ORGANEZATIONAL S’TEREGTYPES OF EMPLOYMENT ENTERWEWERS REVEALED BY B'IGQATA PAWERNS QF ACCEPTED“ VS. [EEjECTED APPLECANFS Thesis for the Dogma as? M. A. MICHIGAN STATE UNIVERSITY Virginia. Cope Green 1967 "‘- ABSTRACT ORGANIZATIONAL STEREOTYPES OF EMPLOYMENT INTERVIEWERS REVEALED BY BIODATA PATTERNS OF ACCEPTED VS. REJECTED APPLICANTS by Virginia Cope Green This study attempted to identify employment interviewer stereotypes through the analysis of biographical information collected from bona fide job applicants during 1964 and 1965 by the Apollo Support Department of the General Electric Company. The criterion was the interviewers' decision to offer or not to offer employment to an applicant. The inter- viewers were not aware of the subjects' (job applicants') responses made on an "interview supplement form" (actually a biographical information form) administered during selection processing but nonetheless could, and undoubtedly did, pick up fairly parallel information in their employment interviewing. Two separate analyses were carried out, one for the pro- fessional and technical applicants and the other for the hour— ly applicants. These two separate analyses served as a check on the extent to which the interviewers who might show stereo— types in selecting for one common job in a company would also show stereotypes in selecting for another, quite different common job in the same company. In other words, some degree Virginia Cope Green of generality of employment interviewer stereotyping over two jobs could be ascertained. Design of the Study Sample. Operation P & T investigated the differences between Group I, the professional and technical applicants who were interviewed and offered employment by General Electric (N = 56), and Group II, the professional and techni- cal applicants who were interviewed and rejected for employ- ment by General Electric (N = 55). Operation H investigated the differences between Group III, the hourly applicants who were interviewed and offered employment by General Electric (N = 49), and Group IV, the hourly applicants who were interviewed and rejected for employment by General Electric (N = 36). Method. An item analysis was performed on the biographi— cal data. Specifically, chi-square was used to measure the discrepancies between an expected distribution and the ob- served distribution of responses for each item. Results and Discussion Only one of the five a priori hypotheses was upheld by the analysis, while four were rejected. The number of sig- nificant item differences between the criterion groups was slightly above the chance level in the case of the professional Virginia Cope Green and technical applicants, but it was below the chance level in the case of the hourly applicants. Because the item dif- ferences between criterion groups were not clearly defined with either of the two samples, only Hypothesis V, which predicted that the patterns of significant items for the pro- fessional and technical applicants would be different from the patterns of significant items for the hourly applicants, was upheld. The rejected hypotheses were: I. For the professional and technical applicants, a larger proportion of the sub- jective items than of the objective biographical items would discriminate. II. More items would discriminate for the professional and technical applicants than for the hourly applicants. III. For'the hourly applicants, items concerning the applicant's attitude toward the job situation would dis- criminate better than other types of items. IV. For the professional and technical applicants, items concerning the applicant's attitude toward the job situation, his "academic" life, and his social and leadership qualities would discrimi- nate better than other types of items. Among the few tendencies which did appear were the following: (1) the professional-technical applicant who was offered employment by General Electric tended to be more out- going socially and more self-confident than the professional- technical applicant who was rejected for employment, and Virginia Cope Green (2) the hourly applicant who was offered emplOyment tended to claim to enjoy routine or detailed work more than the hourly applicant who was rejected. It was concluded that the department's selection process successfully randomized any individual interviewer stereo- types to the extent that no strong organizational stereotypes were evident. Approved: 2%J4&L-w/ if, ggibtégvfir, i'\‘ ; ' Date: ‘y,,1g,g' . 19.4- '7 I LN Thesis Committee: Frederic R. Wickert, Chairman Eugene Jacobson James L. Phillips ORGANIZATIONAL STEREOTYPES OF EMPLOYMENT INTERVIEWERS REVEALED BY BIODATA PATTERNS OF ACCEPTED VS. REJECTED APPLICANTS BY Virginia COpe Green A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1967 C? 45 9L0 (7/95/“77 L) ACKNOWLEDGEMENTS I wish to express my thanks to Dr. Frederic R. Wickert, my major professor and chairman of my thesis committee, for the many hours he devoted to over-seeing the planning and writing of this thesis. I also thank Dr. Eugene Jacobson and Dr. James Phillips for their thoughtful guidance and criticism, especially in formulating the design of the study. I am indebted to Dr. Herbert H. Meyer of the General Electric Company's Behavioral Research Service and to members of the Richardson Foundation for arranging to have data sent to me for analysis, and I am grateful to Mr. Glenn A. Bassett of the Apollo Support Department of General Electric for providing the data and the answers to many questions concern- ing it. Finally, I wish to thank my husband, Gary, for his help and encouragement throughout the writing of this thesis. V.C.G. ii TABLE OF CONTENTS Page ACKNOWLEDGMENTS. . . . . . . . . . . . . . . . . . ii LIST OF TABLES . . . . . . . . . . . . . . . . . . v LIST OF APPENDICES . . . . . . . . . . . . . . . . Vi I. INTRODUCTION . . . . . . . . . . . . . . . . . . . 1 Brief Description of the Study . . . . . . . . 1 General Background and Usefulness of the BIB 2 Background of the Present Study. . . . . . . . 5 Purpose of the Study . . . . . . . . . . 7 Need for the Study in the Light of General Employment Practices . . . . . . . . . . . . 8 Confusion Concerning What Takes Place in the Interview and What the Interview Contributes. . . . . . . . . . . . . . 8 Discrepancies between the Characteristics Employers Say They Look for in Employees and the Characteristics They Actually Use. . . . . . . . . . . . . . . . . . . 16 A Study of the Relationship between Psycho- logical Test Results and the Decision To Offer or Not To Offer Employment . . . . . . 18 Review of Other Related Studies: A Background from Which to Draw Hypotheses for the Present Analysis . . . . . . . . . . . . . . 20 Biographical Data as a Discriminator of Pro- fessional Interest Groups. . . . . . . . . . 22 Categorization of Items in the Present Study . 25 Development of Hypotheses. . . . . . . . . . . 26 Hypothesis I . . . . . . . . . . . . . . . . 26 Hypothesis II. . . . . . . . . . . . . . . . 27 Hypothesis III . . . . . . . . . . . . . . . 28 Hypothesis IV. . . . . . . . . . . . . . . . 29 Hypothesis V . . . . . . . . . . . . . . . . 50 II. DESIGN OF THE STUDY. . . . . . . . . . . . . . . . 52 Sample . . . . . . . . . . . . . . . . . . . . 52 Method . . . . . . . . . . . . . . . . . . . . 55 iii TABLE OF CONTENTS - continued Page III. RESULTS AND DISCUSSION. . . . . . . . . . . . . 41 Introduction. . . . . . . . . 41 Support or Rejection of the Hypotheses by the Analysis: Interpretation of Patterns of Responses by Categories of Items . . . 41 .Summary ovaindings Concerning Cate- gories of Items . . . . . . . . . . . . 48 Interpretation of Patterns of Responses by Individual Items. . . . . . . . . . . . 49 Professional and Technical Applicants . 49 Hourly Applicants . . . . . . . . . . . 55 IV. SUMMARY AND CONCLUSIONS . . . . . . . . . . . . 56 Design of the Study . . . . . . . . . . . . 56 Sample. . . . . . . . . . . . . . . . . . 56 Instrument. . . . . . . . . . . . . . . . 57 Method. . . . . . . . . . . . . . . . . . 57 Results and Discussion. . . . . . . . . . . 57 Conclusions . . . . . . . . . . . . . . . . 59 Suggestions for Further Research. . . . . . 61 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . 65 APPENDICES . . . . . . . . . . . . . . . . . . . . . 66 iv TABLE 5.1 5.2 3.5 LIST OF TABLES Comparison of objective vs. subjective items with reapect to ability to discriminate between professional-technical criterion groups . . . . Comparison of items belonging to categories 2, 5, and 7 vs. items belonging to all other cate- gories in ability to discriminate between pro- fessional-technical criterion groups. . . . . . Ability of all items to discriminate between professional-technical criterion groups vs. ability of all items to discriminate between hourly criterion groups . . . . . . . . . . . . Comparison of items belonging to category 2 vs. items belonging to all other categories in ability to discriminate between hourly cri- terion groups . . . . . . . . . . . . . . . . . Page 45 44 46 47 APPENDIX I. II. III. IV. V. VI. VII. VIII. LIST OF APPENDICES Page The Questionnaire. . . . . . . . . . . . . 67 The Interview Schedule . . . . . . . . . . 81 Categorization of Objective vs. Subjective Items. . . . . . . . . . . . . . . . . . . 85 Detailed Categorization of Items . . . . . 85 Analysis of Multiple-Answer Questions. . . 88 Tabulation of Responses. . . . . . . . . . 95 Contingency Tables, x2 Values, and Levels of Significance for Operation P & T (Groups I & II). . . . . . . . . . . . . . 105 Contingency Tables, x2 Values, and Levels of Significance for Operation H (Groups III & Iv). . . . . . . . . . . . . . . . . 113 vi I . INTRODUCTION Brief Description of the Study This study will attempt to identify collective inter- viewer stereotypes through the analysis of biographical information, using as the criterion the company's decision to offer or not to offer employment to an applicant. The methodology is much the same as that which has been used in several earlier analyses of biographical data, using a test of statistical significance to determine which of the bio- graphical items discriminate significantly between the criterion groups. The choice of criterion is the feature which sets this study apart, since it changes the focus of interest from the applicants' post-employment, on-the—job behavior to the interviewers' decision concerning whether or not to offer employment. Two separate analyses will be done, the first for a group of professional-and-technica1 applicants and the second for a group of hourly applicants, both in the same company (an aerospace component of a large American electrical manu— facturer). In each analysis, a search will be made for the items or groups of items which discriminate between the appli- cants who were offered employment and those who were not. Finally, the items that discriminate for the professional and technical applicants as opposed to the hourly applicants will be contrasted. General Background and Usefulness of the BIB The history of attempts to use biographical information purely as a selection device, rather than a way to get at employment interviewer stereotypes, dates back to preAWorld War I days, when application blanks were develOped to aid in the selection of insurance salesmen (Ferguson, 1961). The use of information concerning personal history actually came into its own during World War II, with the military's using it for a variety of selection and classification purposes. Near the conclusion of World War II, the Personnel Research Section of the U. S. Adjutant General's office claimed the Biographical Information Blank (BIB) to be one of the most promising tools for identifying officers who had been most valuable to the service (Owens, Glennon, and Albright, 1962). Mandell (1964) believes that the biographical informa- tion blank or weighted application form is one of the most useful American contributions to selection procedure. (The terms "biographical information blank," "personal history," "weighted application," "scored life history," and some others have not always been used in a rigidly defined manner during the past twenty years. General usage in the most recent literature, however, seems to imply that "weighted application blank" or "scored life history" refer rather strictly to a questionnaire concerning only factual information, while