Type of Work .28 .36 Overall Job _. __ Satisfaction .37 .24 Compliance -.29" -.17" Identification .31 .33 Internalization .25 .24 Org. Comit Dues. .39 .30 Intention to Turnover .00 -.06 Propensity to Search for -.07 -.04 Another Job Rather than Retire Age .14"' .11. Sex1 -.04 -.04 Race2 .04 .01 1136 Table 10 6 1.00 .06 7 a 9 10 1.00 -.20" 1.00 .03 -.13" -.06 -.01 .09 .01 .13 .10‘ ranged from 318 to 295 using pairwise deletion to maximize data. :.One-tailed p g .05 One-tailed p g .01 5. 17. 18. 19. 20. 21. 22. 24. 25. 26. 27. 28. 29. 30. 1If this work benefit is present, it is coded as 1. Otherwise, it is Company Tenure Job Tenure Hourly Salary No. Hours/Week No. Sick Days No. Vacn. Days Presence of Health Insurnce1 Presence of Retirement Pension Presence of Life Insurnce. Presence of Profit Sharing Plan Data People Things No. Days Absent in Last 3 Months coded as 0. ' One-tailed p g .05 .' One-tailed Q 5 .01 °.02 .02 .00 -.04 -.03 .05 107 Table 10 (Continued) -.03 -.01 -.01 .05 .06 .05 .01 -.02 -.01 .03 -.06 .01 -.03 -.07 9 10 -.02 .02 .01 .07 -.01 -.01 '.01 -.02 -.03 - 02 -.04 -.05 .02 -.01 .08 .06 .03 .04 -.01 .03 .07 .05 .00 -.02 06 .08 10 .10 11 .02 .05 -.03 40' 12 .02 .03 .03 -.07 -.02 -.02 '.07 -.08 .01 .05 .01 .04 .04 .05 .08 .05 .00 14. 15. 16. 17. 18. 19. 20. 21 22. 23. 24. 25. 26. 27. 28. 29. 30. 14 Age 1.00 Sex -.28“ Race .08 Company Tenure -.26'. Job Tenure -.23" Hourly Salary -.37" No. Hours/Week - .48 .I No. Sick Days -.35 No. Vacn. Days -.35 Presence of '. Health Insurnce. -.48 Presence of .. Retirement -.40 Pension Presence of Life Insurnce. - .42 Presence of Profit Sharing .00 Plan Data -.10' People .02 Things .11. No. Days Absent .11' in Last 3 Months :.One-tailed p g .05 One-tailed p g .01 1.00 -.15 .04 .06 -.06 -.01 .12' -.05 16 1.00 -.03 :.01 -.07 .01 ICNB Table 10 (Continued) 17 .35 .10' -.01 -.10' -.03 -.01 18 -.02 -.10' .00 19 -.12' -.02 -.05 20 1.00 -.09 .01 21 22 1.00 .44" 1.00 .51" .65" .32" .54" .40" 48" .11' .09 15" .01 .02 -.10 -.10 .02 .04 .01 23 24 25 1.00 -.07 -.04 -.07 -.06 -.03 -.04 26 1.00 -.08 -.06 .10' .04 1119 Table 10 (Continued) 27 28 29 30 27. Data 1.00 28. People .68" 1.00 29. Things -.21" -.10' 1.00 30. No. Days Absent .10' .07 -.06 1.00 in Last 3 Months :.One-tailed p g .05 One-tailed p g .01 110 Examination of Hypotheses Regression analysis was used to test many of the hypotheses. Several tables are presented and described below to provide the results of each results for the tests of Hypotheses 3a, 3b, and 3c are presented. As these tables clearly demonstrate, the dependent variable was regressed onto the independent variables in three analytical steps. First, the dependent variable was regressed onto the control variables. The beta weights of the control variables which were significantly related to the dependent variable are starred (i.e., ') in the Step 1 results in each of the tables. Second, the dependent variable was regressed onto the independent variables in Step 2. A statistically significant change in R2 noted at the bottom of the Step 2 results in each table denotes that these independent variables accounted for a statistically significant amount of the dependent variable’s variance over and above the amount of variance explained by the control variables entered in Step 1. The beta weights of the independent variables which were significantly related to the dependent variable after the control variables were statistically controlled are starred in the Step 2 results in each of the tables. Finally, the dependent variable was regressed on the moderator - independent variable product in Step 3 to test for interaction effects. A statistically significant change in R2 noted at the bottom of the Step 3 results in each table denotes that these interaction effects accounted for a statistically significant amount of the dependent variable’s variance over and above the amount of variance explained by the control variables entered in Step 1 and the independent variables entered in Step 2. The beta weights of the interaction effects which were significantly related to the dependent variable after the control variables and independent variables were 111 statistically controlled are starred in the Step 3 results in each of the tables. Hypothesis 1. Hypothesis 1 stated that satisfaction with income and satisfaction with social interactions at work would be significantly related to overall job satisfaction for all older part- and full- timers. Table 11 provides partial support for this hypothesis, since satisfaction with pay was significantly related to overall job satisfaction (8 =.15; p 5 .01). However, the main effect for satisfaction with social interactions at work was not statistically significant, despite the fact that a simple bivariate significant relationship was reported in Tables 8 and 9. Apparently, when the control variables and interrelated satisfaction scales are combined in a single regression equation, this facet of satisfaction does not explain a statistically significant unique amount of variance in overall job satisfaction. While no hypotheses were presented regarding the other facets of satisfaction, Table 11 demonstrates that overall job satisfaction was also significantly related to satisfaction with the perceived meaningfulness of work (i.e., satisfaction with the type of work I do) (8 = .37; p g .01) and satisfaction with the manner in which work fills up one’s time (8 = .21; p g .01). While the control variables only explained ten percent of the total variance in overall job satisfaction, the facets of satisfaction accounted for an additional 35 percent of the variance. The control variable which had the strongest significant relationship with overall job satisfaction was age, suggesting that even among the senior employees sampled here, the older respondents were generally more satisfied with their jobs than the "younger" respondents. It is important to remember that while part-timers were significantly IJlZ Table 11 Regressing Overall Job Satisfaction on Control Variables, Facet Satisfaction Indices, and Work Status Moderator Step 1 Step 2 Step 3 LLS Let; Bite. 5% Sggp 1 -- Control Variables 1. Age .23" .05 .04 2. Pension Plan1 .10 .05 .05 3. Health Insurance1 -.25' -.10 -.09 4. Company Tenure -.20' -.08 -.09 5. Hourly Salary .22. .05 .06 6. People (DOT Code) -.06 -.10 -.11 7. Data (DOT Code) .05 .06 .07 3. Sex2 .11 .03 .03 9. Number of Vacation Days .09 .05 .05 10. Profit Sharing Plan1 .12 .05 .05 11. Life Insurance1 .03 -.01 -.02 12. Things (DOT Code) -.01 -.04 -.04 13. Number of Sick Days .01 .02 .04 Sggp 2 -- Mpin Effects 14. Satis. with the Type of Work I Do .37“ .33 15. Satis. with the Manner in Which Work Fills Up Time .21" .29 16. Satis. with Pay .15" .15 17. Satis. with Promotional Opportunities .09 .11 18. Satis. with Supervision .11 .12 19. Satis. with Social Interactions at Work .01 .01 20. Work Status (19.1. or 5.7.)3 .07 .16 §£gp 3 -- Interpgtion Effgpgg 21. Work Status x Satis. with the Manner in Which Work Fills Up One's Time -.30 22. Work Status X Satis. with the Type of Work I Do .22 F 2.28 11.26 10.39 df 13,280 20,273 22,271 32 .096 .452 .457 Change in 32 .096 .356" .005 113 Table 1] (Continued) 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0. 2 Males are coded as 1 and females are coded as 2. 3 Full-timers are coded as 1 and part-timers are coded as 0. '95.6 "2:.m 114 older than full-timers in this sample (see Table 6), part-timers did not express more overall job satisfaction than full-timers (see Table 5). Hence, the statistical significance of age in Table 11 was not indicative of a statistically significant difference in overall job satisfaction between part- and full-timers. Other statistically significant control variables suggested that those employees with a higher hourly salary, less company tenure, and, surprisingly, lack of a work-sponsored health insurance plan were most satisfied with their jobs. Hypotheses 2a and 2b. Hypothesis 2a suggested that perceived meaningfulness of work (i.e., satisfaction with the type of work I do as measured in the survey instrument in Appendix 8) will account for significantly more variance in overall job satisfaction among older full-timers than among older part—timers. Table 11 shows that this hypothesis was not supported, suggesting that the type of work one conducts was just as important to part—timers as to full-timers. Hypothesis 2b stated that satisfaction with the manner in which work fills one’s time should account for significantly more variance in overall job satisfaction for older part-timers than for older full- timers. However, this interaction term in Table 10 was not significant. This suggests that the manner in which work fills up one’s time is just as important to full-timers as it is to p‘rt-timers. Neither one of the interaction terms included in this regression equation even approached statistical significance when they were examined separately. Therefore, hypotheses 2a and 2b must be rejected. Hypothesis 3 . 3b. and BC. I previously mentioned that three of the measures of commitment (organizational identification, internalization, and the OCQ) were highly correlated and could not be 115 simultaneously entered into the same regression equation. Therefore, hypotheses 3a, 3b, and 3c were tested by conducting three hierarchical discriminant function analyses. Each of these functions only differed in terms of the commitment measures which were included; all the other variables were entered into each of the three discriminant functions. Instead of entering all the scale scores in one regression, Table 12 provides the results of a discriminant function analysis using all the control variables, satisfaction measures, and just the organizational compliance and identification measures of commitment. Table 13 presents the results when just organizational compliance and internalization were allowed to enter the equation. Finally, the results of only allowing the organizational compliance and OCQ measures of commitment to enter the discriminant function equation are presented in Table 14. Since organizational compliance was not highly correlated with the other measures of commitment in Tables 8 and 9, it was allowed to enter each discriminant function with one of the other measures of commitment in Tables 12, 13, and 14. Tables 12 through 14 also provide the results of two types of discriminant function analyses. The left side of each table displays the standardized canonical discriminant function coefficients after all the variables have been forced to enter in the order specified by the research model (i.e., control variables enter first, which are followed by satisfaction measures, then commitment measures, and finally the withdrawal measures). This is the type of discriminant function analysis (i.e., hierarchical analysis) which was proposed earlier. In addition, the right side of each table displays the standardized canonical discriminant function coefficients after a stepwise procedure was used to 1115 Table 12 Discriminant Function Analysis Predicting Work Status1 Using Organizational Compliance and Organizational Identification Org. Compliance Org. Identification Discrilimt chtion Usim All variables Entered in This Analysis” .68 Hourly Salary" -.27 Satis. with the Manner in Which Work Fills Up One's Time" .22 Pension Availabilityz" -.19 Sex3" -.17 Age“ -.16 Satis. with Pay" .15 Life Insurance Avail.2" .15 Number of Vacation Days .13 Org. Compliance“ .13 Health Insurance Avail.2 .12 People -- 001 Code. .11 Satis. with the Type of Work I Do .09 Overall Job Satisfaction .09 Data -- DOT Code .09 Things -- 001 Code. .07 Company Tenure .05 Intention to Turnover .04 Satis. with Social Interactions at Work .04 Satis. with Supervision .03 Number of Sick Days .03 Satis. with Promotional Opportunities .01 Propensity to Search for Another Job or Retire .OO Org. Identification .00 Profit Sharing Plan2 Eigenvalue = 1.92 Eta2 = .66 Wilks' Lambda = .34 Chi2 s 295.96" D.F. = 24 % Classified Correctly = 90.34% F to enter Disgriminant Function' 2J7" .02 Discrilifmt Fulction Usim Best variables in This Analysisc .65 Hourly Salary" -.24 Satis. with the Manner in Which Work Fills Time" .20 Pension Availability2" -.20 Sex3" -.13 Age" -.15 Satis. with Pay“ .15 Life Insurance Avail.2" .12 Number of Vacation Days' -.16 Org. Compliance" .14 Health Insurance Avail.2" .18 Satis. with the Type of Work I 00" .09 Things -- DOT Code Eigenvalue = 1.88 Eta2 = .65 Wilks' Lambda = .35 chi2 = 293.45" 0.3. = 12 % Classified Correctly = 90.55% 117 Table 12 (Continued) ‘ This is the F—to—enter after all preceding variables (control and satisfaction variables) have been forced to enter the equation. The default F-to-enter for SPSS must be at least 1.00 or larger. 1 F values of 1.59 and 1.92 correspond to independent significance eve s of .05 and .01, respectively, for this size sample. ” These are the standardized canonical discriminant function coefficients after all variables have been forced to enter in the order specified by the research model (Figure 3 0 These are the standardized canonical discriminant function coefficients after a stepwise procedure was used to onl enter the best discriminating variables at each step and to reeva uate and possibly remove variables entered in previous steps. This procedure maximizes the difference between part- and full-time employees using the fewest number of discriminating variables. .4 Full-timers are coded as 1 and part-timers are coded as 0. There were 172 part—timers and 118 full-timers in this analysis. 2 If this work benefit is present, it is coded as 1. Otherwise, it is coded as . Males are coded as 1 and females are coded as 2. a p g .05 a ' p g .01 11!! Table 13 Discriminant Function Analysis Predicting Work Status1 Using Organizational Compliance and Organizational Internalization Org. Compliance Org. Internalization Discriminait Falction Usim All variables Entered in This Analysisb .66 Hourly Salary" -.28 Satis. with the Manner in which Work fills Up One's time" .22 Pension Availability2" -.19 3ex3" -.16 Age" -.15 Satis. with Pay" .15 Life Insurance Avail.2" .15 Number of Vacation Days .15 Health Insurance Avail.2 -.13 Org. Compliance“ .12 Overall Job Satisfaction .12 People -- DOT Code‘ .10 Satis. with the Type of Work I Do. .08 Data -- DOT Code .08 Things -- DOT Code -.07 Company Tenure -.O6 Intention to Turnover .05 Satis. with Social Interactions at Work -.03 Satis. with Supervision -.02 Org. Internalization .02 Number of Sick Days .01 Satis. with Promotional Opportunities .01 Propensity to Search for Another Job or Retire -.01 Profit Sharing Plan2 Eigenvalue = 1.94 Eta2 = .66 Wilks' Lambda = .34 chi2 = 291.95" D.F. = 24 % Classified Correctly = 89.82% F to enter Discriminant Function‘ 2.39" .05 Discriminant Falction Usim Best variables in This Analysisc .63 Hourly Salary" -.27 Satis. with the Manner in which Work Fills time" .21 Pension Availability2" -.22 Sex3.. -.17 Age. -.17 Satis. with Pay- .14 Life Insurance Avail. .13 Number of Vacation Days. .16 Health Insurance Avail.2" -.16 Org. Compliancen .11 Overall Job Satisfaction 2.. .14 Satis. with the Type of Work I 00.. Eigenvalue = 1.90 Eta2 = .65 Wilks' Lambda = .35 ch12 = 294.55" D.F. = 12 % classified correctly = 33.06% 119 Table 13 (Continued) ‘ This is the F-to-enter after all preceding variables (control and satisfaction variables) have been forced to enter the equation. The default F-to-enter for SPSS must be at least 1.00 or larger. 1 F values of 1.59 and 1.92 correspond to independent significance eve s of .05 and .01, respectively, for this size sample. ” These are the standardized canonical discriminant function coefficients after all variables have been forced to enter in the order specified by the research model (Figure 3). ° These are the standardized canonical discriminant function coefficients after a stepwise procedure was used to onl enter the best discriminating variables at each step and to reeva uate and possibly remove variables entered in previous steps. This procedure maximizes the difference between part- and full-time employees using the fewest number of discriminating variables. ‘ Full-timers are coded as 1 and part-timers are coded as 0. There were 168 part—timers and 117 full—timers in this analysis. 2 If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0 3 Males are coded as I and females are coded as 2. 120 Table 14 Discriminant Function Analysis Predicting Work Status Using Organizational Compliance and Organizational Commitment (OCQ) F to enter Discriminant Function‘ Org. Conpliance 2.77" Org. Comnit. Questionnaire .64 Discriminant Faiction Using Best Variables in This Discrilimmt Falction Usim All Variibles Entered in This Analysish Analysis" .68 Hourly Salary“ .65 Hourly Salary" -.27 Satis. with the Manner in -.24 Satis. with the Manner which Work Fills Up One's 11me" in which Work Fills 11me" .22 Pension Availability2" .20 Pension Availabilityzn -.19 Sex3" -.20 Sex3 -.13 Age" -.13 Age " -.16 Satis. with Pay" -.15 Satis. with Pay" .15 Life Insurance Avail.2" .15 Life Insurance Avail.2" .15 Nuhber of Vacation Days .12 Number of Vacation Days" .13 Health Insurance Avail. .14 Health Insurance Avail.2' -.13 Org. Compliance“ -.16 Org. Compliance“ .12 People -- 001 Code' .10 Satis. with the Type of .18 Satis. with the Type of work 1 Do Work 1 no" -.09 Data -- DOT Code .08 Things -- DOT Code' .09 Things -- DOT Code -.08 Company Tenure .08 Overall Job Satisfaction -.05 Satis. with Supervision -.05 Intention to Turnover .04 Satis. with Social Interactions at Work .04 Org. Commitment (OCO) .03 Number of Sick Days .03 Satis. with Promotional Opportunities .01 Propensity to Search for Another Job or Retire .00 Profit Sharing Plan2 Eigenvalue = 1.92 Eta2 = .66 Eigenvalue =1.88 Eta2 = .65 Wilks' Lambda = .34 Wilks' Lambda = .35 Chi2 = 296.07" D.F. = 24 Chiz = 293.45" 0.r. = 12 X Classified Correctly = 90.00% % Classified Correctly = 90.55% 121 Table 14 (Continued) ' This is the F-to-enter after all preceding variables (control and satisfaction variables) have been forced to enter the equation. The default F-to—enter for SPSS must be at least 1.00 or larger. 1 F ¥alues of 1.59 and 1.92 correspond to independent significance eve s of .05 and .01, respectively, for this size sample. ” These are the standardized canonical discriminant function coefficients after all variables have been forced to enter in the order specified by the research model (Figure 3). ° These are the standardized canonical discriminant function coefficients after a stepwise procedure was used to onl enter the best discriminating variables at each step and to reeva uate and possibly remove variables entered in previous steps. This procedure maximizes the difference between part— and full—time employees using the fewest number of discriminating variables. Full-timers are coded as 1 and part-timers are coded as 0. There were 172 part-timers and 118 full—timers in this analysis. If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0 Males are coded as 1 and females are coded as 2. ' p g .05 a ' p g .01 122 enter the best discriminating variables at each step (irrespective of the order in which they appear in the research model) and to reevaluate variables entered in previous steps and possibly remove them after new variables have entered the equation. While this type of analysis was not related to any of the hypotheses, it was conducted because of its practical value; it seeks to provide maximum explanation of the differences between part— and full-time employees using the fewest number of discriminating variables. The F-to-enter value presented at the top of each table provides evidence of whether each measure of commitment would significantly discriminate between part- and full—timers after all control variables and satisfaction scale scores have already been entered into the discriminant function. Hence, the F-to—enter values provide the tests of hypotheses 3a, 3b, and 3c, since they reflect mean differences in each measure of commitment after all control and satisfaction variables have been partialled from the commitment scores. An F—to-enter value of 1.59 and 1.92 corresponds to independent significance levels of p g .05 and p g .01, respectively, for the sample size in the present study. However, since some of the commitment measures in the separate discriminant function analyses were highly correlated and yielded the same information, somewhat higher F—to-enter values were adopted in an attempt to adjust for the fact that the tests of statistical significance were not completely independent. The data presented at the bottom of each table provides some information about how well each discriminant function (i.e., hierarchical and stepwise) distinguishes between older part- and full- time employees. The eigenvalue is simply the ratio of the between- groups to within—groups sums of squares. Large eigenvalues are 123 associated with "good" functions. Eta2 is the ratio of the between- groups sum of squares to the total sum of squares and represents the proportion of the total variance attributable to differences among the groups. The square root of this value is a measure of the degree of association between the discriminant scores and the groups. This value is called the canonical correlation. Wilks’ Lambda is the ratio of the within-groups sum of squares to the total sum of squares. A lambda of 1 occurs when all observed group means are equal. Values close to 0 occur when most of the total variability is attributable to differences between the means of the groups. Thus, large values of lambda indicate the group means do not appear to be different, while small values indicate that group means do appear to be different. Wilks’ Lambda can be transformed to a variable which has approximately a chi-square distribution. A significance test can then be performed to determine whether in the populations from which the samples are drawn there is a statistically significant difference between the group means. Finally, the percent of cases classified correctly by the discriminant function provides another measure of the function’s effectiveness. This provides evidence of how well the function would classify cases into part- and full-time groups (e.g., where 50-51 percent of the cases would be classified correctly on the basis of chance in the current study depending on the measure of organizational commitment which was used to classify the cases). Hypothesis 3a suggested that part—timers would express a significantly higher level of organizational compliance than full- timers. There was strong support for this hypothesis, since the F-to- enter for organizational compliance is at least 2.77 in Tables 12 through 14, which is larger than the p g .05 and p 5 .01 F-to-enter 124 values for independent significance tests. Since part-timers are coded as 0 and full-timers are coded as 1, the negative sign on the standardized canonical discriminant function coefficient for organizational compliance provides evidence that the significant difference is in the hypothesized direction. Furthermore, organizational compliance was one of the variables that entered the stepwise discriminant function analysis in Tables 12 through 14, suggesting that it was one of the best discriminating variables. Since the F-to-enter for organizational identification was only .02 in Table 12, the F-to-enter for organizational internalization was only .05 in Table 13, and the F-to-enter for the OCQ was only .64 in Table 14, there was no support for hypotheses 3b or 3c which predicted that older full-timers would express significantly higher levels of identification, internalization, and commitment on the OCQ than older part-timers.6 In fact, examination of mean differences in Table 5 indicates that part-timers had higher levels of internalization, identification, and scores on the OCQ than full-timers, although these differences were small. The lack of support for hypothesis 3c suggests that in this sample there was no difference between part- and full- timers’ expressed levels of organizational commitment on the 0C0, a measure of commitment which has been used by many other researchers and cited frequently in the literature. Table 6 and Tables 12-14 suggest that a few significant covariates account for a large amount of the variance in the differences between older part- and full-timers. The part-timers received a significantly lower hourly salary and less fringe benefits than the full-timers. 6The F-to-enter values for these three measures were still not statistically significant when they were not entered with organizational conpliance. 125 Part-timers were also significantly older and comprised of a greater proportion of females than the full-timers. Statistical power is lost with the addition of each covariate in the analysis. Therefore, statistical results supporting any of these hypotheses (e.g., support for hypothesis 3a) are quite strong. The covariate analysis presented here and elsewhere suggests that future studies examining differences between older part- and full-timers may proceed by measuring fewer covariates; hourly salary, age, sex, and presence of a pension plan appear to do a sufficient job of capturing the major differences between older part- and full—timers. Hypothesis 4a. Hypothesis 43 states that organizational compliance should significantly mediate the relationship between overall job satisfaction and withdrawal intentions (intention to be absent and intention to turnover) for older part-timers but not for older full- timers. Since the intention to be absent measure was discarded for its low internal consistency, the hypothesis was only examined for the intention to turnover measure. The test was conducted in four steps. First, I examined whether work status (i.e., part- or full-time) significantly moderated the relationship between overall job satisfaction and organizational compliance. Table 15 demonstrates that it did not. This table and Table 10 provide evidence that while overall job satisfaction and organizational commitment were significantly correlated, this relationship was negative (3 = -.20; 8 = —.19; p g .01). This suggested that those employees who exhibited the least organizational compliance were the most satisfied, and vice versa. This finding contradicted my expectations. 126 Table 15 Regressing Organizational Compliance on Control Variables, Overall Job Satisfaction, and Work Status Moderator Step 1 Step 2 Step 3 l!§ £232 £232 £232 Stgp 1 -- Control Variables .. 1. Sex .20.. -.20 -.20 2. Hourly Salary -.26. -.16 -.16 3. Life Insurance2 -.19. -.17 -.17 4. People (DOT Code% -.18 -.17 -.18 5. Health Insurance .20 .17 .15 6. Age -.11 -.O9 -.08 7. Company Tenure .15 .10 .11 8. Number of Vacation Days .04 .07 .08 9. Data (DOT Code) .11 .12 .12 10. Number of Sick Days -.07 -.07 -.07 11. Pension (Retirement) Plan2 -.07 -.04 -.O4 12. Things (DOT Code) -.03 -.O3 -.03 13. Profit Sharing Plan2 .03 .06 .06 Step 2 -- Main Effects . 14. Overall Job Satisfaction -.19 ' -.25 15. Work Status (P.T. or F.T.)3 -.13 -.34 Stgp 3 -- Interaction Effects 16. Work Status x Overall Job Satisfaction .23 F 2.28 2.92 2.82 df 13,292 15,290 16,289 32 .092._ .131_. .135 Change in 32 .092 .039 .004 1 Males are coded as 1 and females are coded as 2. 2 If this work benefit is present, it is coded as 1. Otherwise, it is coded as O. 3 Full-timers are coded as 1 and part-timers are coded as O. :.p 5 .05 p 5 .01 127 Second, Table 16 demonstrates that work status did not significantly moderate the relationship between organizational compliance and intention to turnover. Compliance was not even significantly related to intention to turnover, since the change in R2 over the variance accounted by the control variables was not statistically significant. This provided some preliminary evidence that hypothesis 4a would have to be rejected, since compliance would have to be significantly related to intention to turnover in order for it to mediate the satisfaction—turnover relationship. An examination of whether work status moderated the direct relationship between overall job satisfaction and intention to turnover was conducted in the third step. Table 17 demonstrates that this moderator was statistically significant and accounted for slightly more than 2% of the variance in turnover intentions over and above the amount of variance explained by the control variables and facets of satisfaction. Hence, the significance of this moderator appears to be of a magnitude which can be regarded as quite meaningful. Using the beta weights for work status, overall job satisfaction, and the moderator term presented in Table 17 to plot this relationship, Figure 4 demonstrates that the regression line for part—timers has a negative slope, while the regression line for full-timers has a slight positive slope. If an employee’s intention to turnover is highly predictive of actual turnover, this could suggest that attempting to influence older part-timers’ levels of overall job satisfaction may be as important if not more important than influencing full-timers’ levels of overall job satisfaction if one is trying to influence future turnover. 128 Table 16 Regressing Intention to Turnover on Control Variables, Organizational Compliance, and Work Status Moderator Step 1 Step 2 Step 3 QLS. 3.93 ms 19;! 555p 1 -- Control Variables 1. Hourly Salary .09 .16 .16 2. Pension (Retirement) Plan1 '.13 -.11 -.11 3. Profit Sharing Plan‘ -.20" -.20 -.20 4. Number of Vacation Days -.19 -.19 -.19 5. Data (DOT Code) -.08. -.O9 -.O9 6. Sex2 .14 .16 .16 7. Things (DOT Code) .09 .10 .10 8. Age .08 .08 .08 9. Life insurance1 .07 .10 .10 10. Company Tenure .15 .13 .13 11. Health insurance1 -.06 -.07 -.07 12. Number of Sick Days .06 .07 .07 13. People (DOT Code) -.06 -.04 -.O4 §§ep 2 -- Mpin Effects 14. Organizational Compliance .11 .12 15. Work status (9.1. or f.1.)3 -.07 -.07 Stpp 3 -- Interaction Effggts 16. Work Status X Organizational Compliance -.01 F 1.86 1.92 1.80 df 13,289 15,287 16,286 112 .077_ .091 .091 Change in 32 .077 .014 .000 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0. 2 Males are coded as 1 and females are coded as 2. 3 Full-timers are coded as 1 and part-timers are coded as 0. :.p g .05 p g .01 129 Table 17 Regressing Intention to Turnover on Control Variables, Overall Job Satisfaction, and Work Status Moderator Step 1 Step 2 Step 3 Lb: 39A 9.932 M Stpp 1 -- Control Variables 1. Hourly Salary .09 .19 .20 2. Pension (Retirement) Plan1 -.13_. -.10 -.11 3. Profit Sharing Plan1 -.20 -.17 -.17 4. Number of Vacation Days -.19 -.16 -.14 5. Data (DOT Code) -.08. -.07 -.O6 6. sex2 .14 .16 .15 7. Things (DOT Code) .09 .10 .09 8. Age .08 .13 .13 9. Life Insurance1 .07 .08 .08 10. Company Tenure .15 .10 .12 11. Health Insurance1 -.06 -.11 -.14 12. Number of Sick Days .06 .06 .06 13. People (DOT Code) -.06 -.O7 -.08 §£gp 2 -- Mpin Effects .1 14. Overall Job Satisfaction -.27 -.42 15. Work Status (3.1. or f.1.)3 -.09 -.53 §£§p 3 -- Interpption Effects 16. Work Status x Overall Job .- Satisfaction .54 F 1.86 3.33 3.67 df 13,289 15,287 16,286 R2 .077. .143. .170.. Change in R2 .077 .071 . .022 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as O. 2 Males are coded as 1 and females are coded as 2. 3 Full-timers are coded as 1 and part-timers are coded as O. :.p 5 .05 2:5-01 130 H 0" a." "II'II.’ ‘89""I"'II"II'|' IIII Illa" '5'. a8- .' 5'l"'-"l '|-| '3' I8- III! '1' ale- III _ _ 6 4 EXPECTED .14 Full-'timers _ _ 2 0 TURNOVER ~2.52 2 4 _ _ INTENTIONS 6 _ High OVERALL JOB SATISFACTION Figure 4 Regression Lines of Overall JOb Satisfaction on Expected Turnover Intentions for Part- and Full-Timers 131 While satisfaction had a negative statistically significant relationship with turnover as expected (8 = -.27; p g .01), it was not expected that satisfaction would have a stronger direct relationship with turnover in Table 17 than it had with organizational compliance in Table 16. This suggested that the ordering of the variables in the model may have been misspecified. The fourth and last step described below examined whether this was true. The top half of Table 18 displays the results of first regressing intention to turnover on organizational compliance, and then adding overall job satisfaction followed by the overall job satisfaction x work status interaction term to the equation. If compliance mediated the satisfaction-turnover relationship, the addition of satisfaction to the equation should not provide a statistically significant increase to the change in R2. The bottom half of Table 18 reverses the order in which compliance and satisfaction entered the regression equation. If compliance mediated the satisfaction-turnover relationship, the addition of compliance to the equation in Step 4 should provide a statistically significant increase to the change in R2. The results presented in Table 18 provide evidence to the contrary. Overall job satisfaction significantly mediated the relationship between organizational compliance and intention to turnover, instead of compliance mediating the satisfaction-turnover relationship as expected. Furthermore, the negative relationship between satisfaction and turnover intentions is stronger for part-timers than for full-timers. While I believe these results are interesting and I will discuss them below, they provide evidence that hypothesis 4a must be rejected. 132 Table 18 Exploring Whether Organizational Compliance Mediates the Relationship Between Overall Job Satisfaction and Intention to Turnover Step 1 Step 2‘ Step 3‘ Step 4 ile 2232 2222 92:2 2223 St 1 -- Control Variables 1. Hourly Salary .09 .20 2. Pension (Retirement) Plan1 -.13.. -.11 3. Profit Sharing Plan1 -.20 -.17 4. Nulber of Vacation Days -.19 -.15 5. Data (DOT Code) -.O8. -.O7 6. Sex2 .14 .17 7. Things (DOT Code) .09 .09 8. Age .08 .13 9. Life Insurance1 .07 .09 10. Company Tenure .15 .11 11. Health insurance1 -.06 -.15 12. Number of Sick Days .06 .06 13. People (DOT Code) -.06 -.07 Step 2 . 14. Organizational Compliance .12 .07 .06 Step 3 .. 15. Overall Job Satisfaction -.26 -.41 16. Work Status -.08 -.57 Stgp 4 17. Overall Job Satisfaction X _. Work Status .53 F 1.86 2.02 3.20 3.51 df 13,289 14,288 16,286 17,285 R2 .077 .039 .152 .173 Change in R2 .077' .012' .062" .021" 1:33 Table 18 (Continued) Step 1 Step 2‘ Step 3' Step 4 L!§ 2232 52:5 2232 22:2 Sggp 1 -- Control Variables 1. Hourly Salary .09 .20 2. Pension (Retirement) Plan1 -.13“ -.11 3. Profit Sharing Plan1 -.20 -.17 4. Number of Vacation Days -.19 -.15 5. Data (DOT Code) -.08. -.07 6. sex2 .14 .17 7. Things (DOT Code) .09 .09 8. Age .08 .13 9. Life Insurance1 .07 .09 10. Company Tenure .15 .11 11. Health Insurance1 -.06 -.15 12. Number of Sick Days .06 .06 13. People (DOT Code) -.06 -.07 Step 2 1. 14. Overall Job Satisfaction -.27 -.42 -.41 15. Work Status -.09 -.58 -.57 §ssié 16. Overall Job Satisfaction X Work Status .54" .53 Step 4 17. Organizational Compliance .06 F 1.86 3.33 3.67 3.51 df 13,289 15,287 16,286 17,285 it2 .077. .143_. .170.. .173 Change in R2 .077 .071 .022 .003 ' Only Step 1 and Step 4 beta weights for the control variables appear since these are the only weights that remain unchanged (and are relevant to both analyses (i.e., Satis ---> Commit ---> Turnover and Commit ---> Satis ---> Turnover) 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0. 