a "biographical information blank" or "personal history questionnaire" can include items with as wide a sc0pe as might be covered in an interview, such as questions concern- ing attitudes and preferences. Since the questionnaire in— volved in this study does include items of a subjective nature, it will be referred to as a BIB and the information which it conveys as biographical data or personal history data.) The use of a BIB, according to Mandell, is best described as a method for scientifically evaluating informa- tion that could be gained in an interview. It can be a valuable source of information either by itself or as a supplement to an interview. Bassett (1963) points out that biographical data such as age, education, and employment history should be just as useful for selection purposes as any published test, since each of them meets the basic definition of a "test" by pro- viding a sample of the individual's behavior. In addition to this factual information, the BIB can also provide other clues to the individual's background. Guion (1965) believes that the subjective items which make the BIB "somewhat more ingenious in tapping psychologically significant character- "1 may also make it more valu- istics of background behavior able as a predictive device than an ordinary application blank. lRobert M. Guion, Personnel_Testing (New_York: McGraw- Hill Book Company, 1965), p. 420. ’ Biesheuvel (1965) agrees with Guion concerning the im- portance of attending to subjective information. He empha- sizes that jobs are typically described in terms of attributes which are at least partly non-cognitive, such as leadership, sociability, and supervisory ability, among others. As the job level becomes higher and higher, these noncognitive traits become increasingly important, since requirements concerning education and training have already eliminated all the candi— dates who fall below certain standards. That is, the candi- dates for a high-level job may all be very much alike accord- ing to the available objective information; the decision must be made on the basis of the subjective. Guion (1965) and Mandell (1964) both regard as a strong point of the BIB the fact that it apparently cannot be faked as readily as interest or personality inventories. Perhaps this is the result of putting together in one questionnaire some items which could be verified quite easily by the pro- spective employer, some which could be verified by going to a little more trouble, and so on, until some of the items are virtually impossible to verify. If the individual feels compelled to answer the immediately verifiable items truth- fully, perhaps it is easier for him to answer all of the items as truthfully as he can. Whatever the reason for the truthfulness, Guion and Mandell are agreed that the available evidence shows surprisingly little distortion for the BIB as compared to other types of tests. ,Background of the Present Study Dr. Herbert H. Meyer of the General Electric Company's Behavioral Research Service arranged for biographical data collected during 1964 and 1965 by the Apollo Support Depart- ment of General Electric to be sent to Michigan State Uni- versity to serve as the subject of a graduate thesis problem. The data are interview supplement forms completed by 201 applicants for employment with the Apollo Support Department (see Appendix I). The questionnaire (i.e., the interview supplement form) was developed by Dr. Meyer, using some items that are prob- ably as old as the use of biographical data itself and some new items which have resulted from research conducted by the Behavioral Research Service. The questionnaire was not used in selection: the data were simply collected for analysis at a later date. The application blank and the interview, which were important to the selection procedure, gave extensive information concerning personal history, with the interview carrying the most weight in the final selection decision. The items on the questionnaire cover much the same in- formation that an interviewer might expect to obtain by "Foundation Questions" I through IV of the interview schedule (see Appendix II). It is impossible to determine to what depth these foundation questions were explored for each appli- cant since the interviews were not systematically recorded during the time that the interview supplement forms were being filled out, but it seems safe to assume that the interviewers covered most of the areas included in the scope of the questionnaire. All the applicants had to pass multiple interview hurdles involving several different personnel interviewers and one or more of the many line managers responsible for the specific Operations to which the applicants were to be assigned if hired. Ten to twelve interviewers worked in the Employment and Recruiting Office alone, and each candidate who went through that office was interviewed by one or more of these interviewers and again by one or more managers. It is certain that each applicant was interviewed at least twice. The im- pression gained from communications with the Employment Office was that any interviewer bias was rather randomly distributed among the interviewers. The patterns of biodata items for accepted as compared with rejected candidates reflect, then, collective stereotypes typical of the organization rather than stereotypes of individual interviewers. The personnel interviewers were trained by General Electric and were provided with the Foundation Questions to guide their interviews. The managers had had little more training in selection interviewing than is customary in in- dustry. They had been provided with a handbook, "A Guide to Effective Interviewing and Selection," published by the Apollo Support Department in May, 1964. However, there is no infor- mation with respect to how closely the managers followed the outlines provided for them in the handbook (Bassett, 1966). Purpose of the Study The purpose of this study is to investigate to what ex- tent both individual items and clusters of similar items were answered significantly differently by applicants who were offered employment and those who were rejected by the company. Finding items which do discriminate between the two groups would serve a two-fold purpose: 1. It would make it possible for the employment manager to determine to what extent the interviewers and managers are making their selection decisions on the basis of characteristics important to the company. There is evidence (which will be cited in the next section) that interviewers' decisions are sometimes made on the basis of information which even the interviewers themselves are reported to regard as irrelevant when later given feedback on the elements that apparently entered into the decisions they made. 2. If the selection decisions are apparently being made on the basis of characteristics which the company hopes to emphasize, then a clear-cut statement of exact— ly what factors are important can be formulated and perhaps a biodata form could be used at the beginning of the selection procedure to serve as a screening device to predict eventual selection or rejection. If, however, characteristics which on §_priori grounds should not be the deciding factors in the selection decision appear to be getting consideration or are being stressed too heavily, the implication may be for the interviewers to re-evaluate the information itself and/or the weight they are attaching to each piece of information. Thus, the study is essentially a search for interviewer stereotypes as those stereotypes are reflected in applicant responses to the interview supplement form. It is an attempt to identify differences between those candidates selected and those rejected. The selection-rejection criterion is the characteristic which sets this study apart from other studies of biographical data. Needyfgr the Study in the Light of General Employment Practices Two major problems in the area of employee selection apart from validity itself indicate that a study of this type might prove useful. The first is the general confusion con- cerning (a) what actually takes place during the interview and (b) what information the interview can or should contribute to the selection decision. The second problem concerns some evidence of discrepancies between the factors companies and their interviewers say they are looking for and the factors which they actually use. Confusion Concerning What Takes Place in the Interview gnd‘What theygnterview Contributes. An extensive survey of the literature concerning the employment interview was under- taken by Ralph Wagner in 1949. Wagner proposed that the interview is very much like a multiple regression technique, since both procedures take different aspects of an individual's personality into consideration, attach some weight to each bit of information according to its relevance, and make a decision on the combination of the factors. He emphasized that the interviewer may not even realize he is trying to combine all factors in this way. Wagner's principal conclusions at the end of his survey of the literature were simple but sweeping: He said that a great deal of confusion existed as to what could or could not be accomplished by the selection interview and that research on the interview was much needed. The next comprehensive reviews of the literature concern- ing interviews were not published until 1964 (Mayfield) and 1965 (Ulrich and Trumbo). Both reviews expressed a continuing concern that while the interview continues to be the most widely used method for selecting employees, there is still no extensive, systematic research on how it can best be uti- lized. Both said that the situation had not changed appre- ciably since Wagner concluded in 1949 that there was much confusion and that research was urgently needed. Mayfield gave the following example of extremely elementary research that has not been done: while it is generally believed that the more structured an interview is the more reliable and valid it is, he could find no study in which the amount of 10 structure had been varied systematically to see what effect it would have on the results. Mayfield sees as the most promising development in recent research the increasing attention being paid to decision-making in the selection interview. By far the most detailed study of this type took place from 1954-1964 at McGill University under the direction of E. C. Webster. Many of the findings of the McGill studies imply that it is extremely important that companies and their interviewers establish exactly what it is that they are looking for in their employees and that they then try to be very systematic in finding these qualities. The following are some of the general findings (as summarized by Webster, 1964) together with related findings and more detailed comments where appropriate: V 1. Interviewers tend to develop a stereotype of a good candidate and then to try to match men to the stereotype. The stereotype is often not a carefully conceptualized description of what the interviewer is seeking. 2. A bias is established very early in the inter- view and will play a large role in determining whether the final decision will be favorable or unfavorable. Webster regards this as the single most important find— ing of the McGill studies, and it was verified by at least four of the researchers working under him. 11 3. Interviewers are more influenced by unfavorable than by favorable information. Springbett (1958) a130 found that, because of the unequal weighting of infor- mation, the interviewer's impression was more likely to change from favorable to unfavorable than vice versa. He concluded that the interview is primarily a search for negative information. Additional evidence of the emphasis which has long been placed on negative information is Uhrbrock's (1948) "flow chart of selection procedure," which he based on observations of actual practice and not on some ideal of how selection interviewing should be done. The flow chart is an outline of ten steps involved in selecting a new employee. Each of the steps ends either in re- jection of the applicant or in his advancement to the next step in the procedure. Further, Uhrbrock stated that each of the interviews indicated in the flow chart should have a sub-series of terminal points where it is decided whether to reject the candidate or continue to screen him. Uhrbrock estimated that usually 5% to 10% of the applicants should survive the initial stages of screening and be referred to the department head or foreman who had requisitioned personnel. Bassett (1962) is also concerned that the screening , process too often concentrates more on rejection than selection. He believes that selection procedures are 12 often just a means of eliminating a certain proportion of the applicants, and that the ones who remain may not be the best ones at all. In fact, the company may auto- matically reject a certain number of applicants at each level until the executive who supposedly makes the final decision has only one person to "choose." Bassett thinks that such procedures encourage the applicant to try to hide his weaknesses rather than to try to sell his strong points, and that selection procedures cannot be very effective until there is a balance between the negative and the positive aspects of choosing employees. It appears that much more work is needed to determine the relative worth of negative and positive information and to establish the proper balance between "selection" and "rejection" techniques in the screening process. 