2 Males are coded as 1 and females are coded as 2. :.p g .05 p 5 .01 134 Hypothesis 4b. This hypothesis suggested that organizational identification and internalization each significantly mediate the relationship between overall job satisfaction and intention to turnover for full-timers but not for part-timers. Tables 19 and 20 demonstrate that work status did not significantly moderate the relationship between satisfaction and identification or satisfaction and internalization, respectively. Both measures of commitment, however, had a positive statistically significant relationship with overall job satisfaction as expected. Tables 21 and 22 also demonstrate that work status did significantly moderate the relationship between each measure of commitment and turnover. Figures 5 and 6 display the plots of these respective relationships. Like the plot which was presented in Figure 4, Figures 5 and 6 demonstrate that the negative relationship between expected turnover intentions and measures of organizational identification and internalization are significantly stronger for older part-timers than for older full—timers. The question of whether identification and internalization mediate the relationship between satisfaction and turnover was examined next. Table 23 explores this question for organizational identification and Table 24 examines this issue for organizational internalization. Since the results presented in Table 23 demonstrate the the addition of satisfaction in Step 4 of the top half of the table provided a statistically significant increase in R? over and above identification as well as the identification x work status interaction and the addition of identification in Step 4 of the bottom half of the table provided a statistically significant increase in R? over and above satisfaction as well as the satisfaction x work status interaction, there was no evidence 1235 Table 19 Regressing Organizational Identification on Control Variables, Overall Job Satisfaction, and Work Status Moderator Step 1 Step 2 Step 3 m ass gets Bets §£gp 1 -- Control Variables _. 1. Age .27 .14 .14 2. Pension (Retirement) Plan1 .09 .05 .05 3. Data (DOT Code) .10 .06 .07 4. People (DOT Code) -.04 -.02 -.02 5. Number of Vacation Days -.12 -.17 -.17 6. Life insurance1 .03 .07 .07 7. Things (DOT Code) '05.. .05 .05 3. Profit Sharing Plan1 .19 .12 .12 9. Hourly Salary .13 .01 .01 10. Number of Sick Days -.03 -.O3 -.O4 11. Company Tenure -.05 .05 .06 12. Sex .00 -.05 -.05 13. Health -.05 .09 .08 §§gp 2 -- Mpin Effects .1 14. Overall Job Satisfaction .58 .54 15. Work status (P.T. or 3.1.)3 -.04 -.16 Sggp 3 -- Interpption Effects 16. Work Status X Overall Job Satisfaction .14 F 2.69 13.68 12.86 df 13,292 15,290 16,289 R2 .107_. .414.. .416 Change in R2 .107 .303 .001 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0. 2 Males are coded as 1 and females are coded as 2. 3 Full-timers are coded as 1 and part-timers are coded as 0. p g .01 136 Table 20 Regressing Organizational Internalization on Control Variables, Overall Job Satisfaction, and Work Status Moderator 1y; Bet Step 1 -- Control Variables 1. Number of Vacation Days -.19 2. Pension (Retirement) Plan1 .14“ 3. Age .20 4. Data (DOT Code) .10 5. People (DOT Code) -.07 6. Things (DOT Code) .10 7. Health Insurance1 -.19 8. Company Tenure .09 9. Life Insurance .08 10. Profit Sharing Plan .17 11. sex2 .07 12. Hourly Salary .10 13. Number of Sick Days .00 Step 2 -- Mpin Effects 14. Overall Job Satisfaction 15. work Status (P.1. or f.1.)3 Step 3 -- Interpption Effects 21. Work Status X Overall Job Satisfaction r 230 df 1L283 R2 JMB“ Change in R2 .095 1 is coded as 0. Step 1 Step 2 Beta Beta -.23 -.23 .11 .11 .09 .09 .07 .07 -.05 -.05 .10 .10 -.04 -.05 .19 .19 .07 .07 .11 .11 .03 .03 .00 .00 -.01 -.01 .54" .52 -.05 -.09 .05 10.53 9.34 15,232 16,281 -359.. .359 .264 .000 If this work benefit is present, it is coded as 1. 2 Males are coded as 1 and females are coded as 2. Otherwise, it 3 Full-timers are coded as 1 and part-timers are coded as 0. Step 3 137 Table 21 Regressing Intention to Turnover on Control Variables, Organizational Identification, and Work Status Moderator Step 1 Step 2 Step 3 m EELO 22 9:2 Stgp 1 -- Control Variables 1. Hourly Salary .09 .17 .16 2. Pension (Retirement) Plan1 -.13._ -.10 -.10 3. Profit Sharing Plan1 -.20 -.15 -.17 4. Number of Vacation Days -.19 -.21 -.20 5, Data (DOT Code) -.08. -.06 -.05 6. Sex2 .14 .14 .13 7. Things (DOT Code) .09 .11 .12 8. Age .08 .13 .15 9. Life Insurance1 .07 .10 .08 10. Company Tenure .15 .13 .15 11. Health Insurance1 -.06 -.O6 -.06 12. Number of Sick Days .06 .05 .06 13. People (001 Code) -.06 -.07 -.03 §ggp 2 -- Mgin Effects .. 14. Organizational Identification -.25 -.39 15. work status (P.1. or f.1.)3 -.1o -.53 Sypp 3 -- Interpption Effects 16. Work Status X Organizational .. Identification .47 F 1.86 3.01 3.26 df 13,290 15,288 16,287 R2 .077. .136.. .154.. Change in 32 .077 .053 .013 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as O. 2 Males are coded as 1 and females are coded as 2. 3 Full-timers are coded as 1 and part-timers are coded as O. :.p g .05 p 5 .01 138 Table 22 Regressing Intention to Turnover on Control Variables, Organizational Internalization, and Work Status Moderator Step 1 Step 2 Step 3 w m Let; M St 1 -- Control Variables 1. Hourly Salary .08 .16 .16 2. Pension (Retirement) Plan1 -.14" -.09 -.1O 3. Profit Sharing Plan1 -.22 -.19 -.19 4. Number of Vacation Days -.17 -.20 -.19 5. Data (DOT Code) -.O7. -.06 -.06 6. Sex .14 .15 .15 7. Things (DOT Code) .10 .12 .12 8. Age .08 .10 .11 9. Company Tenure .16 .17 .17 10. Life Insurance .07 .09 .08 11. Health Insurance1 -.05 -.07 -.O8 12. Number of Sick Days .05 .05 .05 13. People (DOT Code) -.06 -.07 -.07 Sggp 2 -- Mgin Effects .. 14. Organizational Internalization -.19 -.31 15. Work Status (P.T. or P.1.)3 -.11 -.52 Sygp 3 -- Interpction Effects 16. Work Status X Organizational . Internalization .44 F 1.92 2.54 2.64 df 13,282 15,280 16,279 R2 .031. .120.. .132. Change in R2 .031 .033 .012 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as O. 2 Males are coded as 1 and females are coded as 2. 3 Full-timers are coded as 1 and part-timers are coded as O. :.p 5 .05 p g .01 SUM Commit ---> Turnover and Conlnit ---> Satis ---> Turnover) 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0. 2 Males are coded as 1 and females are coded as 2. 2.2 5 -°5 p 5 .01 143 Table 24 Exploring Whether Organizational Internalization Mediates the Relationship Between Overall Job Satisfaction and Intention to Turnover Step 1 Step 2‘ Step 3‘ Step 4‘ Step 5 HS mammm §£pp 1 -- Control Variables 1. Hourly Salary .08 .19 2. Pension (Retirement) Plan1 -.14 -.11 3. Profit Sharing Plan1 -.22" -.13 4. Number of Vacation Days -.18 -.15 5. Data (DOT Code) -.07 -.05 6. Sex2 .14' .15 7. Things (DOT Code) .10 .11 8. Age .16 .14 9. Life Insurance1 .07 .12 10. Company Tenure .07 .08 11. Health Insurance1 -.05 -.13 12. Number of Sick Days .05 .05 13. People (DOT Code) -.06 -.08 Step 2 14. Organizational Internalization 1.19" -.31 -.19 -.12 15. Work Status -.11 -.52 -.51 -.68 Step 3 16. Organizational Internalization X Work Status .44 .44 .15 Step 4 17. Overall Job Satisfaction -.22" -.35 Step 5 18. Overall Job Satisfaction X Work Status .48 F 1.91 2.52 2.62 3.16 3.25 df 13,281 15,279 16,278 17,277 18,276 R2 .031. .119_. .131. .162_. .175. change in R2 .031 .033 .012 .031 .012 VS §£gp 1 -- Control Variables 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. Hourly Salary Pension (Retirement) Plan1 Profit Sharing Plan1 Number of Vacation Days Data (DOT Code) Sex2 Things (DOT Code) Age Life Insurance‘ Company Tenure Health Insurance Number of Sick Days People (DOT Code) Step 2 14. 15. Overall Job Satisfaction Work Status Step 3 16. Overall Job Satisfaction X Work Status Stgp 4 17. Organizational Internalization Step 5 18. Organizational Internalization x Work Status 144 Table 24 (Continued) Step 1 Beta Step 3‘ Step 4‘ Step 5 Beta Beta Beta -.41 -.37 '.35 -.60 -.61 '.68 L91 13gm1 .mn. Jm1 16,278 3.55 3.43 3.25 17,277 18,276 .169. .174 .175 .023 . .004 .001 ‘ Only Step 1 and Step 5 beta weights for the control variables appear since these are the only weights that remain unchanged (and are relevant to both analyses (i.e., Satis ---> Commit ---> Turnover and Commit ---> Satis ---> Turnover) ‘ If this work benefit is present, it is coded as 1. Otherwise, it is coded as O. 2 Males are coded as 1 and females are coded as 2. :.p 5 .05 p 5 .01 145 for a mediating relationship. Rather, satisfaction was positively related to identification and both satisfaction and identification were negatively related to intention to turnover, each accounting for a unique proportion of the variance in the turnover score. Therefore, satisfaction appeared to be directly related to both identification and turnover intentions. Once again, the relationships between satisfaction aggsggrnover intentions as well as identification and turnover intentions were significantly stronger for older part-timers than older full-timers. The results for organizational internalization were Similar to those which referred to organizational compliance in Table 19. The data in Table 25 clearly Show that satisfaction mediated the relationship between organizational internalization and intention to turnover instead of the expected relationship of internalization mediating the satisfaction-turnover relationship. Furthermore, the negative relationship between satisfaction and turnover intentions was stronger for the older part-timers than the older full-timers. These unexpected findings will be discussed shortly. Hypothesis 4c. Hypothesis 4c stated that Porter, et al.’s (1974) OCQ mediated the relationship between overall job satisfaction and intention to turnover for full-time employees. While no hypothesis was presented regarding this relationship for part-time employees, the possibility of such mediation was explored. Table 25 demonstrates that satisfaction had a statistically significant positive relationship with the OCQ, and work status did not moderate this relationship. The results presented in Table 26 show that the OCQ had a statistically significant negative relationship with turnover as expected. Additionally, this relationship was moderated by Regressing Organizational Commitment (OCQ) on Control Variables, Overall Job Satisfaction, and Work Status Moderator 146 Table 25 Step 1 Step 2 Step 3 l “J Eat—8 Est—8 Sgpp 1 -- Control Variables 1. Age .29" .15 .15 2. Pension (Retirement) Plan1 .14 .03 .09 3. Life Insurance1 .12 .11 .11 4. Data (DOT Code) .03 -.O1 -.01 5. Halber of Vacation Days -.12 -.17 -.18 6. Health Insurance‘ -.16 .OO .00 7. Hourly Salary .16 .01 .01 3. Sex2 .05 .00 .00 9. Number of Sick Days .05 .04 .04 10. Company Tenure .01 .12 .12 11. Profit Sharing Plan1 .13' .05 .05 12. Things (DOT Code) .01 .02 .02 13. People (DOT Code) .06 .08 .08 Step 2 -- Main Effects 14. Overall Job Satisfaction .64" .66 15. Work Status (P.T. or F.T.)3 .OO .08 Step 3 -- Interaction Effects 16. Work Status X Overall Job Satisfaction -.09 F 2.30 16.61 15.55 df 13,290 15,288 16,287 R2 .094 .464 .464 change in 32 .094" .370“ .001 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0. 2 Males are coded as 1 and females are coded as 2. 3 Full-timers are coded as 1 and part-timers are coded as O. 147 Table 26 Regressing Intention to Turnover on Control Variables, Organizational Commitment (OCQ), and Work Status Moderator Step 1 Step 2 Step 3 Hi Pet; 3.111! .3133 §£gp 1 -- Control Variables 1. Hourly Salary .09 .20 .19 2. Pension (Retirement) Plan1 -.13 -.O7 -.08 3. Profit Sharing Plan1 -.20" -.15 -.16 4. Number of Vacation Days -.18 -.23 -.21 5. Data (DOT Code) -.08 -.07 -.08 6. Sex2 .14' .16 .15 7. Things (DOT Code) .10 .11 .10 8. Age .08 .18 .20 9. Life insurance1 .07 .13 .10 10. Company Tenure .15 .15 .15 11. Health Insurance1 -.06 -.11 -.10 12. Number of Sick Days .06 .08 .08 13. People (DOT Code) -.06 -.O4 -.04 Step 2 -- Main Effects 14. Organizational Commitment (OCO) -.39" -.52 15. Work Status (P.T. or m.)3 -.09 -.73 Step 3 -- Interaction Effects 16. Work Status X Organizational Commitment (OCQ) .68 F 1.87 5.41 5.72 df 13,288 15,286 16,285 R2 .073 .221 .243 Change in R2 .078. .143" .022" 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as O. 2 Males are coded as 1 and females are coded as 2. 3 Full-timers are coded as 1 and part-timers are coded as O. p g .05 'p;g.01 148 work status. Figure 7 displays the plots of the separate regression lines for part- and full-timers. The part-timer regression line has a steeper negative slope, suggesting that the negative relationship between the OCQ and intention to turnover was significantly stronger for part-timers than full-timers. This may suggest that if one is attempting to influence intention to turnover, it may even be more important to influence part-timers’ levels of commitment on the OCQ than full-timers’ levels of commitment. This moderator term accounted for slightly more than 2% of the variance in turnover intentions over and above the amount of variance explained by the control variables, OCQ score, and work status variable. Hence, the significance of this moderator appears to be of a magnitude which can be regarded as quite meaningful Table 27 provides the actual test of the hypothesis by examining the mediating role of the OCQ. The statistically significant OCQ x work status and satisfaction x work status moderator terms described above were also included in this analysis. The first section of the table demonstrates that overall job satisfaction does not add a statistically significant increment to R? beyond the control variables, work status, and OCQ main effects as well as the OCQ x work status moderator term. On the contrary, the second half of the table demonstrates that the OCQ adds a statistically significant increment to Rzlabove and beyond the amount accounted for by overall job satisfaction and the satisfaction x work status moderator. Therefore, not only is hypothesis 4c supported, but the data suggests that the OCQ mediates the relationship between satisfaction and turnover for part-timers as well as full—timers. This structural relationship is contrary to the findings obtained for identification and internalization. UMHOF‘IWNF‘I 50m Commit ---> Turnover and Commit ---> Satis ---> Turnover) 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0. 2 Males are coded as 1 and females are coded as 2. ' p g .05 .. pg .01 152 A graphical summary of the results obtained for Hypotheses 4a, 4b, and 4c is presented in Figure 8. The indicated paths between satisfaction, commitment, turnover intentions, and the moderating role of work status are displayed four times, once for each of the four measures of commitment which was explored above. The sign of each of the paths is denoted by a ’+’ or ’-’. Figure 8, which shows a subset of the larger exploratory model displayed in Figure 3, serves as a summary of the results presented above in Tables 15-27. Hypothesis 5a. This hypothesis proposed that organizational compliance would account for significantly more variance in turnover intentions among part-timers who think they would search for other jobs if they left their present jobs within five years than for those part- timers who think they would completely retire if they left their present jobs within five years. Unfortunately, the interaction term and its corresponding change in R2 in Table 28 were not statistically significant, so this hypothesis was rejected. Hypothesis 5b. Hypothesis 5b suggested that organizational identification, internalization, and Porter, et al.’s (1974) OCQ should each account for significantly more variance in intention to turnover among full-timers who think they would search for other jobs if they left their present jobs within five years than for those full-timers who think they would completely retire if they left their present jobs within five years. Tables 29, 30, and 31 present the results of the tests of this hypothesis for the measures of identification, internalization, and the OCQ, respectively. Since the interaction terms and corresponding changes in R2 were not statistically significant, this hypothesis must be rejected. While the power to test this hypothesis may be low since I did not obtain responses from 156 full-timers, the 153 Organizational. (') Overall (I) Turnover Compliancej Job lnieniions Satisfaction I Work Sloius (stronger for Port—Timers) Overall Job (_) Work Sioius (sironger for P.T.) Solisfoclion / Turnover Intentions Organizational (-) \ ldeniificoiion Work Siolus (stronger for P.T.) Orgonizoiionol (4.) Overolr (-) Turnover \' Job / lnleniions lniernolizoiion Solisfociion I -— ~— Work Sioius (stronger for P.T.) .---—_—----—-———-———-——t—_—-_——_—--—.—--——a—--_—“—-”p-— overall Job (4') Orgonizoiionol (- Turnover Solisfociion Commiimeni __. lnieniions Questionnaire l (SiY'voonrgel’SJrPdeT’J) Figure 8: Graphical summary of results obioined from hypotheses 4a, 4b, and 4c Regressing Intention to Turnover on Control Variables, Organizational Compliance, and Propensity to Search for Another Job or Retire Moderator QL §£gp 1 -- Control Variables 1. Things (DOT Code) 2. Sex1 3. Hourly Salary 4. Life Insurance 5. Number of Vacation Days 6. People (DOT Code) 7. Data (DOT Code) 8. Profit Sharing Plan2 9. Number of Sick Days 10. Health Insurance2 11. Company Tenure 12. Pension (Retirement) Plan2 13. Age 2 §£gp 2 -- H25" Effects 14. Organizational Compliance 15. Propensity to Search for Another Job or Retire3 §£§p 3 -- Interaction Effects 16. Propensity to Search for Another Job or Retire X Organizational Compliance F df R2 Change in R2 lli4 Table 28 for Part-Timers Step 1 Step 2 Step 3 Beta Beta Beta .15 .17 .20 .10 .10 .11 .01 .02 .03 -.06 -.02 -.01 -.15 -.15 -.20 -.12 -.09 -.09 -.07 -.05 -.06 -.12 -.12 -.12 .11 .12 .14 -.11 -.1o -.11 .07 .04 .05 .08 .03 .05 -.os -.05 -.06 .09 -.16 -.19' -.42 .37 1.05 1.43 1.54 13,164 15,162 16,161 .077 .117 .132 .077 .040' .016 1 Males are codes as 1 and females are coded as 2. 