4. Interviewers search for information to support or reject their own hypotheses about an applicant, and, as soon as they find anything to support their ideas, they direct their attention to a new area. Crowell (1961) found that interviewers generally try to confirm early impressions by changing the emphasis they place on dif- ferent aspects of the information received. She further found that even the choice of areas to be explored in the interview reduces the possibility of the inter- viewer's discovering information which is not compatible with his first impressions of the applicant ("selective perception"). 15 5. The relationship between the degree of empathy felt by the interviewer and the decision to accept or reject the applicant is specific to the individual inter- viewer. Smith (1966) defines empathy as "the tendency of a perceiver to assume that another person's feelings, "2 Sydiaha thoughts, and behavior are similar to his own. (1962) found that empathy can be a major source of error in interview decisions, since interviewers tend to attribute characteristics to the applicants which the applicants do not attribute to themselves. 6. Experienced interviewers tend to rank applicants in the same order, but they differ in the percentage of applicants they will accept. Any one interviewer is generally highly consistent in the proportion he will accept. These are the major findings of the McGill studies. It is important to note that the sole concern of the studies was to learn about decision-making processes as they affect the interview. It was irrelevant to the investigation whether or not the decisions were valid as to their accuracy of pre- diction. Mayfield regards Webster's method of paying attention only to how the interviewers make their decisions as a promis- ing approach to the study of the selection interview. He agrees that if it can be shown that interviewers make different 2Henry Clay Smith, Sensitivity to People (New York: McGraw-Hill, 1966), p. 19. 14 decisions after listening to the same interview because they pay attention to different information, then steps can be taken to lead interviewers to attend to the same information. But if the interviewers make different decisions because they weight the same information differently, then an attempt can be made to get interviewers to examine their own weight— ing behavior. Mayfield believes that the structure of the interview can help to eliminate both of these problems. Ulrich and Trumbo (1965), as well as Mayfield, are con- cerned with the inadequacy of evidence concerning the useful- ness of the interview as a selection device. They, too, see some hope in the research now developing and are particularly interested in studies concerning the accuracy of information obtained by interview. They cite the following two studies as examples of research that has been needed for some time: 1. Metzner and Mann (1952) interviewed and admin— istered questionnaires to employees of an electric utility in an attempt to find out how much agreement there was between the employees' replies to an open— ended interview and a fixed-alternative questionnaire. An employee's response to a particular question in the interview was rated 1 through 5 by judges so that it could be checked for agreement with responses 1 through 5 to the same question on the questionnaire. Metzner and Mann found some discrepancies between infor- mation gathered by the two separate methods, Specifically 15 that (a) agreement between interview and questionnaire responses depended on the similarity of the wording of the questions, (b) agreement depended even more on the context in which the question was asked than on the exact wording, and (c) degree of agreement was also related to the employee's level of work, with white collar employees showing more agreement between interview and questionnaire responses than blue collar employees. 2. Bennett, Alpert, and Goldstein (1954) also did a study to determine how much agreement in information there was between answers to a questionnaire and responses to a taped personal interview. Their technique was to interview each subject extensively, and they covered the same general areas that were covered by the questionnaire. The interviewer and another judge, using the tape- recorded interview, then tried to predict how the sub- ject would respond to the questionnaire. The investi- gators found discrepancies between the predictions (made by the interviewer and the judge) and actual responses to the questionnaire, but they believe that their findings do not warrant any extensive conclusions as to the nature of or the reasons for the discrepancies. Ulrich and Trumbo regard these two studies as being really concerned with the accuracy of interpersonal perceptions. They believe that further research in this direction will have important implications for the interview. They view Sydiaha's 16 empathy studies as another promising beginning to the in- vestigation of accuracy of perceptions in the interview. Discrepancies between the Characteristics Employers Say TheyLook for in;§mployees and the Characteristics They Actually Use. Several studies have been done in an attempt to identify the characteristics of college graduates that are stressed by campus recruiters and companies in selecting graduates for jobs. Most of these investigations were mail surveys which focused on the company. Dickinson (1955) obtained employers' ratings of seven characteristics of college graduates as they relate to a successful career in Administrative, Technical, Accounting, Teaching, and Sales jobs. The factors to be rated 1 to 7 were sociality, intel- ligence, conscientiousness, drive, training, judgment, and physical traits. Dickinson, with a usable return of 66% on 659 questionnaires, found that the factor rated by employers as being most important was not the same for all five occu- pational fields. However, intelligence was consistently rated first or second and was the most important trait in the composite ratings for all fields. The physical traits factor (looks, bearing, manner, speech) was consistently rated sixth or seventh and was the least important factor on the composite ratings. Sullivan (1961), in a similar study, found that drive and initiative were listed as the personal qualities most im- portant to the companies, and that academic grades and evidences 17 of leadership on the campus were the most important infor- mational items. Johnson (1956) found that 92% of the companies who responded to his questionnaire listed high grades and extra- curricular activities as most important. Carroll (1966) conducted a study at the University of Minnesota in 1961 which was directed at the student seeking a job rather than at the campus interviewers or the companies. Nineteen personal and biographical characteristics of the students were correlated with five criteria representing their success in obtaining visits and job offers from the companies. The five criteria were the number of job offers received, the number of visit offers received, the visit- interview ratio, the job offer-visit ratio, and a composite measure of the first four criteria. Since these correlations reflect actual student success at getting employer consider- ation regarding a job, Carroll believes that they indicate more accurately the importance attached to student traits than the surveys asking the companies which traits they con- sider to be important. While the earlier surveys of companies and campus inter- viewers indicated that intelligence, grades, leadership positions, and extracurricular activities were all important to them, Carroll found that none of these factors was related to the students' actual job-seeking success. And while all of the previous studies said that appearance was not important, 18 Carroll found that appearance was the most important single factor in terms of the students' success. Marital status and office experience were the only other characteristics which were significantly related (x2, p.g .05) to any of the five criteria. These findings, which represent actual student job-seeking experience, indicate that companies en- gaged in college recruiting are apparently not emphasizing the characteristics which they say are important to them in selecting new employees. A Study of the Relationship between Psycho- lggical Test Results and the gecision To ijer_9£yNot¥To Offerygmployment The next study to be reviewed does not deal specifically with biographical information, but it does involve the vali- dation of various tests against a criterion of whether or not an individual would be offered employment by a company. Hodgson (1964) described an extensive study in which over 500 applicants for technical and professional positions were being considered by the Caterpillar Tractor Company. Each of the applicants was given a battery of tests for mental alertness, mechanical comprehension, and inventories of various person- ality factors. Communication between the test interpreter and the interviewing department was temporarily cut off in order to determine how much the psychological tests, especial- ly the personality inventories, were related to other factors considered important to employment purposes. Without any communication with the interviewing department, the test 19 interpreter evaluated the results of the psychological tests. And, without any knowledge of the results of the test battery, the interviewing department made a tentative decision to hire or not to hire. This tentative decision was based on any information thought useful by the interviewing department other than the results of the psychological tests. Hence, there was no way the test results could have influenced the initial decision whether or not to offer employment. The test results were then studied in terms of the appli— cants who were offered employment versus those who were re- jected by the company. For the Research, Service, and Engineering departments, the tests of ability discriminated sharply between those who were offered employment and those who were rejected, but the personality factors were not sig- nificant. Only a few slight tendencies concerning the per- sonality factors were indicated. In the Research Department, those offered employment tended to be more introspective and reflective in thinking than those who were not offered employment. In the Service Department, those offered employ- ment tended to be less nervous, more outgoing, more aggres- sive, and more stable than those who were rejected. In the Engineering Department, there were no differentiating person- ality tendencies at all. For the Sales Department, nearly the opposite was true. The tests of ability did not discriminate at all between the offered and not-offered groups, but several personality 20 factors discriminated sharply. Those offered employment were significantly more aggressive, more self-confident, less nervous, more socially extroverted, and less subject to depression or cycling moods than those who were rejected by the interviewers. Review of Other Related Studies: A Background from‘Which to Draw Hypotheses for the Present Analysis As was mentioned