2 If this work benefit is present, it is coded as 1. Otherwise, it is coded as O. 3 A higher positive score signifies a greater intention to search for another job rather than retire. A higher negative score signifies a greater intention to retire rather than search for another job. 'gg.m 155 Table 29 Regressing Intention to Turnover on Control Variables, Organizational Identification, and Propensity to Search for Another Job or Retire Moderator for Full—Timers Step 1 Step 2 Step 3 l me 32 3.93. §£gp41 -- Control Variables 1. Age .28" .28 .28 2. Hourly Salary .17 .18 .18 3. Data (DOT Code) -.14 -.08 -.07 4. Pension (Retirement) plan1 -.26" -.23 -.24 5. sex2 .22 .20 .21 6. Profit Sharing plan1 -.3o" -.23 -.23 7. Halber of Vacation Days -.22 -.27 -.27 8. Company Tenure .22 .23 .22 9. Life Insurance1 .11 .10 .09 10. Number of Sick Days .04 .02 .01 11. Health Insurance1 .11 .09 .09 12. People (DOT Code) .03 -.01 -.01 13. Things (DOT Code) .02 .03 .05 Ste -- Hain Effects 14. Organizational Identification -.15 -.03 15. Propensity to Search for Another Job or Retire3 -.18 -.02 §tgp 3 -- Interaction Effects 16. Propensity to Search for Another Job or Retire x Organizational Identification -.25 F 2.47 2.76 2.62 df 13,108 15,106 16,105 a’ .229 .231 .285 Change in 32 .229" .052.. .005 i If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0. 2 Hales are codes as 1 and females are coded as 2. 3 A higher positive score signifies a greater intention to search for another job rather than retire. A higher negative score signifies a greater intention to retire rather than search for another job. 9 g .05 ' 9,5 .01 156 Table 30 Regressing Intention to Turnover on Control Variables, Organizational Internalization, and Propensity to Search for Another Job or Retire Moderator for Full-Timers Step 1 Step 2 Step 3 LV. 933 39$ BLte §££P 1 -- Control Variables 1. Age .29" .26 .26 2. Hourly Salary .17 .17 .18 3. Pension (Retirement) Plan1 -.26' -.24 -.26 4. Data (DOT Code) -.13 -.07 -.06 5. sex2 .20 .21 .21 6. Company Tenure .24 .28 .27 7. Profit Sharing plan1 -.3o" -.26 -.26 8. Number of Vacation Days -.22 -.28 -.29 9. Life insurance1 .11 .07 .07 10. Health Insurance1 .11 .10 .11 11. Number of Sick Days .03 .02 .02 12. People (DOT Code) .04 -.02 -.02 13. Things (DOT Code) .03 .05 .08 Step 2 -- Main Effects 14. Organizational Internalization -.08 .09 15. Propensity to Search for Another Job or Retire3 -.19' .15 Step 3 -- Interaction Effects 16. Propensity to Search for Another Job or Retire x Organizational Internalization -.42 F 2.47 2.59 2.53 df 13,106 15,104 16,103 :2 .232 .272 .282 Change in R2 .232" .040 .010 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0. 2 Males are coded as 1 and females are coded as 2. 3 A higher positive score signifies a greater intention to search for another job rather than retire. A higher negative score signifies a greater intention to retire rather than search for another job. 157 Table 31 Regressing Intention to Turnover on Control Variables, Organizational Commitment (OCQ), and Propensity to Search for Another Job or Retire Moderator for Full—Timers Step 1 Step 2 Step 3 Lye 3_9t_8 Lesa M2 §£SP 1 -- Control Variables 1. Age .29" .30 .30 2. Hourly Salary .17 .20 .21 3. Pension (Retirement) Pla -.26" -.21 -.23 4. Data (DOT Code) -.13 -.06 -.06 5. Sex2 .21 .22 .23 6. Profit Sharing Plan1 -.30“ -.23 -.22 7. Number of Vacation Days -.22 -.29 -.30 8. Company Tenure .22 .25 .25 9. Life insurance1 .10 .11 .11 10. Health Insurance1 .11 .07 .07 11. Number of Sick Days .03 .03 .02 12. PeOple (DOT Code) .04 -.01 .01 13. Things (DOT Code) .04 .07 .09 Stee 2 -- Hain Effects 14. Organizational commitment (one) -.24" -.09 1S. Propensity to Search for Another Job or Retire3 -.1a. -.21 Steg 3 -- Interaction Effects 16. Propensity to Search for Another Job or Retire x Organizational Commitment (0C0) -.44 F 2.44 3.23 3.12 df 13,107 15,105 16,104 :2 .229 .316 .324 Change in p2 .229" .087" .003 1 If this work benefit is present, it is coded as 1. Otherwise, it is coded as 0. 2 Males are coded as 1 and females are coded as 2. 3 A higher positive score signifies a greater intention to search for another job rather than retire. A higher negative score signifies a greater intention to retire rather than search for another job. 158 low increments to R2 associated with the interaction terms in these tables (i.e., R2 change = .005 — .010) suggests that none of these interaction terms would have been statistically significant even if many more subjects had been included in the analysis. For example, in order to obtain a statistically significant increase in R2 for the largest increment reported in any of these tables (i.e., .010 for the propensity x internalization interaction term in Table 29), approximately 300 older full—timers would have to be sampled. This assumes, of course, that responses of older full-timers who were not sampled would be similar to the responses provided by the older full—timers who did complete the questionnaire. DISCUSSION In this section, three basic issues are discussed. First, the findings are summarized and possible explanations for the occurrence of the results obtained are provided. Second, limitations of the present study are presented. Finally, implications for this study are provided and directions for future research suggested. Summary of Results The present investigation explored several research issues. First, an examination was conducted to determine if several facets of job satisfaction would be significantly related to overall job satisfaction, and if two of these facets (satisfaction with the perceived meaningfulness of work and the nature in which work fills up one’s time) would differentially account for a statistically significant amount of variance in overall job satisfaction for older part— and full-time employees. Second, the relationship between overall job satisfaction, organizational commitment, and turnover intentions was compared and contrasted using four different measures of organizational commitment. The presence of differential relationships between the measures of satisfaction, commitment, and turnover intentions for older part- and full-time workers was also explored. Third, mean differences on several —159- —160— demographic, job descriptive, and attitudinal measures were examined to determine if any of them would successfully discriminate between part— and full-time employment status. Finally, an examination was conducted to determine if the relationship between organizational commitment and turnover intentions was significantly stronger for those employees who were more likely to search for other positions if they left their present jobs than for those employees who were more likely to retire. Results indicated that statistically significant bivariate relationships existed between each of the facets of satisfaction (pay, supervision, promotional opportunities, perceived meaningfulness of work, social interactions, and manner in which work fills one’s time) and overall job satisfaction. However, when all of these facets and the control variables were entered into a single regression equation, satisfaction with promotional opportunities, supervision, and social interactions no longer accounted for statistically significant amounts of variance in overall job satisfaction. The amount of variance in overall job satisfaction accounted for by satisfaction with the manner in which work fills one’s time and satisfaction with the perceived meaningfulness of work was not significantly different for older part- timers than for older full—timers. Several demographic and job descriptive differences between the part— and full-timers in this sample were discovered when all the demographic, job descriptive, and attitudinal measures were entered into a single discriminant function analysis. Part—timers were more likely to be female, older (even among a sample of employees 55 years of age or —l61— older), and receive a lower hourly salary and fewer job benefits than full-timers. The part-timers expressed more compliance towards their organizations but did not differ from full-timers in their levels of organizational identification, internalization, or scores obtained from the Organizational Commitment Questionnaire. Part—timers were also more satisfied with the manner in which work filled up their time but slightly less satisfied with the perceived meaningfulness of their work. However, there were no statistically significant differences between part— and full-timers’ reported levels of overall job satisfaction. Organizational commitment as measured by the Organizational Commitment Questionnaire significantly mediated the relationship between overall job satisfaction and turnover intentions for all older employees. The negative relationship between the OCQ and intention to turnover was significantly stronger for older part—timers than older full-timers. Overall job satisfaction significantly mediated the relationship between commitment and turnover intentions for all employees when organizational compliance and internalization were used as measures of commitment. These mediated relationships were not identical, however, since compliance was negatively related to job satisfaction and internalization was positively related to job satisfaction. In both instances, satisfaction had a stronger negative relationship with turnover intentions for older part-timers than older full—timers. The fourth measure of commitment examined in the present study, organizational identification, had significant direct effects on both overall job satisfaction and turnover intentions. No mediated relationship was obtained with this measure of commitment. —162- Finally, relationships between all the measures of commitment and turnover intentions were not significantly different for those employees who were more likely to search for other positions if they left their present jobs than for those employees who were more likely to retire. Possible explanations why this and other pertinent hypotheses were not supported are provided below. Determinants of Overall Job Satisfaction While hypotheses were not offered regarding whether some of the facets of job satisfaction would be related to overall job satisfaction, it was expected that satisfaction with social interactions at work would be significantly related to overall job satisfaction for all older employees. However, the zero order correlation between satisfaction with social interactions at work and overall job satisfaction was only .19 for part—timers and .29 for full~timers. This hypothesis was not supported when satisfaction with each of the measured job facets plus the control variables were included in a single regression equation. Data presented in Table 5 demonstrates that part— and full—timers’ mean scores on satisfaction with social interactions at work were quite high (5.06 and 4.92, respectively) with fairly low standard deviations (.98 and .94, respectively). Therefore, it seems possible that this hypothesis was not supported because nearly all the respondents reported that they liked the people with whom they work. —163- Nhile satisfaction with social interactions may be an important facet of overall job satisfaction, it did not adequately discriminate between employees who reported overall satisfaction or dissatisfaction with their jobs in the current sample. For instance, while 28% of the full—time respondents were slightly to strongly dissatisfied with their jobs, only eight percent of the full—time respondents were slightly to moderately dissatisfied with their social interactions at work. Similarly, while 21% of the part-time respondents were slightly to moderately dissatisfied with their jobs, only seven percent of the part— time respondents were slightly to moderately dissatisfied with their social interactions at work. It is possible that this hypothesis would have been supported if there had been greater variance on the measure of satisfaction with social interactions at work. This support might be accomplished in three ways in future research. First, additional items might be added to this scale in an effort to increase its reliability and variance. Second, the wording of items like A7 ("I like the people with whom I work") might be altered so they do not promote socially desirable responses (e.g., 57% of the respondents who answered item A7 chose response option six). For example, this item might be changed to something like, "I would prefer to work with other people instead of the people with whom I currently work." Of course, making significant changes to the content of items could hinder one’s ability to make meaningful generalizations to results obtained by others who have used similar items from frequently cited inventories such as the Job Descriptive Survey (Smith, et al., 1969) or the Quality of Employment -164- Surveys (e.g., Quinn & Shepard, 1974; Quinn & Staines, 1979). Additionally, the type of data collected in the present study does not provide any information regarding whether the items tended to promote socially desirable responses or whether the respondents who completed the questionnaire really were very pleased with the people with whom they worked. Finally, more variance (and probably a lowering of the mean) could be introduced by including data from employees who appear to be more dissatisfied with their social interactions at work. Future research might attempt to incorporate supervisory ratings of how well employees appear to get along with their coworkers. This information would not only help locate employees who are dissatisfied with social interactions at work, but would also provide a way of assessing whether supervisors agree with employees’ responses on the satisfaction with social interactions at work scale. This assessment could be useful in estimating the effects of social desirability response bias and method bias resulting from self-report data. It was also hypothesized that satisfaction with the perceived meaningfulness of work (the type of work I do) would be more strongly related to overall job satisfaction for older full-timers than older part-timers. The fact that this facet was significantly and positively related to overall job satisfaction for all the respondents suggests that the perceived meaningfulness of work was as essential to the part— timers in the current sample as it was to the full-timers. This finding is important, because it suggests that the older part-timers in the ~165- current sample would have been less satisfied with their jobs if they had been given tasks which were less significant and involving. Even though the part-timers spent less time on their jobs per week, the perceived meaningfulness of their work was still as much a component of their overall job satisfaction as it was for the full-timers. If the results from this sample can be generalized to other samples of older employees, it suggests that managers should not be tempted into giving older part-timers more tedious and less involving tasks than older full— timers simply because part—timers spend less time at work. If they have the required skills and abilities, many older part-timers may be more satisfied with their jobs if they are given additional work responsibilities, provided that these responsibilities do not require an additional time commitment to work. Finally, it was hypothesized that satisfaction with the manner in which work occupies one’s time would be more strongly related to overall job satisfaction for older part-timers than older full—timers. Failure to find support for this hypothesis was surprising, especially since older part-timers expressed significantly more satisfaction with this facet than older full—timers (see Table 5) and the internal consistency reliability of this scale was low for the older full-timers (.49). Perhaps the failure to gain support for this hypothesis resulted from the limited sample of occupations represented in the present study. Table 1 demonstrated that only seven respondents stated that they work more than 45 hours per week. Full—time employees in professional white— collar occupations who typically work 50 or more hours per week might be more likely to find that work does not give them time to pursue other -166- desired activities. The full-timers in the present study may have still found that their work hours were flexible enough to be quite satisfying. Therefore, future research might reexamine this hypothesis with older full—timers employed in more demanding occupations which tend to require greater time commitments. Differences Between Part— and Full-Timers’ Levels of Organizational Compliance The part-timers did not exhibit significantly less identification, internalization, or commitment (as measured by the OCQ) than the full- timers as hypothesized earlier. However, the part-timers expressed significantly more organizational compliance than their full—time counterparts. These results seem to suggest that while the part— and full-timers expressed equal organizational involvement based on a desire for affiliation and involvement predicated on congruence between individual and organizational values, the part-timers expressed greater instrumental involvement for specific, extrinsic rewards. The part- timers received fewer of these rewards (i.e., lower hourly salary and fewer fringe benefits) than the full-timers (see Table 6). When these differences in salary and fringe benefits were statistically controlled, the difference between part- and full-timers’ levels of organizational compliance reported in Table 5 was no longer statistically significant. This suggests that when employees receive the same salary and valued benefits, they feel equally motivated to work hard. An examination of the retained items in the organizational compliance scale (BZ,B4,BS) -167- suggests that part-timers might have been somewhat less motivated to work hard than full-timers because the part-timers did not