earlier, most studies concerning bio- graphical data investigate whether the items will differen- tiate between groups on the basis of a criterion of job tenure or job performance. Although the criterion groups in the present study are determined by whether or not an individual was offered a job by the company, it will be of interest to see what success previous investigators have had in finding items which discriminate between their criterion groups and what types of items the discriminating ones tend to be. Hopefully, this study will reveal whether the types of items which predict an employment offer or rejection are the same types of items which have previously been found to predict on-the-job success or failure, long or short job tenure, or some other characteristic. That is, the study should give some indication as to whether interviewers tend to hire applicants who have the characteristics of success- ful, long-tenure employees according to earlier studies. Shott, Albright, and Glennon (1965) attempted to identify factors associated with long or short tenure in an automated 21 office situation. Thirty-four items were available for analysis, of which nine were found to be significant for women and seven for men. The investigators found that both men and women of long tenure had prior work experience and an average of at least ten months with previous employers, left their previous jobs to seek advancement, and had no un- favorable references. Kirchner and Dunnette (1957), in a study for the Minne- sota Mining and Manufacturing Company, also tested biographi- cal items to find those which discriminated between short- tenure and long-tenure groups. Of the 40 variables tested, 15 differentiated between the two groups. From the 15 sig- nificant items, Kirchner and Dunnette constructed a composite picture of the long-term female employee of the company which included, among others, the following characteristics: She was young, single, from Minnesota, a high school or college graduate with no college years left to complete, had no car or furniture, and had left her last job for advancement reasons. In a study of biographical information as judged against a criterion of job performance, Mandell (1950) administered an information blank consisting of 57 items to about 150 chemists working for the United States Government. The items, concerning such information as amount of graduate work, number of published articles, and standing in college graduat- ing class, did not discriminate between those rated high and 22 those rated low in job performance by their superiors and colleagues. Smith, Albright, Glennon, and Owena (1961) tried to validate a 484-item personal history form given to petroleum research scientists against three different criteria of suc- cess in research work-~ratings of over-all performance, rating of creativity (both ratings made by supervisors), and the number of patent disclosures in a five-year period. The discriminating items suggested to the investigators two broad factors responsible for the success of some scientists as compared with others. The first factor they labeled "academic orientation," since the high groups generally had more education than their colleagues, had obtained it on scholarships or fellowships, had published at least one technical paper, and devoted much time to reading. The second factor they called "self-confidence," because the successful scientists said that they had taken better advantage of their opportunities than had the less successful, considered their achievements superior to those of others with the same education, worked faster than others, and liked to have several projects under way at the same time. Biog;aphical_§ata as a Discriminator of gggfessionalylnterest Groups Albright and Glennon (1961) used the same data as in the Smith, Albright, Glennon, and Owens study, but this time tried to see whether the extensive questionnaire would 23 discriminate between administratively-oriented and research- oriented scientists. The subjects were grouped according to whether they said they preferred research or supervisory work. The investigation yielded 43 items which discriminated significantly between those who were more interested in supervisory work and those more interested in pure research. Albright and Glennon then did a cross-validation study of 65 individuals who did not take part in the original study. Using the 43 previously identified discriminating items, each -person in the cross-validation group was arbitrarily assigned a score of +1 for every response characteristic of the research scientists and a -1 for every response typical of those more interested in administration. Thus, each person had a total score which could be represented on a table and used to make generalizations about the whole group. The mean scores of +6.42 for the research scientists and +1.72 for the administratively-inclinced scientists were in the expected directions and were significantly different (t-test, p < .01). In summary, Albright and Glennon found 43 items which did discriminate between research- and supervisory-oriented scientists, and their findings were confirmed by the cross— validation study. Categorization of Items in the Present Study With a survey of the relevant literature in hand, the next step is to turn attention back to the present study. First, how to work with the questionnaire the company used 24 needs to be examined in some detail. The questionnaire has 64 items. For this reason, it would be difficult to compre- hend and keep track of as many as 64 or more separate statis— tical tests (some items may be tested more than one way) without some guidelines as to the relationship of the items to one another. Peak's (1953) advice was that in any case in which there are many separate events (in this instance, items) which are to be reduCed to simpler form, the first step must be to categorize. She stated that the necessary categorization may involve a simple §_priori decision concern- ing the items. Such an a priori categorization may serve a double func- tion later in the analysis. It will, of course, help to define clearly the clusters of items which are significant in discriminating between the criterion groups. Beyond that, however, in the event that the number of significant items is very near the chance level, it would be much more justi- fiable to say that a group of items belonging to the same category indicates some tendency or tendencies than to make the same assertion about separate items which were not linked to one another on any §_priori basis. Thus, the existence of categories will make it easier to deal with the fact that, of the 64 items, about three will be significant by chance at the 5% level of confidence, about six at the 10% level of confidence, and so on. Since the questionnaire includes both subjective and ob- jective items concerning many different specific aspects of 25 personal history and also, cutting across the subjective- objective dimension, several content dimensions, two separate classes of categorizations, (1) subjective versus objective and (2) content, will be used. Anything of a factual nature which the interviewer could probably verify, if he wanted to devote enough time and effort to doing so, will be classi- fied as objective. The subjective items will be the ones which remain and will call for an opinion, attitude, or pref- erence on the part of the applicant (see Appendix III). The second class of categorization will be much more detailed since it is intended to reflect the content di- mensions present in the questionnaire. This class will indicate specifically which aspects of the applicant's personal history are dealt with by the items in each of the content-type categories (see Appendix IV). The following categories will be of special value in the formulation of the hypotheses: (2) Attitudes toward job, supervision, and fellow employees (5) An academic factor--college experiences, attitudes toward schoolwork, reading habits and preferences (7) Professional attainments, social poise, leadership, way of relating to other persons in general Recall that two separate analyses are to be performed. The first operation will identify any significant differ- ences between the professional and technical applicants who 26 were interviewed and offered employment by General Electric and the professional and technical applicants who were interviewed and rejected for employment. Similarly, the second operation will identify any significant differences between the hourly applicants who were interviewed and offered employment by General Electric and the hourly appli- cants who were interviewed and rejected for employment. Qevelopment ofyHypotheses Now, by taking into account the relevant literature, the two classes of item categorization, and the two groups of subjects, it is possible to state the hypotheses. These hypotheses, it will be remembered, are to be concerned with the stereotypes that interviewers are likely to let govern their selection-rejection decision. gypptheers I. For theyprofessionergand techniee; applicants onlyy a lerger proportion of the subjective items than er;the objective items will discriminate between those whg were prrered employment and those who were not offered employment. Mandell (1964) warns that objective facts can mislead employment interviewers, especially in the case of some applicants who lack desirable characteristics. He gives as an example that relevant courses, good grades, and several publications or patents may be good indications to accept an applicant, but their absence may be because the individual had to take a part-time job during college or has been 27 assigned to projects where there was little opportunity for personal achievement. Further, since the professional and technical applicants all meet certain minimum standards of education and training, the selection decision in their case in particular will per- haps depend more on the subjective information, as Biesheuvel (1965) suggests. Hypothesi§_;1. More of all the items will discriminate significantly for the professional and technical applicants than for the hourly applicants. Several of the questions on the interview supplement form concern college experiences and professional activities which are necessarily more relevant to a college-educated person than to one who has not attended college. Since nearly all of the applicants for professional and technical jobs attended college while very few of the applicants for hourly work did, there are actually fewer items which relate to the hourly candidates than to the professional and tech- nical. Metzner and Mann (1952), in their study of agreement between employees' replies to an interview and to a fixed- alternative questionnaire, found that there were more dis- crepancies for the blue collar workers than there were for higher job-level personnel. Lack of agreement between what the interviewer heard during the interview and what the applicant marked on the questionnaire could further reduce 28 the possibility that the questionnaire will reveal any tendencies as to whether the hourly applicants will or will not be offered employment. Finally, 40 of the 64 items (62%%) are subjective in nature, according to the e_priori categorizations. If Hypothesis I is true and the subjective items do discriminate more effectively than the objective for the professional and technical applicants (indicating that Biesheuvel is correct in saying that subjective information becomes more and more important at higher job levels), then the pre- ponderance of subjective items on the questionnaire should lead to a greater number of significant items for the pro- fessional and technical applicants than for the hourly. Hypothesis IrI. gor the hourly applicants, items con— cernipg the applicant's attitude toward his present or previous 19b, sgpervision, and fellow employees (#2 in the Detailed Categorization of Items, Appendix IV) will dis- criminate more successfully between the two criterion groups than items in the other categories. Shott, Glennon, and Albright (1963) found that such indirect indications of applicant attitude toward the previous job as the length of service the hourly employee had with his previous employer, the reasons he gave for leaving his previous job, and his having no unfavorable references (implying that he had a satisfactory relationship with his previous supervisors or employing companies) were 29 all significantly related to job tenure. It seems logical to suppose that interviewers would be particularly concerned with the applicant's attitudes toward the job situation in making a selection decision. Hypothesis IV. For the profeeeronal and technreel applicantsI item§_concerninq the applicantis attitgde toward. his jobI supervision, and fellow empleyees (#2, Appendix ryL; items concerning hie "academic" life j#5); epdpitemeiconeerp- ing his social and leadership gualitiee;(#7) will discriminate mpre sucpessfully between the criterion qroungfthen the items in the other ceteqoriee. Items concerning the job situation will be just as im- portant for the professional and technical applicants as they are for the hourly applicants, and for the same reasons (see Hypothesis III and justification). No matter how well- educated or well-trained the applicant may be, he still will not be a valuable employee if he cannot function well in the day-to-day job situation. In addition to items about the job situation itself, other items will be important for the professional and technical applicants. Smith, Albright, Glennon, and Owens (1961) identified an "academic orientation" factor as being related to the success of the scientists in their study, and it will be of interest to determine whether or not a similar factor might be related in this case with respect to whether or not the applicant was offered employment. 