feel they were justly rewarded for conducting hard work. Given that the perceived task complexity of part— and full—timer jobs was not significantly different (see Table 6), part—timers may have felt they were not treated as equitably as full-timers regarding their pay and fringe benefits. Perhaps the older part-timers sampled here would have worked harder (or claim to have worked harder) if they received more benefits. However, it is interesting to note that the part-timers’ levels of organizational commitment as measured by identification, internalization, and OCQ are as high as the full—timers’ levels despite the fact that the part- timers received fewer benefits and a lower salary. Clearly, the part— timers in this study reported that they felt as deeply (or superficially) attached to their organizational goals as their full- time coworkers, although they were less likely to exert extra effort on the job for their present levels of renumeration. Relationship Between Overall Job Satisfaction, Organizational Commitment, and Turnover Intentions Three of the four measures of organizational commitment did not mediate the relationship between overall job satisfaction and turnover intentions as expected. Some possible explanations of why unexpected relationships between these measures were obtained are presented below. -168- Reconsidering what the OCQ measures. The OCQ was the only measure of commitment which mediated the relationship between overall job satisfaction and turnover intentions, replicating previous results using the OCQ (e.g., Nilliams & Hazer, 1986). Despite the fact that this relationship was expected in the present study, it is somewhat surprising that the data supported this hypothesis since the OCQ includes some items measured by compliance (e.g., 816), identification (e.g., Bl7 & 821), and internalization (e.g., 820), and none of these three measures were statistically significant mediators of the satisfaction - turnover relationship. Apparently, the OCQ must measure something else besides whatever is measured by these three other instruments in order to account for its statistically significant direct relationship with turnover intentions. Some researchers have suggested that the relationship between the OCQ and turnover intentions is strong because the OCQ includes items which measure intention to turnover. This results in a great deal of overlap and redundancy between these two constructs, and obviously increases their statistical relationship (Morrow, 1983; Reichers, 1985). However, when these redundant items were eliminated from the OCQ in the present study’, the OCQ was still a statistically significant mediator of the satisfaction-turnover relationship. Furthermore, while the elimination of these four redundant items resulted in the satisfaction — OCQ relationship accounting for 20% instead of 24% percent of the variance in turnover intentions reported in Table 27, it still explained 7Items 822, 824, 826, and BBO were eliminated from the OCQ score for this analysis. -169— more variance in turnover intentions than the compliance — satisfaction (17% -see Table 18), identification - satisfaction (18% — see Table 23), or internalization - satisfaction (16% — see Table 24) relationships. What is left of the OCQ after the compliance, identification, internalization, and turnover intention items (i.e., B16, 817, 820, B21, 822, 824, 826, 830) are removed? My impression is that the remaining scale of seven items could just as easily (and perhaps more appropriately) be labelled "organizational satisfaction" instead of organizational commitment. For instance, items B19, 825, and 829 appear to measure a concept similar to global or overall organizational satisfaction, since they ask employees if they are satisfied that they chose their present work organizations over others. Items 823 and B27 seem to measure a particular facet of organizational satisfaction: satisfaction with organizational policies and the degree to which these policies allow one to work efficiently. While the two remaining items (818 & 828) may measure commitment more than satisfaction since they refer to concepts like loyalty and caring about an organization’s fate, my belief is that behavioral indicators of these concepts would be better measures of commitment than these attitudinal measures which appear to beg for socially desirable responses. While the OCQ was developed to include both attitudinal and behavioral components of commitment (Mowday, et al., 1982), I believe that we need to redefine this concept and focus only on the behavioral components measured through sidebets (e.g., Becker, 1960), extrarole behaviors (e.g, Smith, Organ, & Near, 1983), and/or other behavioral —l70— indices. I am not convinced that measures of attitudinal components of commitment are distinct from measures of satisfaction. Furthermore, behavioral components of commitment measured through a paper and pencil instrument like the OCQ appear to measure turnover intentions as much as they measure commitment. All of these problems result in severe construct redundancy between job satisfaction and commitment as well as commitment and turnover intentions. If we accept the argument that the OCQ measures organizational satisfaction, we have gathered preliminary support for the following research model in the present study: Overall Job Satis. —--> Organizational Satis. ---> Turnover Intentions. This model appears plausible, especially since Porter et al. (1974) developed the OCQ to represent a global evaluative link between the employee and the organization, with job satisfaction among the OCQ’s specific components. They further speculated that satisfaction would be associated with aspects of the work environment and thus would develop more quickly than commitment (i.e., organizational satisfaction as measured by the OCQ), which would require a worker to make a more global assessment of his/her relationship to the organization. They believed that this rapid cause of job satisfaction would suggest that it is a cause of commitment (i.e., organizational satisfaction), rather than vice versa. This model might also help explain why the negative relationship between the OCQ and turnover intentions was significantly stronger for older part—timers than older full-timers. Table 6 demonstrated that the -l7l- older part-timers in this study received significantly lower hourly salaries and fewer fringe benefits, and had less job and company tenure than the older full—timers. In addition to working fewer hours per week, these findings may suggest that part-timers are less invested in their jobs/companies and more likely to turnover if they are dissatisfied than full-timers. Older full—timers are more likely to stand to lose pension plans and high salaries than older part-timers. Older part—timers may even be more likely to turnover if they perceive more alternative employment opportunities than older full—timers, since it seems reasonable to me that older people have a better chance at finding another part-time job than another full—time job. However, the data collected in the present study cannot test this assumption. The question still remains why organizational compliance, identification, and internalization did not mediate the overall job satisfaction — turnover intentions relationship. Some possible explanations are offered in the next section. Organizational compliance, identification, and internalization. Many researchers have suggested that job satisfaction is the result of congruence between one’s job-related values and his/her experiences on the job (e.g., Locke, 1969, 1976; Porter & Steers, 1973). A research model of withdrawal behavior proposed by Steers and Mowday (1981) has even suggested that congruence between one’s job expectations and values and job experiences result in positive affective responses to the job, and these affective responses, in turn, have a negative relationship with one’s withdrawal intentions. Reviews of research findings have provided modest support for these relationships by demonstrating that —l72- unmet expectations and values have an ultimate impact on turnover (e.g., Mobley, et al., 1979; Muchinsky & Tuttle, 1979; Porter & Steers, 1973; Wanous, 1977). Reexamination of the items comprising the organizational internalization scale suggests to me that it might measure satisfaction with how well one thinks his/her job related values have been met more adequately than it measures a component of organizational commitment. Therefore, in retrospect, it seems logical to me that responses on the organizational internalization scale measure the degree of perceived congruence between one’s job related experiences and his/her job related values, suggesting why it was found to be a statistically significant predictor of overall job satisfaction in the present study. This measure of organizational internalization was even significantly related to overall job satisfaction when it was included in a single regression equation with the other facets of satisfaction and all the control variables. However, since many of the other facets also significantly accounted for unique amounts of variance in overall job satisfaction in this equation, it appears that overall job satisfaction may comprise more important components of the job satisfaction construct than organizational internalization does. This might explain why overall job satisfaction is a statistically significant direct predictor of turnover intentions, mediating the relationship between organizational internalization and intention to turnover. Organizational identification was highly correlated with internalization, supporting a recent study which provided the results of -173- a factor analysis where these two measures loaded on the same factor but on a different factor than organizational compliance (Caldwell, Chatman, & O’Reilly, 1989). Despite the high correlation between internalization and identification, however, the latter measure did not mediate the satisfaction - turnover intentions relationship in the present study. Instead, both overall job satisfaction and organizational identification had statistically significant direct relationships with turnover intentions when all these measures were included in the same regression equation. A possible explanation for this occurrence may be that the organizational identification measure more closely resembles a self- reported index of past behavior, while organizational internalization only measures one’s attitudes. I believe that talking up an organization to one’s friends is a better proxy measure of one’s commitment (or, at least, organizational satisfaction) than simply stating that one’s values are in sync with an organization’s values. Since future behavioral intentions may be more directly related to previous behaviors than previous attitudes (Ajzen & Fishbein, 1980), it seems logical to me that the measure of identification would account for a statistically significant unique amount of variance in turnover intentions over and above the amount explained by overall job satisfaction. In retrospect, it also seems reasonable to expect satisfaction to account for a statistically significant unique amount of variance in turnover intentions over and above the amount explained by -174- organizational identification. Perhaps one may feel pride and talk up his/her organization to others for specific reasons (e.g., "we do more business than all of our competitors"), but still be dissatisfied with important facets of work (e.g., "the pay is lousy" or "my work hours are not as flexible as I would like"). Since overall job satisfaction encompasses many facets which are likely to be considered before one decides to turnover, it seems reasonable to me that overall job satisfaction would also have a statistically significant relationship with turnover intentions when the identification, satisfaction, and intention to turnover measures are all included in a single regression equation. Organizational compliance did not mediate the satisfaction - turnover intentions relationship as expected. However, as I mentioned earlier, the measure of organizational compliance may reflect one’s perception of equity in terms of the salary and fringe benefits s/he receives. While this perception may have an impact on his/her attitude toward a job and an organization, it may have little influence on an older employees’ turnover intentions. For instance, while an older part-time employee may not perceive a great deal of equity in terms of the salary and fringe benefits s/he receives in comparison to full-time employees, his/her need for income in order to maintain a desired standard of living may be a more important factor in deciding whether s/he continues working or withdraws from the organization. It is my impression that the specific job facets comprising overall job satisfaction could be more important than "perceived equity" in an older employee’s determination of whether to continue working, especially -l75- since many alternative employment opportunities may not be available. This may explain why only overall job satisfaction is a significant predictor of turnover intentions when compliance, satisfaction, and intention to turnover measures are all included in a single regression equation. Of course, the data gathered in the present study is not sufficient to test this proposition. Perhaps future research should examine this issue in greater depth. Commitment 0C Is a Relevant Construct for Older Part—Timers Whether the OCQ really measures commitment or organizational satisfaction, the fact of the matter is a great deal of previous research has demonstrated that the OCQ significantly mediates the relationship between overall job satisfaction and turnover intentions. This relationship was once again supported in the present study. Although the results obtained for the mediated relationships regarding compliance, identification, and internalization were conflicting and difficult to interpret, we cannot ignore the fact that far less research has been conducted with these measures. Therefore, I believe that the results relating to the OCQ are more interesting and important in the present study than the results relating to the other measures of commitment. Perhaps of most interest to future researchers in this area will be the significantly stronger mediated effect of the OCQ for part—timers. We would intuit the relationship to be stronger for full—timers, since -l76- Katz and Kahn’s notion of partial inclusion and our own stereotypes of part-time workers often limit our thinking of organizational commitment as a construct only relevant to full-timers. However, these results suggest that organizational commitment may even be mere relevant to older art-timers in terms of predicting turnover intentions. Apparently, older part-time workers (and maybe other types of part— timers) are not simply interested in punching a time clock and collecting a weekly paycheck. Managers need to realize that treating older part-timers with less respect than full-timers may result in high turnover. Despite the fact that older part-timers spend less time at work, this study suggests that their perceived commitment towards the organization is a powerful indicator of whether or not they intend to stay. Propensity to Search for Another Job or Retire The lack of support for many of the hypotheses suggesting a "commitment" - turnover intentions relationship was a signal that propensity to search for another job or retire would not significantly moderate the relationship between organizational commitment and intentions to turnover. Post hoc analyses demonstrated that this variable did not significantly moderate the relationship between overall job satisfaction and turnover intentions. This would lead us to conclude that this moderator variable was completely useless. However, it is possible that the sample utilized in the present study did not provide an adequate test of this hypothesis. -l77- It was suggested earlier that this moderator may be important to examine because older employees who express a greater likelihood that they would turnover to completely retire rather than find another job might be more satisfied with their present jobs and committed to their respective organizations but desire withdrawal for other reasons more directly related to "abilities" (e.g., health difficulties, transportation problems, etc.). Perhaps the sample of older employees in the present study did not adequately test this assumption, since Table 6 demonstrated that few of the respondents missed any days of work in the last three months for any of these reasons. Therefore, those employees who stated that it was more likely they would retire rather than search for another job in the present study seem to have made this choice based on reasons other than the "abilities" listed above. Given the limited amount of data collected in the present study, it is not possible at this time to suggest what many of these reasons might have been. However, based on the exploratory interviews conducted by this researcher and described in Appendix A, some of these important factors may have included financial status, non-work hobbies and interests, and belief that another employment opportunity would be available”. It is important to emphasize, however, that the subjects who were interviewed by this researcher were not the same people who responded to the survey. Hence, it is impossible to state with any confidence exactly what the 8Although item C8 asks employees to choose between retirement and taking another readily available job , it is unknown if the respondents really accepted the statement that another job would be available if they left their present jobs. —l78- respondents’ motives were regarding their