3O Presumably, the manner in which an applicant financed his college education, his attitudes toward his schoolwork, his reading habits and preferences, and his professional accomplishments will convey to the interviewer the appli— cant's high degree of awareness and his eagerness to learn (or the low degree of these qualities). And, since, the professional or technical employee is more likely than the hourly employee to be called upon to serve in a leadership capacity, it will also be of interest to determine whether the selection decision appears to be related to the applicant's social poise, leadership quali- ties, and way of relating to other persons in general. I Hypothesis V. The patterns of signiricant items rpr the profeselenal and technlcal applicants will be different. in the expected socially stereotyped directione, from the petterns of signirlcant items rpr the hourly applicants. This hypothesis is essentially a generalization of Hypotheses I, II, III, and IV which, if true, necessarily mean that the patterns of significant items will be differ- ent for the two groups. The two groups represent two dif- ferent strata both occupationally and socially, and so it should be expected that there will be differences in the way they respond. Even if Hypothesis IV is not true (i.e., if it has not correctly identified the factors which are particularly relevant to the professional and technical group), it should be noted in what ways the professional 31 and technical applicants do differ from the hourly appli- cants in filling out the same questionnaire. Such findings could be an important consideration in indicating to an interviewer how differently or similarly he should handle applicants from the two groups. II. DESIGN OF THE STUDY Sample The data as received from General Electric were in the following six groups: Group A: N = 46. Professional and technical applicants who were interviewed and offered employment by the Apollo Support Department and who accepted the offer. Group B: N = 11. Professional and technical applicants who were interviewed and offered employment but rejected the company's offer. Group C: N = 34. Professional and technical applicants who were interviewed and rejected by the company. Group D: N = 36. Hourly applicants who were inter- viewed and hired by the company in late 1964 and early 1965 and who were still employed at the beginning of 1966. Group E: N = 25. Hourly applicants who were inter- viewed and hired but had terminated employment within a few months of their hiring. This group includes both voluntary terminations and those who were released by the company for administrative or performance reasons. 32 33 Group F: N = 49. Hourly applicants who were inter- viewed and rejected by the company. The professional and technical applicants had been brought to the Apollo Support Department for their interviews at the expense of the company, and they therefore represented some degree of pre-selection on the basis of their resumes, campus interviews, or other sources of information. The hourly applicants represented no pre-selection of any kind; they had only to present themselves in the employment office to be interviewed and considered for a position. The Depart- ment attempted to get every applicant to fill out the inter- view supplement form. Any oversights which may have occurred are believed to be too few in number to introduce a bias into the sample. The first major decision concerning the sample was to put the professional and technical applicants into one study and the hourly applicants into a quite separate study. The differences between these two groups with respect to bio- graphical data are so great that there is very little logic in treating them in one large group. Twenty-seven of the applicants who filled out the interview supplement form were women, and their question— naires were removed from the sample to avoid complicating the data with biodata responses which might vary largely because of sex differences. Besides the variations which would most surely occur because of sex differences on items 34 such as those concerning height, weight, and amount of time spent on high school athletics, it is entirely plausible that the feminine point of view would be much different from a man's on many of the items which call for an Opinion or attitude statement on the part of the applicant. Since the N for Group B was quite small, Group B was combined with Group A and the whole group treated as those professional and technical candidates who were offered employ- ment. Thus, the first Operation is to investigate the differ- ences between the professional and technical applicants who were offered employment by General Electric and those who were not offered employment. The same was true of Group E, which was combined with Group D and the resulting group treated as hourly applicants who were offered employment. The second operation investigates the differences between this newly formed group and the group of hourly applicants who were rejected for employment by General Electric. In summary form, the two operations and the groups in- volved in each, after the women had been taken out of the sample and the groups combined as just described, are the following: I New Group I. Professional and technical applicants who were interviewed and Offered employment by General Electric. Operation P & T N = 56. New Group II. Professional and technical applicants 35 who were interviewed and rejected for employment by General Electric. N = 33. New Group III. Hourly applicants who were interviewed and Offered employment by General Operation H Electric. N = 49. New Group IV. Hourly applicants who were interviewed and rejected for employment by General Electric. N = 36. Method Thorndike (1949) stated that the biographical data blank is the best example of a test blank in which there is no essential unity in the items themselves, and that it is there- fore most appropriate to inquire into the validity of each single item. Accordingly, an item analysis will be performed on the present data, and the e_priori categorizations will be relied upon to provide a frame Of reference within which to study the results. The chi-square statistic lends itself very well to the type of analysis to be performed, since it is a measure of the discrepancy between an expected distribution and the Ob- served distribution. The chi-square deals with differences between Observed and expected frequencies, rather than values (Albright, Glennon, and Smith, 1963), and it is therefore most valuable when the items are not to be "scored," but 36 instead the differences between groups are to be reflected by the number of persons in each group who responded to an item in a certain way. As the analysis progressed, omissions and errors in answering on the part of applicants were encountered in the data, especially in the questionnaires completed by the hourly applicants. The instructions on the questionnaire said specifically to "mark just one answer for each question unless otherwise indicated," but some Of the applicants had markéd'two or more responses to the questions which were to be answered by a single response. Omissions in the data were simply the result of an applicant's failing to answer one or more items. Since it seemed too hazardous to attempt to interpret what the applicant might have meant by a multi- ple answer to a single-response question, the omissions and errors were left out of the analysis. Every item which an individual answered correctly was included in the analysis, even though he might have answered some other items incorrectly. This arrangement utilized every possible bit of information available without making it necessary to speculate as to what an incorrect response might mean. Because the omissions and errors were omitted, the N for each group varies from item to item and the total observed frequency for each group on any item depends on how many of the group answered that item correctly. The chi- square statistic handles the situation smoothly, since it is 37 the distribution of the responses among the cells of the contingency table for each item which is important, not the total number Of responses. Since the expected cell frequency should be at least 5 in order for the chi-square statistic to be reasonably stable, it was sometimes necessary to combine two or more responses to one question into one cell of the contingency table. Such combinations were made in a way which was psycho- logically as sound as possible. Responses were often com- bined in this manner in other cases, too, not because it was necessary in order to obtain the minimum expected value, but simply because it facilitated the analysis and did not distort the data. In a very few cases, responses which were marked by fewer than five applicants and which could not be logical- ly combined with another response to the same question were left out Of the analysis. England (1961) Offered several suggestions for grouping responses for purposes of analysis. With items in which the responses form a continuum, England suggested setting limits which would provide approximately equal total frequencies in each classification. Or, one could set limits which would put an equal number of the possible responses into each classification. Since generally most items are not continu- ous, England also stated that the researcher could, after becoming well acquainted with his data, try a number of dif- ferent groupings. He believes there is no danger that 38 significance will be artificially created in this way but that there is some danger that significance will be missed, and so combinations were made carefully to avoid masking possible significant differences. Ten Of the items on the interview supplement form could be multiply answered, and these items had to be handled dif- ferently from the other questions. In the case of item 29, for instance, there are 25 possible responses, and each applicant was free to mark as few or as many as applied to him. Since it is not sound to make it possible for the re- sponses of one individual to carry up to 25 times the weight of the responses of another in doing the statistical analy- sis, some other method had to be devised. The solution arrived at was to divide the responses into two to four classes and to credit the applicant with responding pppe to a particular class if he marked any one or more of the responses included in that class. (For instance, if re- sponses a, b, and c to a question were all put in the same class, the applicant was credited with one response to the class if he marked a, b, or c pr any two of them pr all three of them.) This system still makes it possible for the re- sponses of one individual to carry up to four times the weight of the responses Of another, but perhaps that is not necessar- ily a bad