choice to retire versus search for other jobs. Limitations of the Study The present study was an initial attempt to compare and contrast the work attitudes of older part- and full-time employees. However, there are some limitations to the study because of the type of sample and research design that were utilized. These limitations are discussed below. Sample Older part- and full-timers employed in several retail/sales/grocery corporations participated in this study. Hhile data obtained from several corporations may increase the generalizability of results over data obtained from only one company, a failure to specifically recognize differences in corporate policies and cultures may have resulted in a threat to the internal validity of the obtained findings. However, few (if any) companies employ enough older part- and full-time employees to provide all the respondents needed in order to have sufficient power to examine hypotheses similar to those proposed here. Therefore, it is my belief that utilization of multiple convenience samples will remain a necessity in order to conduct field research of this type at the present time. -179- Another limitation arose from the extreme difficulty of finding organizations to participate in this study. Since managers agreed to be involved only if the study required minimal commitment from their staffs, and since the researcher had little control over the distribution of the questionnaires, it was not possible to collect any data to determine if the respondents were significantly different from the nonrespondents in any meaningful and important ways. Finally, it may not be possible to generalize the results obtained from this study to older part- and full-timers employed in other types of occupations. The part- and full-timers in the present study performed similar kinds of work, and this may not be true of part- and full-time workers employed in manufacturing, technological, or professional white-collar positions. The present findings may not even generalize to future samples of older part— and full-time workers employed in retail/sales/grocery organizations if these future employees have attained greater levels of education and previous work experiences which may affect their job expectations and values. Of course, all of the explanations presented earlier regarding the success and failure of the data to support the hypotheses in the present study should only be tentatively accepted; much more future research would need to be conducted on older part— and full-time employees before these findings can be accepted with a great deal of confidence. This is especially true of any unexpected findings, which should be replicated in future research in order to minimize the role sampling error might have played in accounting for these results. -180- Research Design It is not possible to calculate the extent to which response bias may have spuriously inflated the correlations derived in the present study. Some response [or method] bias undoubtedly resulted from the use of just self-report paper and pencil measures, all of which were administered to the same individuals. Method bias may not have been a severe problem in this study since some of the obtained relationships were significantly moderated by work status. In other words, method bias would not in and of itself account for a significantly larger spurious correlation between two variables for one group of subjects than for the other group of subjects. Nevertheless, the presence of this bias in data of this type cannot be avoided. The use of just paper and pencil instruments to measure all the relevant variables in this study may have caused an unnecessary high degree of conceptual redundancy between the job satisfaction and organizational commitment constructs. Earlier it was suggested that future research should emphasize the behavioral components of commitment in studies incorporating measures of both satisfaction and commitment. This may help us make finer distinctions between these two constructs in the future. Furthermore, it is my belief that studies utilizing paper and pencil ratings of satisfaction in conjunction with behavioral indices of commitment derived from observations of others and/or archival organizational data could substantially reduce method bias. -181— The "intention to stay home from work" measure developed by this researcher for the present study was very poor. While it was my impression that the items measured the same content domain and would exhibit high internal consistency reliability, obviously the respondents proved that these assumptions were incorrect. Lindell (1989) has offered an explanation why these items might not have been highly correlated. He suggested that item C4 measures an intention of future behavior, item C5 is a general statement of (past) behavioral regularity, and item C6 is an endorsement of role expectations. Perhaps, then, each item really requires the respondent to focus on different aspects of intentions to attend/stay home from work. This could explain why they were not internally consistent items and precluded my ability to examine this second type of withdrawal behavior in the present study. Additionally, lack of variance on the present three items resulted in a scale with low reliability. My impression is that organizational attendance records should be used as a measure of absenteeism in future research whenever possible. It may be important, however, to follow-up the collection of this type of data with personal interviews in an attempt to make a distinction between voluntary and involuntary incidents of absenteeism. The results obtained in the present study are further limited by the fact that a hypothesized longitudinal process (i.e., overall job satisfaction -—-> organizational commitment -——> turnover intentions) was examined using data from measures that were all collected at the same time. While it is more difficult and time consuming to conduct longitudinal analysis, it is my belief that it is certainly more -182— appropriate for researching process models of this type. A recently developed analytic technique called survival analysis would be an appropriate way to examine longitudinal data since it allows the intensity or rate of the dependent variable to vary over time rather than remain fixed. Morita, Lee, and Mowday (1989) have demonstrated how survival analysis could be applied to a turnover model such as the one proposed here. Finally, it is important to emphasize that while the results of one study may lend support for a particular causal model, other unconsidered research models may also support the pattern of obtained results (Cohen & Cohen, 1983). While future researchers may attempt to replicate some of the results presented in this study, they are encouraged to challenge and extend our body of knowledge by comparing and contrasting the present research model with competing models in an attempt to better understand the turnover process among older employees. Implications and Future Directions Overall, this study provided some preliminary evidence that job attitudes expressed by older part- and full-time employees are quite similar despite the differences in pay and fringe benefits they receive. Furthermore, the relationship between overall job satisfaction, the Organizational Commitment Questionnaire, and turnover intentions in the present study replicates findings obtained by other researchers examining younger full—time employees. This suggests that the -183- withdrawal process in older part— and full-time employees may be the same process that is frequently observed in younger full-time employees. Any generalization to younger employees, however, must be assumed since they were not used as a comparison group in the present study. We can best learn more about older workers by directly comparing them with younger workers. Future research need to move in this direction if we are to form a better understanding of the needs of tomorrow’s "more senior" workforce. It may also be informative to conduct studies comparing the job— related needs and desires of older part-time workers with the job- related needs and desires of some of the other groups of part—time workers mentioned at the outset of this paper. Hhile teenagers, working mothers, and senior citizens may all hold part—time jobs for different reasons, no one has conducted comparative research on these groups of employees. My impression is that the organizational compliance scale may not be a good indicator of organizational commitment, but it may prove to be an interesting way of examining different groups of part— timers’ perceptions of the degree to which they are justly and equitably rewarded for their efforts. This could be useful in order to determine if particular groups of workers (e.g., part-time teenagers) tend to feel they are treated more unjustly by their organizations than other groups of part-timers. Of course, future research examining the construct validity of this three item organizational compliance scale should first be conducted before research hypotheses incorporating this measure are developed. -184- Additional research might also examine which employees are typically used as "comparison others" when older part—timers form perceptions of equity and justice. The idea of comparing oneself to others is a central theme of many psychological theories such as social information processing theory, social cognition theory, and equity theory of motivation. It may be useful to understand if older part-timers base their perceptions of equitable treatment by comparing themselves to other older workers or younger part-time workers in the same organization, or even to older part—time workers employed by different organizations. It is my belief that research examining this issue might provide some insight into the process through which older part—time employees develop their job expectations. Finally, emphasis on paper and pencil measures of organizational commitment should be replaced by more research on behavioral indices of commitment. However, even behavioral representations of the construct should not be utilized until this construct is more clearly defined. Reichers (1985) has indicated that a single definition of commitment that applies equally well to all employees may not be realistic. Instead, employees may experience several different commitments to the goals and values of multiple groups within and outside of the organization. This would suggest that a global measure of organizational commitment such as the OCQ may not be very precise or meaningful. Reichers believes that development and progress in commitment research should include an attempt to understand commitment from -185- subjects’ own perceptions and definitions of the concept. Similarly, it is my impression that initial exploratory interviews with subjects may help researchers develop creative behavioral indices of commitment that are tailored to specific organizations and differentiate between employees who supervisors and coworkers would rate as strongly and weakly committed. Following this course of action should help us draw a clearer distinction between the concepts of job satisfaction and organizational commitment. If we cannot successfully make this distinction, I believe we should reconsider the utility of further research on the organizational commitment construct. APPENDICES APPENDIX A INTERVIEW FINDINGS APPENDIX A INTERVIEW RESPONSES Listed below are descriptions of some of the clients and employees of a state’s senior employment service. The researcher met the clients at their orientation meeting on July 21, 1988. One of the program’s employees was interviewed the following week. All the names have been changed below to maintain the clients’ and employees’ rights to confidentiality. Tony -- He is a senior citizen who directs this state agency. He says he likes what he is doing primarily because it is so different from what he used to do in his earlier career. He used to be a Transportation Manager for a chemical company in New York, and is quite pleased to be away from the "rat race." Heidi —- She is responsible for client intake at this state agency. She had always done secretarial work for family businesses, first working full—time in her father’s office and later working part—time in her husband’s home business. She has not had much schooling (she had to quit in the ninth grade because of illness). Her husband was quite ill -186— -187— before he died and all their savings were spent on medical expenses. Heidi found that she had to find a job shortly after his death in order to survive. She first worked as a cashier in a department store and took her current job when it became available. She likes this job because she "likes the people. [She gets to] see different people and different circumstances.“ The variety of the people and circumstances she encounters is what satisfies her the most. Dolly -- She used to be a restaurant hostess. Since her husband passed away, she feels a need to get out of the house and meet new people. Jill —- She has worked since 1941 in many different secretarial positions. Now that she has been retired for 2 1/2 years, she finds that Social Security "doesn’t stretch" to meet her expenses and allow her to live comfortably. Rena -— She retired from her job as a manager of a gift shop. She claims she needs something to do; she has always liked to keep busy. Many of her old friends have moved away so she feels somewhat lonely. She can always "use the money" from work, although this is not her prime motivation for securing employment. She definitely does not want a desk job; she wants a job where she can constantly meet new people. This is one of the reasons why she would like to consider another sales position. She also claims to want challenge from her work, not the "boredom associated with repetitious office work." However, she does not want another managerial position since she doesn’t want to become —188— "so involved [with her job] that it will be a 24-hour job [like she used to have]." Given the costs of transportation and the taxes taken out of her pay she wants to work at least 3 days each week. One day a week “is not worth the effort." Don —- Don just turned 55 and was previously an industrial oil salesman for approximately 20 years. His last employer was a recycler of motor oil and was put out of business by the Environmental Protection Agency. In addition to losing his job, he had back surgery last year which would make another sales position where he has to spend a great deal of time sitting in his automobile difficult. He did try such a job last autumn where he had to travel approximately 100 miles per day. It was "too much to take at that time." Pam -— Her mother always wanted to be a nurse, so she spent her earlier years going to nursing school and training only to find that she hated nursing. She tried office work but found she didn’t like it because she is a "people person." She has always enjoyed creating things, so she opened a flower and jewelry shop with a friend where she could use her creative talents. Unfortunately, the shop became "all work and no play." She wants to find a part—time job where she can be with people. Bob -- He had a good job until last October as a collective bargainer. He originally is from the Philippines and came here because his ex-wife was a nurse and felt she would have more work opportunities in the -189- United States. He is now 58 years old and doesn’t think he can handle a job where much travel is involved. Brenda -— She had been working for 35 years and, as a single mother, put two children through college without receiving any child support or alimony. She is a certified teacher and has held eight teaching positions (each for less than 2 years). She has also done some Public Relations work and has found that she is very creative. Since she does not have Social Security or any other retirement pensions, she must find a full—time job in order to "keep the bills paid." She wants to find a job where she can "grow.“ APPENDIX B SURVEY INSTRUMENT APPENDIX B Code# SURVEY OF SENIOR EMPLOYEES General Instructions In this survey you will be asked how satisfied you are with your current job, how committed you are to the company or corporation for which you work, and your future intentions to leave your job. Often, when employees are asked such questions, they automatically state what they think their employers would want to hear. We wish to remind you that THIS SURVEY IS WRITTEN AND SPONSORED BY RESEARCHERS AT MICHIGAN STATE UNIVERSITY, NOT BY YOUR EMPLOYER. YOUR EMPLOYER WILL NEVER SEE YOUR INDIVIDUAL RESPONSES. FOR THIS REASON, WE WILL NEVER ASK YOU FOR YOUR NAME. Therefore, we want you to report your true feelings and opinions. We will start by asking you how satisfied you are with various aspects of your job. Please follow the instructions printed below. —190— -l91- SECTION A FACETS OF JOB SATISFACTION Please indicate the extent to which each of the following statements applies to you. Although some of the items may appear repetitious, we ask that you provide an answer to every item in the survey. Use the following response categories: 1 = Strongly disagree 4 = Slightly agree Moderately agree 2 = Moderately disagree 5 3 = Slightly disagree 6 = Strongly agree Please tell us how much you agree or disagree with each statement by using the above response categories (i.e., by answering each statement with one of the six numbers that best describes your agreement with that particular statement). Do NOT write "Yes" or "No" as an answer to any of the items. Satisfaction with Pay Al. I feel satisfied with my salary ........................... A2. My salary adequately meets my financial needs ............. A3. My salary is less than I deserve for the work I do ........ —192- Satisfaction with Social Interactions at Work A4. People at work generally upset me ......................... A5. The people with whom I work are friendly .................. A6. The people with whom I work take a personal interest in me. A7. I like the people with whom I work ........................ Satisfaction with the Manner in Which Work Fills up One’s Time A8. My job does not give me time to do other things I want to do ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo A9. I like the flexibility of my work hours ................... A10. I am happy that my job gives me something to do with my time. ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo Satisfaction with Supervision All. My supervisor praises me for doing good work .............. -193- A12. I am satisfied with the amount of feedback I receive from my supervisor ................................................ A13. My supervisor is competent in his/her job ................. _etisfaction with Promotional Opportunities A14. I am satisfied with the promotional opportunities available to me OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO A15. There are not enough promotional opportunities available to me ......................................................... A16. I am not moving ahead in my company as fast as I would like ........................................................... Satisfaction with the Type of Work I Do A17. I wish I could conduct more varied types of activities in my job 00000000000000000000000000000000000000000000000000000000000 A18. The work I do on my job is respected by others ..................................................... A19. A20. A21. A22. A23. A24. -194- I am proud to say that the work I do has an impact on others I am happy that my job lets me produce something from beginning to end, with visible results ................................. I feel that the job I do is important to the company for which I work ...................................................... I perform a variety of satisfying activities at work ...... The work I do is routine and boring ....................... I am pleased that my job lets me complete a whole identifiable piece of work ............................................. Overall Job Satisfaction A25. A26. A27. My present job is very close to an ideal job I would like to have ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo My job is very much like the job I wanted when I took it .. All things considered, I am completely satisfied with my job ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo —195- A28. In general, I like my job very much ....................... SECTION 8 FACETS 0F ORGANIZATIONAL COMMITMENT Once again, please indicate the extent to which each of the following statements applies to you. These questions regard feelings you have about the particular organization (company) for which you are now working. Please continue to use the following response categories: 1 = Strongly disagree 4 = Slightly agree 2 = Moderately disagree 5 = Moderately agree 3 = Slightly disagree 6 = Strongly agree Organizational Compliance 81. How hard I work for my work organization is directly linked to how much I am rewarded ........................................ 82. Unless I’m rewarded for it in some way, I see no reason to expend extra effort on behalf of my work organization ............ B3. In order for me to get rewarded at work, it is necessary to express the right attitude ................................ —196- 84. If I were paid more money I would probably work harder in this organization .............................................. 85. In order to get me to work harder for this organization it is necessary to give me more rewards ......................... 86. I do not think I should ever have to work overtime in this organization unless I am paid for it ...................... 87. I do what I am supposed to do in this organization largely because I am rewarded for it ...................................... Organizational Identification 88. I am proud to tell others that I am part of the organization where I work .................................................... 89. I talk up my employer to my friends as a great organization to work for .................................................. 810. I feel a sense of "ownership“ for my work organization rather than just being an employee .................................... -197— Organizational Internalization 811. If the values of my work organization (that is, the company for which I work) were different, I would not be as attached to the organization .............................................. 812. My attachment to my work organization is based primarily on the similarity of my values and those represented by the organization ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo 813. What my work organization stands for is important to me ... 814. Since joining my work organization, my personal values and those of the organization have become more similar .............. 815. The reason why I prefer my work organization to others is because of what it stands for, its values ......................... Organizational Commitment 816. I am willing to put in a great deal of effort beyond that normally expected in order to help my work organization be successful ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo 817. I talk up my work organization to my friends as a great organization to work for .................................. 818. 819. 820. 821. 822. 823. 824. 825. 826. -198— I feel very little loyalty to my work organization ........ I would accept almost any type of job assignment in order to keep working for my present work organization .................. I find that my values and my organization’s values are very similar ................................................... I am proud to tell others that I am part of my work organization ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo I could just as well be working for a different organization as long as the type of work was similar ...................... My work organization really inspires the very best in me in the way of job performance .................................... It would take very little change in my present organization to cause me to leave my present work organization ............ I am extremely glad that I chose my work organization over others I was considering at the time I joined .................... There’s not too much to be gained by sticking with my present work organization indefinitely ................................. 827. 828. 829. 830. —199— Often, I find it difficult to agree with my work organization’s policies on important matters relating to its employees ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo I really care about the fate of my work organization ...... For me, my work organization is the best of all possible organizations for which to work ........................... Deciding to work for my organization was a definite mistake on my part ...................................................... SECTION C WITHDRAWAL INTENTIONS Once again, please indicate the extent to which each of the following statements applies to you. These questions regard feelings you have about staying home from work and/or leaving your job. Please continue to use the following response categories: 1 = Strongly disagree 4 = Slightly agree 2 = Moderately disagree 5 = Moderately agree 3 = Slightly disagree ' 6 = Strongly agree -200- Intention to Turnover C1. Taking everything into account, I shall make a genuine effort to find a new job with another employer within the next year ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo C2. I intend to still be working for my company one year from now ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo C3. I intend to still be working for my company three years from now ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo Intention to Stay Home from Work C4. Provided I do not feel ill, I intend to come to work everyday I am scheduled to do so ........................................ C5. Sometimes I like to take a day off from work, even if I am supposed to work that day ................................. C6. I feel obligated to show up for work if I am on the schedule to do so and if I am not sick ................................... Propensity to Search for Another Job or Retire C7. If I were to leave my present job within the next five years, I would look for another job rather than retire ............. C8. C9. C10. —201- If I were to leave my present job within the next five years, I would retire even if another job were readily available .. Approximately how many days when you were scheduled to work during the last 3 months did you not show up for work? days (fill in number) Please provide all the reasons why you missed the number of days you answered to item C9 by using the table below. Fill in the number of days you missed for each of the reasons listed in the table. The total number of days you missed should equal your answer to C9. Remember, your employer will not have access to this information. Number of days missed because: Number of days Surgery/Hospitalization ............... Actual sickness (flu, bad cold, etc.) ........................... I had sick days coming to me (even though I was not feeling ill). Transportation problem with getting to work ..................... Personal day (birthday, visit from out of town relative, etc.) .. l ooooooooooooooooooooooooo —202— I did not feel like going to work ............................... Appointment that couldn’t be scheduled on a day off (e.g., doctor, dentist, etc.) .................................................. Had to assist family member or neighbor at the last minute (e.g., had to babysit for grandchild, etc.) ..................... TOTAL EQUALS (Remember, the total should equal your response to item E9 above) SECTION D JOB DESCRIPTIVE INFORMATION The following questions refer to different features of your job and the length of time you have held your present job. Please answer all these questions with respect to your present job and employer and not to any different jobs you may have held in the past. D1. When did you start working for this company? month year DZ. D3. D4. 05. 06. D7. -203— When did you start working in your current job for this company? , 19 month year How many hours (on average) do you presently work each week on your job for which you are paid? ............... hours If you currently work part-time and you used to work full-time for this company, when did you start working part—time? month year If you are paid by the hour, what is your hourly salary? S If you are not paid by the hour, please answer the following: My last paycheck was s for (e.g., week, month,...) time period How many of the following are you entitled to per year?: Sick days and/or "personal" days (please put 0 if none) Vacation days (please put 0 if none) D8. D9. 010. -204- Please place a checkmark in the blank next to all the following benefits you currently receive in your job: Group health insurance plan Pension plan Group life insurance plan Pension plan Profit sharing plan Please think of all the employees with whom you work who: 1) are approximately your age, 2) hold similar job positions as you, and 3) work approximately the same number of hours per week as you. How many months would it take before a "good worker" who meets these three qualifications would get a promotion? (If most people who meet these qualifications do not receive promotions, please check that option below.) months Most people do not receive promotions Please place an X on the blank below next to the job title that best describes the type of work you do. If more than one title best describes your job responsibilities, place an X next to each of these titles. If you rotate jobs, check all that apply. Brief suggestions of some major job duties have been included next to all of these job titles to help you decide. Please read through all nine job titles on the next two pages before you choose: ~205- CASHIER - CHECKER (operates cash register to itemize and total customer’s purchases; reviews price sheets to note price changes and sales items; collects money from customer and makes change; may weigh items, bag merchandise, and redeem food stamps and promotional coupons) CASHIER - COURTESY BOOTH (cashes checks for customers and monitors checkout stations; cashes personal and payroll checks, provides information to customers; receives customer’s complaints and resolves complaints when possible; issues cash to stations and removes excess cash; audits cash register tapes; may compile reports, verify employee time records, and prepare payroll) CASHIER — GIFT WRAPPER (may perform duties of cashier - checker listed above but also be responsible for gift wrapping merchandise for pickup or mailed delivery) DEPARTMENT MANAGER (may perform tasks typically conducted by sales attendant, salesperson, and stock clerk (see descriptions below), but also assigns duties to workers and schedules lunch and break periods, work hours, and vacations; trains workers in store policies, department procedures, and job duties; assists sales workers in completing difficult sales; evaluates worker performance -206- and recommends retention, transfer, or dismissal of worker) MERCHANDISE DELIVERER (uses automobile or light truck to deliver merchandise to customer’s home or place of business; collects money from customers or signature from charge—account customers; may also sweep floors, runs errands, and wait on customers) SALES ATTENDANT (aids customer in locating merchandise, obtains merchandise from stockroom when not on floor, keeps merchandise in order, may mark or ticket merchandise, directs or escorts customer to fitting or dressing rooms or to cashier) SALESPERSON (same as sales attendant above, but you may also demonstrate the use of the merchandise and estimate the quantity of merchandise a customer would need. You would actually offer more personal assistance and advice to the customer than the sales attendant) STOCK CLERK (marks order form to order merchandise based on available shelf space, merchandise on hand, customer demand, or advertised specials; periodically counts merchandise to take inventory or examines shelves to identify which items need to be reordered or replenished; unpacks cartons and crates of merchandise, checking —207- invoice against items received; stamps or attaches prices on merchandise or changes price tags, referring to pricelist; stocks shelves with new or transferred merchandise; cleans display cases, shelves, and aisles; may pack customers’ purchases in bags or cartons; may carry packages to customer’s automobile) 9. OTHER -— if you have not seen your job title above, write it in the space below. Also please provide a few of your major job duties: D11. What is the name of the department in which you work (e.g., grocery, mens clothing, checkout, sporting goods, etc.)? Please write the name of the department on the line below: SECTION E BACKGROUND INFORMATION Please place your responses on the lines provided alongside each item, or choose the appropriate option for each of the following items. E1. E2. E3. E4. E5. E6. —208— What is your birthdate? month day year Sex: (please check one of the following) Male Female Race: (please check one of the following) Caucasian Black Asian Hispanic Other Current marital status: (please check one of the following) Married Never been married Divorced Legally separated Widowed Please check the highest level of education you have obtained: High school diploma or below College degree 1-3 years of college Postgraduate degree Please list below the last job you held (not including your current job), the starting and ending date of this job, and the reason why you left this job. TYPE OF JOB STARTING ENDING REASON WHY I LEFT JOB M -209- THANK YOU FOR TAKING THE TIME TO FILL OUT THIS SURVEY. PLEASE PUT THE SURVEY IN THE POSTAGE PAID ENVELOPE ADDRESSED TO MICHIGAN STATE UNIVERSITY AND DROP IN ANY MAILBOX. IF YOU HAVE QUESTIONS OR COMMENTS ABOUT THIS SURVEY, YOU MAY CALL AND ASK TO SPEAK WITH: Scott Cohen 517-349-7694 BIBLIOGRAPHY BIBLIOGRAPHY Ajzen, I. & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, N.J.: Prentice—Hall. Allen, R. E., Keaveny, T. J., & Jackson, J. H. (1979). A reexamination of the preferred job attributes of full—time and part-time workers. Journal of Management, 5, 213—227. Allport, F. H. (1933). Institutional behavior. Chapel Hill: University of North Carolina Press. Atchley, R. C. (1971). Retirement and leisure participation: Continuity or crisis? Gerontologist, 11, 13-17. Atchley, R. C. (1976). The sociology of retirement. New York: John Wiley and Sons. Bachman, J., et al. (1967). Youth in transition: Vol. 1. Ann Arbor, MI: Survey Research Center. Barrett, N. (1984). Part—time: Workers’ mandate. In D. Rothberg (Ed.), Part-time employment in America. McLean, Va.: Association of Part—time Professionals. —210— -211— Bateman, T. S. & Strasser, S. (1984). A longitudinal analysis of the antecedents of organizational commitment. Academy of Management Journal, 21, 95-112. Becker, H. S. (1960). Notes on the concept of commitment. American Journal of Sociology, 66, 32-42. Belsley, D. A., Kuh, E., & Welsch, R. E. (1980). Regression diagnostics: Identifying influential data and sources of collinearity. New York: John Wiley and Sons. Bornstein, J. M. (1983). The prediction of successful job placement for unemployed older workers receiving assistance from a senior employment service. Unpublished Masters Thesis. Michigan State University. Bowlby, J. (1982). Attachment. New York: Basic Books. Bradburn, N. M. & Sudman, S. (1979). Improving interview method and Questionnaire