point since it provides a rough measure of how many different classes of responses an individual who is repre- sentative of his group might be expected to make. Where it 39 was appropriate, the multiple-response questions were also tested for significance according to the sheer number of different responses the applicants made. In short, all of the multiple-answer questions were tested for significance in at least two different ways. See Appendix V for a detailed description Of how each Of the ten questions was handled. The original design of the investigation included a plan to cross-validate, in much the same way as Albright and Glennon did in their study of physical scientists' career aspirations, by using the items which were identified in the initial chi-square analysis as discriminating between the criterion groups. A random-number table was used to select one-fourth of each group in the sample to be taken out for cross-validation purposes. This left groups which were still barely large enough to analyze via the chi-square. However, the differences between the criterion groups were apparently so slight that, with a fourth of the sample taken out, all tendencies disappeared and the initial test yielded nothing of significance. Since the cross-validation plan depended on finding significant items in the initial analysis, there was nothing to cross-validate in this situation. ‘Moreover, the N's were simply too small to try to devise a system for cross-vali- dation which was any more complicated than the simple scheme used by Albright and Glennon. (After the women were removed from the sample, N = 33 for Group II. Only eight question- naires could be held out for cross-validation purposes and 40 still leave a minimum of 25 on which to perform the chi- square analysis.) It seemed better to analyze the whole sample in search of tendencies which would be worth exploring in other studies with larger sample3~than to insist upon using a cross-validation plan which yielded no information. Hence, the cross-validation groups were returned to the sample and the plan to cross-validate was abandoned. Any tendencies which appear will be the result of a chi-square analysis of the whole sample. Because the groups were small, Levy's (1967) statement concerning the relationship between the size of the N and statistical significance may serve as a reminder not to under- estimate the importance of any significant differences. Levy believes that few investigators try to determine the import- ance of significant differences between groups with respect to the size of the groups. He is concerned about this failure, since "a significant result achieved for a small number of cases has a greater potential value than a result from a large number Of cases which achieves the same level of sig- nificance."3 3Philip Levy, "Substantive significance of significant differences between two groups," Psychol. Bulletin, Vol. 67, NO. 1 (1967). p. 37. III. RESULTS AND DISCUSSION _ntroduction A tabulation of the responses to each question by groups is given in Appendix VI. Detailed results of the chi-square tests appear in Appendices VII and VIII. Interpretations of the patterns of responses both by categories and by indi- vidual items are given in this chapter. §ppport or Rejection of the Hypotheses by the Analysis: Interpretation of Patterns Of Responses by Categories of ltems The specific purpose of the analysis was to test the following five hypotheses: H1. For the professional and technical applicants only, a larger proportion of the subjective items than Of the objective items will discriminate between those who were offered employment and those who were not offered employment. H2. More of all the items will discriminate signifi— cantly for the professional and technical appli— cants than for the hourly applicants. H3. For the hourly applicants, items concerning the applicant's attitude toward his present or previous job, supervision, and fellow employees (#2 in the Detailed Categorization of Items, Appendix IV) 41 42 will discriminate more successfully between the two criterion groups than items in the other categories. H4. For the professional and technical applicants, items concerning the applicant's attitude toward his job, supervision, and fellow employees (#2, Appendix IV); items concerning his "academic“ life (#5): and items concerning his social and leadership qualities (#7) will discriminate more successfully between the criterion groups than the items in the other categories. H5. The patterns of significant items for the profes- sional and technical applicants will be different, in the expected socially stereotyped directions, from the patterns of significant items for the hourly applicants. A total of 77 chi-square tests were performed for Opera- tion P & T (see Appendix VII), the operation to investigate possible differences between the professional and technical applicants who were offered employment by General Electric and the professional and technical applicants who were not Offered employment. Of the 77 tests, 11 revealed differences significant at the 10% level or greater. Since 7.7 tests should have been significant at the 10% level by chance alone, the investigation does yield a discrimination which is approximately 44% better than chance significance between the 43 Offered and the not-offered groups. Since the discrimi— nation in this study is at least somewhat better than chance, it will be of value to examine each item.which was shown to be significant and to indicate the directions in which the groups differ. These differences will be explored in detail and in the light of the surveyed literature in the next section. Consider now the findings concerning whether the hypothe- ses dealing with the differences between the two professional and technical groups were supported or rejected by the statistical analysis. (Hypotheses I and IV are discussed at this point, since only the findings Of Operation P & T are necessary to the consideration of these two hypotheses. Hypotheses II, III, and V will be considered after the findings of Operation H have also been presented.) Table 3.1--Comparison of objective vs. subjective items with respect to ability to discriminate between pro- fessional-technical criterion groups. m Objective Subjective _Number of tests performed on items 32 45 Number of tests which discriminated between criterion groups 4 7 Per cent of tests which discriminated between criterion groups 12.50% 15.56% No significant difference between objective and subjective items in ability to diScriminate 44 Hypothesis I is rejected by the analysis, according to the data presented in Table 3.1. Of the 32 tests done on Objective items, four discriminate between the two pro- fessional and technical groups (12.50%). Of the 45 tests performed on subjective items, seven discriminated between the two groups (15.56%). Although a slightly larger per- centage of the subjective than of the objective tests did discriminate, the difference is not significant (chi-square, p g_.10). Hence, it is more accurate to reject the hypothe- sis and conclude that the subjective and the Objective items were very nearly the same in their ability to discriminate between those professional and technical applicants who were offered employment and those who were not Offered employment. Table 3.2--Comparison of items belonging to categories 2, 5, and 7 vs. items belonging to all other categories in ability to discriminate between professional- technical criterion groups. ii _:_ Categories All Other 2, 5, and 7 Categories Number of tests performed on items 34 43 Number of tests which discriminated between criterion groups 7 4 Per cent of tests which discriminated between criterion groups 20.59% 9.30% No significant difference between items in categories 2, 5, and 7 and items in all other categories in ability to discriminate 45 Hypothesis IV is rejected by the analysis. As is shown in Table 3.2, of the 34 tests performed on items belonging to categories (2), (5), or (7) of the detailed e priOri cate- gorizations, seven discriminated successfully between the two Professional and technical groups (20.59%). Of the 43 tests performed on items belonging to any of the other cate— gories, four discriminated between the two groups (9.30%). The difference is not significant (chi—square, p‘g .10). Again, although a larger percentage of the items in the selected categories (2, 5, and 7) discriminated between the criterion groups, the difference is not significant and so the hypothesis must be rejected. A total of 66 chi-square tests were performed for Opera- tion H (see Appendix VIII), the operation to investigate possible differences between the hourly applicants who were Offered employment by General Electric and the hourly appli- cants who were rejected for employment by General Electric. Of the 66 tests, only five discriminated between the two groups at the 10% level or greater. Since 6.6 of the tests could be expected to show significance at the 10% level by chance alone, the two hourly groups are even more alike than chance would predict; that is, the criterion does not separate the two groups in this case as well as a simple random separa- tion Of the persons involved could be expected to do. Since this is the case, the most profitable thing to do may be to describe the tendencies which do appear, remembering that they 46 are pply tendencies which may be of interest in formulating hypotheses for future research. The next step is to consider this statistical infor- mation concerning the hourly applicants in order to reach some conclusions as to the validity of the a_priori hypotheses. Table 3.3--Ability of all items to discriminate between pro- fessional-technical criterion groups vs. ability of all items to discriminate between hourly criterion groups. ——__—__-=====:.-= Professional- Technical Hourly Number of tests performed on items 77 66 Number of tests which discrimi- nated between criterion groups 11 5 Per cent of tests which discrimi- nated between criterion groups 14.28% 7.58% No significant difference in the ability of all items to discriminate between professional-technical criterion groups and the ability of all items to discriminate between hourly criterion groups Hypothesis II is rejected by the analysis, according to the data presented in Table 3.3. Eleven of the 77 statis- tical tests (14.28%) indicated significant differences between the two professional and technical groups, while five of the 66 tests (7.58%) were significant for the two hourly groups. This difference is not significant (chi-square, p.g .10), even though there is a tendency toward better discrimination 47 between the two professional and technical groups than between the two hourly groups as was hypothesized. Table 3.4--Comparison of items belonging to category 2 vs. items belonging to all other categories in ability to discriminate between hourly criterion groups. A All Other Category 2 Categories Number of tests performed on items 11 55 Number of tests which discriminated between criterion groups 2 3 Per cent of tests which discriminated between criterion groups 18.18% 5.45% No significant difference between items in category 2 and items in all other categories in ability to discriminate Hypothesis III is rejected by the analysis (see Table 3.4). Of the 11 tests performed on items belonging to cate- gory (2) of the detailed a priori categorizations, two dis- criminated successfully between the two hourly groups (18.18%). Of the 55 tests performed on items belonging to any of the other categories, three discriminated between the two groups (5.45%). The difference is not significant (chi-square, p‘g .10). Hypothesis III, then, is also rejected on the basis of the significance test although there is again a tendency toward better discrimination between criterion groups by the items in category (2) than by the items in other categories. 