design. San Francisco: Jossey-Bass. Brown, M. (1969). Identification and some conditions of organizational involvement. Administrative Science Quarterly, 15, 346—355. Caldwell, D. F., Chatman, J. A., & O’Reilly, C. A. (1989). Building Organizational Commitment: A Multi-Firm Study. Manuscript under review. —212- Campbell, D. T. & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, 56, 81-105. Carsten, J. M. & Spector, P. E. (1987). Unemployment, job satisfaction, and employee turnover: A meta—analytic test of the Muchinsky model. Journal of Applied Psycholpgy, 12, 374—381. Clark, M. (1972). An anthropological view of retirement. In F. M. Carp (Ed.), Retirement. New York: Behavioral Publications. Clark, M. & Anderson, 8. G. (1967). Culture and aging: An anthropological study of older Americans. Springfield, Ill.: Charles C. Thomas. Cohen, J. (1977). Statistical power analysis for the behavioral sciences (Rev. Ed.). New York: Academic Press. Cohen, J. & Cohen, P. (1983). Applied multiple regressionlcorrelation analysis for the behavioral sciences (2nd Ed.). Hillsdale, N.J.: Lawrence Erlbaum Associates. Cook, T. D. & Campbell, D. T. (1979). Quasi-experimentation: Design and analysis issues for field settings. Boston: Houghton Mifflin. -213- Cotton, J. L. & Tuttle, J. M. (1986). Employee turnover: A meta- analysis and review with implications for research. Academy of Management Review, 11, 55-70. Cooley, C. H. (1922). Human nature and the social order. New York: Scribners. Cronbach, L. J. (1960). Essentials of psychological testing (2nd Ed.). New York: Harper & Row. Crowne, D. & Marlowe, D. (1964). The approval motive. New York: Wiley. Deutermann Jr., W. V. & Brown, S. C. (1978,June). Voluntary part-time workers: a growing part of the labor force. Monthly Labor Review, 3- 10. Dictionary of Occupational Titles (4th Ed.) (1977). Washington, D.C.: U.S. Department of Labor. Dunham, R. 8. & Smith, F. J. (1979). Organizational surveys. Glenview, Ill.: Scott, Foresman. Edwards, A. L. (1970). The measurement of personality traits by scales and inventories. New York: Holt. Eisdorfer, C. (1972). Adaptation to loss of work. In F. M. Carp (Ed.), Retirement. New York: Behavioral Publications. —214— Eisenberger, R., Huntington, R., Hutchison, S., & Sowa, D. (1986). Perceived organizational support. Journal of Applied Psychology, 11, 500—507. Farrell, D. & Rusbult, C. E. (1981). Exchange variables as predictors of job satisfaction, job commitment, and turnover: The impact of rewards, costs, alternatives, and investments. Organizational Behavior and Human Performance, 28, 78-95. Ferris, K. R. & Aranya, N. (1983). A comparison of two organizational commitment scales. Personnel Psychology, 36, 87—98. Fiske, D. W. (1982). Convergent-discriminant validation in measruements and research strategies. In D. Brinberg & L. H. Kidder (Eds.), Forms of validity in research. San Francisco: Jossey-Bass. Frease, M. & Zawachi, R. A. (1979). Job sharing: An answer to productivity problems. The Personnel Administrator, 24, 35-39. Friedmann, E. A. & Havighurst, R. J. (1954). The meaning of work and retirement. Chicago: University of Chicago Press. Friedmann, E. A. & Orbach, H. L. (1974). Adjustment to retirement. In S. Anieti (ed.), American handbook of psychiatry (Vol. I). New York: Basic Books. -215- Gannon, M. J. & Nothern, J. C. (1971). A comparison of short—term and long-term part-time employees. Personnel Psychology, 25, 687-696. Glick, W. H., Jenkins, G. D., & Gupta, N. (1986). Method versus substance: How strong are underlying relationships between job characteristics and attitudinal outcomes? Academy of Management Journal, 29, 441—464. Gove, W. R. & Geerken, M. R. (1977). Response bias in surveys of mental health: An empirical investigation. American Journal of Sociology, 82, 1289-1316. Greenhaus, J. H., Seidel, C., & Marinis, M. (1983). The impact of expectations and values on job attitudes. Organizational Behavior and Human Performance, 31, 394—417. Hackett, R. D. & Guion, R. M. (1985). A reevaluation of the absenteeism - job satisfaction relationship. Organizational Behavior and Human Decision Processes, 35, 340-381. Hackman, J. R. & Oldham, G. R. (1975). Development of the Job Diagnostic Survey. Journal of Applied Psycholpgy, 69, 159—170. Hackman, J. R. & Oldham, G. R. (1976). Motivation through the design of work: Test of a theory. Organizational Behavior and Human Performance, lg, 250-279. -216- Hall, D. T. & Gordon, F. E. (1973). Career choices of married women: Effects on conflict, role behavior, and satisfaction. Journal of Applied Psychology, 55, 42—48. Hall, D. T. & Mansfield, R. (1975). Relationships of age and seniority with career variables of engineers and scientists. Journal of Applied Psychology, 59, 201—210. Hammer, T. H., Landau, J. C., & Stern, R. N. (1981). Absenteeism when workers have a voice: The case of employee ownership. Journal of Applied Psychology, 55, 561—573. Herzberg, F. (1966). Work and the nature of man. Cleveland: World Publishing Co. Hom, P. W. (1979). Effects of job peripherality and personal characteristics on the job satisfaction of part-time workers. Academy of Management Journal, 2;, 551—565. Hrebiniak, L. G. & Alutto, J. A. (1972). Personal and role—related factors in the development of organizational commitment. Administration Science Quarterly, 11, 555—572. James, L. R. & Brett, J. M. (1984). Mediators, moderators, and tests for mediation. Journal of Applied Psychology, 55, 307-321. -217— Johns, G. & Nicholson, N. (1982). The meaning of absence: New strategies for theory and research. Research in Organizational Behavior, A, 127-172. Jondrow, J. M., Brechling, F., & Marcus, A. (1983). Older workers in the market for part-time employment. National Commission for Employment Policy. Kahne, H. (1985). Reconceiving part-time work. Totowa, N.J.: Rowman & Allanheld. Katz, D. & Kahn, R. L. (1978). The social psychology of organizations (2nd Ed.). New York: John Wiley and Sons. Kelman, H. C. (1958). Compliance, identification, and internalization: Three processes of attitude change. Journal of Conflict Resolution, 2, 51-60. Klecka, W. R. (1975). Discriminant Analysis. In N. H. Nie, C. H. Hull, J. G. Jenkins, K. Steinbrenner, & D. H. Bent (Eds.), Statistical Package for the Social Sceinces (2nd Ed.). New York: McGraw—Hill. Koch, J. T. & Steers, R. M. (1978). Job attachment, satisfaction, and turnover among public sector employees. Journal of Vocational Behavior, 12, 119-128. —218- Kraut, A. (1975). Predicting turnover of employees from measured job attitudes. Organizational Behavior and Human Performance, 15, 233-243. Lawler, E. E. & Hall, D. T. (1970). Relationship of job characteristics to job involvement, satisfaction, and intrinsic motivation. Journal of Applied Psychology, 53, 305—312. Leon, C. & Bednarzik, R. W. (1978,0ctober). A profile of women on part- time schedules. Monthly Labor Review, 3-12. Lindell, M. (1989). Personal communication. Locke, E. A. (1969). What is job satisfaction? Organizational Behavior and Human Performance, 5, 309—336. Locke, E. A. (1976). The nature and causes of job satisfaction. In M. D. Dunnette (Ed.), Handbook of industrial and organizational psychology. Chicago: Rand-McNally. Lodahl, T. M. & Kejner, M. (1965). The definition and measurement of job involvement. Journal of Applied Psycholggy, 55, 24—33. Loether, H. J. (1964). The meaning of work and adjustment to retirement. In A. B. Shostak & W. Gomberg (Eds.), Blue collar world. Englewood Cliffs, N.J.: Prentice-Hall. Lofquist, L. H. & Dawis, R. V. (1969). Adjustment to work. New York: Appleton-Century—Crofts. -219- Logan, N., O’Reilly III, C. A., & Roberts, K. H. (1973). Jobs atisfaction among part-time and full-time employees. Journal of Vocational Behavior, 3, 33-41. Lowenthal, M. F. (1971). Intentionality: Toward a framework for the study of adaptation in adulthood. Aging and Human Development, 2, 79— 95. Lowenthal, M. F. & Chiriboga, D. (1973). Social stress and adaptation: Toward a life-course perspective. In C. Eisdorfer & M. P. Lawton (Eds.), The psychology of adult development and aqinq. Washington, D.C.: American Psychological Association. Maslow, A. H. (1970). Motivationand personality. New York: Harper and Row. Mead, G. H. (1930). Mind. self. and society, edited by C. W. Morris. Chicago: University of Chicago Press. Miller, H. E. & Terborg, J. R. (1979). Job attitudes of part-time and full-time employees. ngrnal of Applied Psychology, 55, 380-386. Mitchell, T. R. (1985). An evaluation of the validity of correlational research conducted in organizations. Aeademy of Management Review, 15, 192-205. Mobley, W. H. (1977). Intermediate linkages in the relationship between job satisfaction and employee turnover. Journal of Applied Psychology, 5;, 237-240. -220- Mobley, W. H., Griffeth, R. W., Hand, H. H., & Meglino, B. M. (1979). Review and conceptual analysis of the employee turnover process. Psychological Bulletin, 55, 493-522. Mobley, W. H., Horner, S. 0., & Hollingsworth, A. T. (1978). Intermediate linkages in the relationship between job satisfaction and employee turnover. Journal of Applied Psychology, 55, 408-414. Moore, W. E. (1963). Man. time. and society. New York: John Wiley. Morita, J. G., Lee, T. W., & Mowday, R. T. (1989). Introducing survival analysis to organizational researchers: A selected application to turnover research. Journal of Applied Psychology, 15, 280-292. Morrison, M. H. (1982). Flexibility in retirement: U.S. and international experience. In M. H. Morrison (Ed.), Economics of aging: The future of retirement. New York: Van Nostrand Reinhold. Morrow, P. C. (1983). Concept redundancy in organizational research: The case of work commitment. Academy of Management Review, 5, 486-500. Mowday, R. T., Porter, L. W., & Steers, R. M. (1982). Employee- organization linkages: the psychology of commitment. absenteeism, and turnover. New York: Academic Press. ~221- Mowday, R. T., Steers, R. M., & Porter, L. W. (1979). The measurement of organizational commitment. Journal of Vocational Behavior, 11, 224- 247. Muchinsky, P. M. & Tuttle, M. L. (1979). Employee turnover: An empirical and methodological assessment. Journal of Vocational Behavior, 15, 43—77. Nardone, T. J. (1986,February). Part—time workers: Who are they? Monthly Labor Review, 13-19. Nicholson, N. (1984). A theory of work role transitions. Administrative Science Quarterly, 22, 172-191. Nollen, S. D. & Martin, V. H. (1978). Alternative work schedules. Part 2: Permanent part-time employment. New York: Amacon. Nunnally, J. C. (1978). Psychometric Theory (2nd Ed.). New York: McGraw-Hill. O’Reilly, C. (1983,August). Corporations, cults, and organizational culture: Lessons from Silicon Valley firms. Paper presented at the forty—third meeting of the Academy of Management, Dallas, Texas. O’Reilly, C. & Caldwell, D. (1981). The commitment and job tenure of new employees: Some evidence of postdecisional justification. Administrative Science Quarterly, 25, 597—616. -222- O’Reilly, C. & Chatman, J. (1986). Organizational commitment and psychological attachment: The effects of compliance, identification, and internalization on prosocial behavior. Journal of Applied Psychology, 11, 492-499. Owen, J. D. (1978, June). Why part—time workers tend to be in low-wage jobs. Monthly Labor Review, 11—14. Personick, V. A. (1983,November). The job outlook through 1995: Industry output and employment projections. Monthly Labor Review, 24— 35. Pfeffer, J. & Lawler, J. (1980). Effects of job alternatives, extrinsic rewards, and behavioral commitment on attitude toward the organization: A field test of the insufficient justification paradigm. Administrative Science Quarterly, 25, 38-56. Plewes, T. (1984). Profile of the part—time worker. In D. Rothberg (Ed.), Part-time employment in America. McLean, Va: Association of Part—time Professionals. Pond, S. B. & Geyer, P. D. (1987). Employee age as a moderator of the relation between perceived work alternatives and job satisfaction. Journal of Applied Psychology, 12, 552—557. Porter, L. W. (1963). Job attitudes in management: 11. Perceived importance of needs as a function of job level. Journal of Applied Psychology, 51, 141—148. -223— Porter, L. W. & Steers, R. M. (1973). Organizational, work, and personal factors in employee turnover and absenteeism. Psychological Bulletin, 59, 151-176. Porter, L. W., Steers, R. M., Mowday, R. T., & Boulian, P. V. (1974). Organizational commitment, job satisfaction, and turnover among psychiatric technicians. Journal of Applied P§ychology, 55, 603-609. Quinn, R. P. & Shepard, L. J. (1974). The 1972—73 Quality of Employment Survey. Ann Arbor. The Institute for Social Research, University of Michigan. Quinn, R. P. & Staines, G. L. (1979). The 1977 Quality of Employment Survey. Ann Arbor. The Institute for Social Research, University of Michigan. Reichers, A. E. (1985). A review and reconceptualization of organizational commitment. Academy of Management Review, 15, 465-476. Reichers, A. E. (1986). Conflict and organizational commitments. Journal of Applied Psychology, 11, 508-514. Rhodes, S. R. (1983). Age-related differences in work attitudes and behavior: A review and conceptual analysis. Psychological Bulletin, 55, 328-367. -224- Rotchford, N. L. & Roberts, K. H. (1982). Part-time workers as missing persons in organizational research. Academy of Management Review, 1, 228-234. Rusbult, C. E. & Farrell, D. (1983). A longitudinal test of the investment model: The impact of job satisfaction, job commitment, and turnover of variations in rewards, costs, alternatives, and investments. Journal of Applied Psychology, 55, 429-438. Salancik, G. (1977). Commitment and the control of organizational behavior and belief. In B. Staw & G. Salancik (Eds.), New directiona in organizational behavior. Chicago: St. Clair. Salancik, G. & Pfeffer, J. (1978). A social information processing approach to job attitudes and task design. Administrative Science Quarterly, 25, 224-235. Sanford, N. (1955). The dynamics of identification. Psychological Review, 52, 106-118. Schein, E. H. (1978). 5areer dynamics: Matching individual and organizational needs. Reading, MA: Addison-Wesley. Schein, E. H. (1980). Organizational psychology (3rd Ed.). Englewood Cliffs, N.J.: Prentice-Hall. —225- Seashore, S. E. (1986). Surveys in organizations. In J. W. Lorsch (Ed.), Handbook of organizational behavior. Englewood Cliffs, N.J.: Prentice—Hall. Smith, P. C., Kendall, L. M., & Hulin, C. L. (1969). The measurement of satisfaction in work and retirement. Chicago: Rand—McNally. Smith, P. C., Organ, D., & Near, J. (1983). Organizational citizenship behavior: Its nature and antecedents. Journal of Applied Psycholggy, 55, 653—663. Sofer, C. (1970). Men in mid—career. London: Cambridge University Press. Spector, P. E. (1987). Method variance as an artifact in self- reported affect and perceptions at work: Myth or significant problem? Journal of Applied Psychology, 12, 438—443. Steers, R. M. (1977). Antecedents and outcomes of organizational commitment. Administrative Science Quarterly, 22, 46—56. Steers, R. M. (1984). Introduction to Organizational Behavior. (2nd Ed.). Glenview, Ill.: Scott, Foresman. Steers, R. M. & Mowday, R. T. (1981). Employee turnover and post- decision accomodation processes. Research in Organizational Behavior, 5, 235—281. -226- Steers, R. M. & Rhodes, S. R. (1978). Major influences on employee attendance: A process model. Journal of Applied Psychology, 55, 391- 407. Stevens, J. M., Beyer, J., & Trice, H. M. (1978). Assessing personal, role, and organizational predictors of managerial commitment. Academy of Management Journal, 21, 380—396. Stinson Jr., J. F. (1986,November). Moonlighting by women jumped to record highs. Monthly Labor Review, 22-25. Super, D. E. (1980). A life—span, life—space approach to career development. Journal of Vocational Behavior, 15, 282—298. Super, D. E. & Bohn, M. J. (1970). Occupational psychology. Belmont, CA: Wadsworth Publishing. Van Maanen, J. & Schein, E. H. (1977). Improving the quality of work life: Career development. In J. R. Hackman & J. L. Suttle (Eds.), Improving life at work. Santa Monica, Ca: Goodyear. Vroom, V. H. (1964). Work and motivation. New York: Wiley. Wakefield, D. S., Curry, J. P., Mueller, C. W., & Price, J. L. (1987). Differences in the importance of work outcomes between full-time and part-time hospital employees. JoUrnal of Occupational Behavior, 5, 25— 35. -227- Wanous, J. P. (1974). Individual differences and reactions to job characteristics. Journal of Applied Psychology, 55, 616-622. Wanous, J. P. (1977). Organizational entry: Newcomers moving from outside to inside. E§ychological Bulletin, 55, 601-618. Wanous, J. P. (1978). Realistic job previews: Can a procedure to reduce turnover also influence the relationship between abilities and performance. Personnel Psychology, 51, 249-258. Werther Jr., W. B. (1975). Part-timers: Overlooked and undervalued. 5gsiness Horizons, February, 13-20. Williams, L. J. & Hazer, J. T. (1986). Antecedents and consequences of satisfaction and commitment in turnover models: A reanalysis using latent variable structural equation methods. Journal of Applied Psychology, 11, 219-231. Winer, B. J. (1971). 5Latistical principles in experimental design. (2nd Ed.). New York: McGraw-Hill. Winkler, J. D., Kanouse, D. E., & Ware, J. E. (1982). Controlling for acquiescence response set in scale development. Journal of Applied Psychology, 51, 555-561. -228- Wright, J. D. & Hamilton, R. F. (1978). Work satisfaction and age: Some evidence for the ’job change’ hypothesis. Social Forcea, 55, 1140- 1158. Youngblood, S. A. (1984). Work, nonwork, and withdrawal. Journal of Applied Psychology, 55, 106-117. Zedeck, S. (1971). Problems with the use of "moderator" variables. E§ychologigal Bulletin, 15, 295-310. QRIES "lilllllllllllllll 2