48 Hypothesis V appears to be supported by the analysis. Two questions (but not the same two) from category (2) were significant for both operations, but, other than that, none of the other significant items for the professional and technical applicants were even in the same categories as the remaining significant items for the hourly applicants. Hypothesis IV names categories (5) and (7) as the other categories which would be especially important to the pro- fessional and technical applicants. There was a tendency toward support of the hypothesis, but the differences were not statistically significant and the hypothesis had to be rejected. Still, if the p_priori categorizations in general are taken as the basis for deciding whether or not the patterns are the same or different, apparently they are quite different, with a slight tendency (but not a significant trend) for factors concerning academic outlook, social poise, leadership, and way of relating to other persons to be more important, as expected, for the professional and technical applicants. Another possibility is that since the question- naire did not even discriminate at the chance level for Operation H, perhaps it is safer to conclude that there simply is no pattern of significant items for the hourly applicants to be compared with the patterns of significance for the professional and technical applicants. Thus, the "patterns“ are necessarily different. Summary ofygindinge Concerning Categories Ofpltems. Only one of the five hypotheses was upheld and four were rejected 49 by the analysis of the data. In each case of a rejected hypothesis, there was a tendency, on the basis of sheer ob- servation and the comparison of percentages, toward the hypothesized differences, but none of the differences was statistically significant. Still, the tendencies need to be examined for each question which did discriminate suc- cessfully between criterion groups, since they may provide some insight as to what might occur if other similar studies were conducted. The function of the next section, then, is essentially to describe the differences which did occur in the hope that the descriptions may be of value in suggesting further research along these lines. lpterpretation of Patterns or Responses by:lndiyldual Items The focus in this section will be the individual signifi- cant items rather than categories Of related items. A "composite picture" of the criterion groups (such as Kirchner and Dunnette developed in the 1957 study of long- versus short-tenure employees of the Minnesota Mining and Manufactur- ing Company) will bring to light each detail which revealed a significant difference between the groups. Professional and Technical Applicants. A composite picture of the professional or technical applicant who was Offered employment by General Electric includes the following characteristics: He preferred a job which allowed a great amount of contact with other people; believed that a super- visor's ability to deal with people effectively was less 50 important than some other aspects of the supervisory job (ability to keep the pressure on until the job is done, technical knowledge, etc.); was more likely than an appli- cant who was not offered employment to have spent four hours a week or fewer on athletics during his last two years in high school; derived his greatest pleasure while in high school from some activity other than achieving academic success and recognition; was more likely to enjoy discussion courses in college than an individual who was not offered employment; was more likely than a rejected applicant to say that he had little time for reading although he reads as much as he can, or that about the only reading he does is the newspaper and occasionally a few magazines; was thoroughly interested whenever he visited a museum: was more likely to have made one or more public Speeches during the past year than an applicant who was rejected: had helped to organize at least one athletic team, charity campaign, literary club, or other similar community activity; was more likely than a rejected applicant to have taken his first trip of over 100 miles alOne when he was 15 or younger: and he was more likely than a rejected applicant to want other people to think that he was "tough but fair." It is important to note that several of the distinctions included in the composite picture say only that the applicant who was offered employment "was more likely to" answer in a certain way than an applicant who was rejected for employment. 51 This does not necessarily mean that even a simple majority of the applicants who were offered employment answered in that way--only that significantly mpre_of them answered the item in that particular manner than Of the applicants who were rejected. For instance, in answer tO Item 21, 14 of the applicants who were Offered employment said that they preferred discussion to any other type of course in college, while 29 chose some other type of course as their favorite. However, only three of the rejected applicants preferred discussion courses, while 22 chose some other type Of course as most enjoyable. The difference is significant and, therefore, the applicant who was offered employment was "more likely to enjoy discussion courses in college than an individual who was not offered employment." Several of the significant items which make up this composite picture suggest that the individuals who were Offered employment tended to be more outgoing or gregarious in the job situation, in school, and in the community than those who were rejected. The fact that the individual who was offered employment generally preferred a job which allowed a great amount of contact with other people, derived his greatest pleasure while in high school from some activity other than achieving academic success and recognition (when all Of the other choices necessarily involved interaction with other students at least to some extent), had helped to organize various community activities, plus the evidence that 52 he was more likely than a rejected candidate to enjoy dis- cussion courses in college and to have made public speeches, lead one to believe that the interviewers may have sought out socially active, seemingly personable individuals who enjoy other people in general. Hodgson (1964), in his ex- tensive study, found slight tendencies in one department (Service Department) for the individuals who were Offered employment to be less nervous, more outgoing, more aggressive, and more stable than those who were rejected. Also, with reference to the "self-confidence" factor identified by Smith, Albright, Glennon, and Owens (1961), it is interesting to note that significantly more of the appli- cants who were Offered employment than of those who were rejected wanted other people to think of them as "tough but fair." It is true that only 11 of the Offered group answered the question in this way, but it must be taken into account that wanting others to think Of them as "capable" was almost a stock answer for both groups of professional and technical applicants. Only two of the not-Offered group preferred to have others think of them as "tough but fair.“ The difference, significant a < .10 level, perhaps indicates that these persons wanted to be recognized as having leadership ability. Curiously, while Smith, Albright, Glennon, and Owens (1961) found that devoting much time to reading was an important aspect of their "academic orientation" factor which discriminated between scientists chosen on a success Versus 53 non—success criterion, in this study‘the applicants who were Offered employment were more likely than the rejected appli- cants to say that they had little time for reading or that about the only reading they did was the newspapers and occasionally a few magazines. Slightly over half of those who were Offered employment admitted that they read very little, while over two-thirds of those who were rejected said that they devoted considerable time to reading. Perhaps those who were offered employment really did spend less time reading because they were more inclined to be active Socially than those who were rejected, or perhaps those who were of- fered employment were Simply more honest about their reading habits, but both of these possibilities are purely Specula- tion on the Slim basis provided by other tendencies. Hoprly Applicants. A composite picture of the hourly applicant who was offered employment by General Electric includes these characteristics: he had left his previous job because of the boss, the hours of work, a reduction of the work force, or just "other" reasons instead of for reasons directly related to advancement: said that he rather enjoys routine "once he gets the hang of it"; was most greatly influenced in his choice of major field by the possibility of achieving personal satisfaction: thought he was best described as socially introverted or a dreamer: and though he was least described as disliking routine or detailed work. 54 Since there are actually fewer Significant differences between the criterion groups for Operation H than even a chance separation of the individuals would predict, generali- zations concerning the tendencies above must be made with great caution. One particular point of interest, though, is that two of the five tendencies indicate that the hourly applicant who was offered employment typically rather likes routine work. One of the characteristics states Specifically that he rather enjoys routine "once he gets the hang Of it,” while the other points out that he considers himself leper described as disliking routine or detailed work. In this case, perhaps the interviewers were attempting to seek out those who would be content to carry on routine work without becoming bored or restless. If the work required of an hourly employee in the department stays much the same day after day, perhaps the interviewers were selecting those who could adjust best to the Situation. The tendencies listed in this section may provide at least some direction for the development of future question- naires which perhaps should handle professional and technical applicants differently from hourly applicants. Congruent with Hodgson's discovery of different personality tendencies in each department involved in his Study (when only profes— Sional and technical applicants were studied) is the fact that even greater differences in tendencies will appear for professional and technical applicants as compared to hourly 55 applicants. A questionnaire aimed at the professional and technical applicants might be designed to learn more about the applicants' sociability and self-confidence, if this is actually the kind of information upon which the company wants its selection decisions to be based. The question- naire for the hourly applicant might delve more deeply into the candidate's attitudes toward routine work, his patience in working with details, and his willingness to stay with a fairly unchanging job long enough to make it worthwhile for the company to train him. IV. SUMMARY AND CONCLUSIONS This study attempted to identify interviewer stereo- types through the analysis of biographical information collected during 1964 and 1965 by the Apollo Support Depart- ment of the General Electric Company, using as the criterion the company's decision to offer or not to Offer employment to an applicant. Two separate analyses were done, one for the professional and technical applicants and the other for the hourly applicants. The investigation was designed to serve a two-fold pur- pose: (1) to allow the company to determine to what extent the interviewers and managers made selection decisions on the basis of characteristics important to the company, and then (2) either to formulate a statement Of what factors are important or to re-evaluate the emphasis being placed on various pieces Of biographical information. pesignppr the Study Sample. Operation P & T investigated the differences between Group I, the professional and technical applicants who were interviewed and Offered employment by General Electric (N = 56), and Group II, the professional and tech- nical applicants who were interviewed and rejected for employment by General Electric (N = 33). 56 57 Operation H investigated the differences between Group III, the hourly applicants who were interviewed and offered employment by General Electric (N = 49), and Group IV, the hourly applicants who were interviewed and rejected for employment by General Electric (N = 36). lpstrument. The source of the biographical information was an interview supplement form which was completed by 201 applicants for employment with the Apollo Support Department. Method. An.item analysis was performed on the biographi- cal data. Specifically, a chi-square test was used to in- vestigate the discrepancies between an expected distribution and the Observed distribution of responses for each item. In many cases two or more responses to one question were combined into one cell of the contingency table in order to insure stability in the chi-square test or Simply to facili- tate the analysis. Questions which could be multiply answered were Specially handled, and each multiple-answer question was tested in at least two different ways. Result§_andpiscupsion Only one of the five e_priori hypotheses was upheld by the analysis, while four were rejected. The number of Significant differences between the criterion groups was Slightly above the chance level in the study of the profes- sional and technical applicants, but it was below the chance level in the study of the hourly applicants. Because the differences between criterion groups were not clearly defined 58 in either study, only Hypothesis V, which predicted that the patterns of Significant items for the professional and technical applicants would be different from the patterns of Significant items for the hourly applicants, was upheld. Briefly, the rejected hypotheses were: I. For the professional and technical applicants, a larger prOportion of the subjective items than Of the objective items will discriminate. II. More items will discriminate for the professional and technical applicants than for the hourly applicants. III. For the hourly applicants, items concern- ing the applicant's attitude toward the job Situation will discriminate better than other types of items. IV. For the professional and technical applicants, items concerning the applicant's attitude toward the job Situation, his "academic" life, and his social and leadership qualities will discrimi- nate better than other types of items. Attention was then focused on getting as much informa- tion as possible from the tendencies which did appear. It seems that the professional or technical applicant who was offered employment by General Electric tends to be more out- going socially and is perhaps more self-confident than the professional or technical applicant who was rejected for employment. Also, the hourly applicant who was Offered employment tends to claim to enjoy routine or detailed work more than the hourly applicant who was rejected for employ- ment. 59 Conclusions This study did reveal some tendencies, at least for the professional and technical applicants, which appear to be worth more investigation. The hypotheses for this study were based primarily on the kinds of factors which have been found to be important in studies of biographical data against criteria of jOb performance or job tenure. The e_priori categorization (detailed) of the items and the hypotheses essentially separated the life of the individual into parts-- education, work experience, hobbies, community activities, and SO forth--as has been typical in studies of biographical information. However, this study suggests that such divisions should not be the primary concern in investigations of biographical data as a predictor of whether or not the person will be Offered employment. It suggests that investigators should search for single traits or characteristic behaviors which would cut across all these dimensions in the person's day-to- day living. Perhaps, as this study suggests, an entire investigation of professional and technical persons could be devoted to determining to what extent interviewers are more likely to offer employment to individuals who are outgoing and self-confident in school, in their work, and in their communities. The tendencies which appeared in this investi- gation might Show up more clearly in studies with larger N's. AS for the hourly applicants, the results tended toward upholding the earlier Speculation that the questionnaire 60 does not apply to them as much as to the professionally- Or technically—trained applicants, even though the difference was not Significant. Further work with hourly applicants might begin with a search for items which apply more di- rectly to them. If Metzner and Mann (1952) were correct in concluding that there is more discrepancy for blue-collar employees than for white-collar employees between the way they answer a questionnaire and what they say during an inter- view, then a questionnaire may never be very useful as a predictor of eventual selection or rejection. But this dis- crepancy is something which must be investigated much more deeply than Metzner and Mann studied it before discarding the questionnaire as a possible predictive device for hourly applicants. In conclusion, let it be noted that the lack of any clearly defined organizational stereotypes may not be con- trary to what the company would want. If it is true that the interviewers, collectively, did not follow some prescribed list of company-set values, then it is equally true that they did not adhere to some other set of values which Should not be determining factors in the selection decision. Having several interviewers, including both personnel interviewers and line managers, contribute to the selection decision for each applicant is apparently a successful method of random- izing individual interviewer stereotypes, and this method may be a strong point of the department's selection process. 61 This is a value judgment which must be made within the com- pany by those responsible for personnel selection in terms of how they want their selection process to function; it is not the purpose of this study to determine whether it is good or bad that no strong organizational stereotypes were identified. The study was designed to identify such stereo- types if they existed so that the company might reinforce desirable stereotypes or try to get rid of undesirable stereotypes. Suggestions for Further Research The next step in the attempt to use biographical data in order to identify interviewer stereotypes might be to try questionnaires more Specific to the employment group until the factors most affecting selection are identified, and it is determined whether or not these factors can be singled out early in the selection procedure. Such an approach to the problem may appear to be a step backward when the general effort in studies of biographical data has been toward in- creased generalization. However, the emphasis has always been on the applicant. In Shifting the emphasis to finding out what the interyiewer is doing, it may be best to move backward temporarily in order to establish some bases from which to generalize. This could very well be faster in the long run than searching too generally and hoping to stumble onto some basic principles. By breaking the research down into more manageable pieces, investigagors may be able, in 62 relatively Short order, to identify factors common to several occupational groups and then proceed toward increased generals ization concerning interviewer stereotypes. BIBLIOGRAPHY Albright, L. E., and Glennon, J. R., 1961. Personal history correlates of physical scientists' career aspirations. J. Appl. xflpcmmsa mmm .comummspumSmcmumamHuHss m mmumoflch* III mmm. a mme.m mam.md m ma m.p.n mam.dm «we.>m mm mm m.o.m w III . www.md Shd.mm ma mm £.m.o mes a a ksfl.aa www.mm ma mm 6.. m III mmo. a mmm.m NS$.¢H m ea O.p.o N>¢.¢N mNm.H¢ em Ne Q.m w . . omm.ma omo.mm mm am A mmo v mas m a omo.ma omm.m~ oa an m m mee.m Smm.efi Ha NH mGOmmmn III WNN. 0H H mHOE .HO N 0>MU A..NV Smm.om m¢¢.mm ma mm common H m>mw Agav *N mma.m~ asm.ma mm as A no .a.m III sum. H .m.m.o.o.m A.NV ewm.m mme.ma m SH A A.dv *m mmo.oa www.md 0H ha a Any In: mew. N mmS.OH mom.mfi 0a ma m HO .o.p Amy med.¢d «mm.mm ma mm n no .M.O.Q.m Adv *m -u- mma. a omm.sa omn.om ma mm m.m.m.e omn.aa omm.md 0a Hm O.Q.m a m Ozam> .m.p HH H HH H Amvmmcommom coaumwso NX . moaocosooum mowocmsvonm pmpommxm Uo>Hmeo AHH % H mmbomwv 8 £ m ZOHBdmme mom MUZGUHmHZOHm ho mflm>mq QZd .mMDA€> NX .mmamdfi MOZmOZHBZOO HH> NHQmemd 106 poscHucoo III Hmo.H www.mm mOH.mm mm Sm v.0 mOH.m 5mm.eH S SH m.m.n.m «H III HHO. mmfiom Hmmofi m m “40.6.0 Hmm.>N mmH.NH mm Ne Q.m mH III mum. Hmm.dm mem.HH mm OH m.m.p.o mem.m Hmm.OH m NH Q.m NH III woo. mem.> SmH.mH m MH m.m.p.o SmH.mN mem.mm mm mm m HH mwm.dH mmm.em NH SN mOSHx III omH.H bosom HO m A.mv www.mH mem.om om mm xno3 mo mchx OHOE no 4 A.HV *OH www.mm www.me Hm ed u no .m.o . .msflsflsflsfim Amv III was. oam.mH oms.mm NH mm u so .o.c.o Ame smo.mm www.mm Hm mm E no .x.m Redd... 3 *3 III mmo. Sem.HH mmm.Hm NH Hm O.O.O.Q www.mH Sem.mm mH em 8 m . . ooo.mH ooo.mN HH mm m.p.o.p mo v OHm e ooo.mH ooo.mm Hm mm m m III . oom.wm oom.me mm me o mmo H oom.m oom.OH m m 9 S m ode> . .m.p HH H HH H Amvomcommom GOHumOOO mx mOHUQOSUOHm mmHosmsooum pouoomxm Uo>uomno “H / poocHusoo I HIIIIIIIIIII 107 possHusoo In- Nmm. H HHA.m~ www.ma mm om n no .o.o.n.o A.mv mom.oH aHs.mH NH mH o A.HC *om In- new. H mmo.mm Hem.ow Hm me o no 9 AND Ham.mm mmo.ma mm «a n no .o.o.o AHV *om .I. owe. H Hou.Hm mmm.>m mm em o.o.o.n mm~.0H HOA.AH 0H 8H m mH In- mmH. H omo.sm OHm.ma pm we ooHnno H no 0 A.mc OHm.m omo.m m m mooHnno onos no N A.HC *oH II- emu. H Hmm.mH moo.mm 4H mm o INC moo.sH Hmm.sm oH mm o no .o.n.m AHC *mH . . mam.m moa.mH mH mH o 0H v Hes N H moa.H~ mmm.om 8H 04 o.n.m 5H . . omH.mH a>m.mm mm mm o.o.o . oH v moo m H Hem.mH omH.mm m om . n.o oH In- one. H msm.mH www.mm oH mm onos no m A=NC mmo.sH msm.m~ sH om moHnH>Hnoo Ho3mm no N A=HV *mH II- mHo. H omm.mm oms.mm mm mm onos no N A.mv oms.m omm.oH 0H 6H an>Hnoo H no 0 A.HC *mH In- moo. H www.mm ans.am mm mm n no .o.o.o.o Ame ems.sm www.ma mm no H no .a.m.n 1H3 *mH m osHm> .m.p . HH H HH H Amvomsommom SOHumosO Nx mOHUSOOUOHm mOHosoonnm pouoomxm po>uomno COUCHDCMMHMHHHHHHHHHHHHN 108 poscHusoo www.0H www.mH OH mH .s AHV www.mm me>.om mm mm .H III mme. m .o Amv mmm.>m «OH.om mm as .n Ame HSH.Hn www.mm mm mm .m AHV *mm III mmH. H m>m.wH mNH.mN mH mm o.p mNH.MH mum.HN «H Hm n.m mm III mmH. H Nom.mH mmH.om mH Hm n mmH.¢H Nom.NN mH mm m.o.p.o.m SN ooo.m ooo.eH m HH m III 0mm. N Hmo.m mOm.mH OH mH o.p mom.eH Hmo.wm «H Sm O.Q.m mm .0 umooxo III oncommou has poxumfi 3mm Om OOSHm SN me O 0 AT «x .m v mooHo> oonoomxm H m n.o mm III Ono. H Hmw.mm mHm.¢m mm mm o no .p.o.n A.mv mHm.mN Hme.mm om mm o A.HV *Hm -I- okm.H H sHs.m www.mH HH 0H o no a Ame www.mm SH>.m¢ mm mm .O.m AHV *HN III Nmm. H www.mH moH.HN HH mm o.p.o.n SOH.>H www.mm mH mm m mm III mHo. H NOH.OH mmm.0m HH om o.p.o.n www.mH moh.em mH mm m mm . . omN.w omS.OH m HH 0 0H v mom m H oms.mH omm.mm mm mm n.o Hm m osHm> .m.p .HH H HH H Amvomsommom GOHumoso Nx .mOHUSODUOHm mOHocosomHm pouoomxm po>nomno poocanoo a HH> xHozmmme 109 COSSHDSOU III me. HNH.HH mSm.mH NH mH o.p.o.n mSm.HN «NH.Sm HN mm m on NmH.m mom.m H HH C III mmo.H mNm.m SSo.mH m «H o mmm.¢H mHH.mN SH wN m mm mmN.OH HHS.mH OH mH p.o III mmH. mem.m HmH.SH m mH Q mmH.m mom.mH OH NH o.m Hm III mmm. omS.mH omN.mN NH HN Hosmm no N A.Nv omN.SH omS.mN mH Hm mchx OHOE HO m A.Hv *nm www.mH eHn.mm 0N mm H Amy In: mmo. mmH.Sm aHm.ma mm on o no .n.o.n Ame mNm.SH HSH.on SH Hm O HO .o.m AHV *mm . . HwH.HH mmm.wN OH mN p. we v New m mmm.mH HoH.Hm mm Sm n.o mm III Smo. Som.m www.mH m mH o.p mmH.SH Som.mN mH mN o.m Hm II- mom.H mSm.S mmH.mH OH HH pom H no 0 AND mNH.mN mSm.H¢ mN we ofios m SH mnofl OHOE no N AHV *Om III HHm.N SHN.NN mmS.Sm mH He ouoE HO S A.Nv mmS.OH SHN.mH HH mH mOHuH>Huoo Hozmm no w A.HV *mN m oSHo> HH H HH H Locomaommmm SOHnmmso Nx mOHUSOOUOHm mOHososoonm pouoomxm po>uomno ' 'i ill il COSCHUGOO I HH> NHQmem¢ 110 poscHucoo III mmN. H mmH.mH Hnm.om SH Nm o.p Hmm.eH mmH.mN mH HN O.Q.m we III . ¢m¢.HH mom.mH m NN o.p.o mHm H H oom.Hm HmH.om Hm Hm n.o HH .mOHuH>Huom on we H no 0 A.NV OHOE no N poxumfi 3mm Om OOSHm m m mOHuH>Huom o AI.NX .m v mOSHm> pouoomxm OHOE no N A.HV *me . . oem.m 0mm.mH mH HH o ANV Ho v Hmo o H oom.mm oao.mm mH ma o no .o.n.o HHS *mH III mmH. H mmH.HN www.mm NN mm o.o.Q mom.HH mmH.ON HH HN m NH III . www.4H NMH.mN mH MN o.p.o.p Hmm H H NmH.mH mom.Hm mH mm o HH . . HHo.OH www.mH m HN o.o.o.p 0H v moo m H www.mm HHo.mm Sm mm m 0H .m umooxo uncommon III was poMHmE chmHom N SHSO ooch N o Q o.Al mx .m.v mosHm> pouoomxm Hm mm m mm .m umooxo III oncommou was pmxumfi Bow Om OOSHm m S o.o.o.n o AT.NX .mv mmsHm> pouoomxm SN me o mm .. . omN.mN omS.mm mH me a mo v mom H H omS.m omm.oH HH NH o.o.o Sm m osHm> .m.p HH H HH H Amvomcommom SOHumoso Nx mOHOGOOOOHm mOHocmsooum pouoomxm po>uompo UOSSHHSOO I HH> xHazmmmd 111 OOOSHuGO0 III . SNH.SH MS0.0N HN ON v.0.n SoH N H mSm.mH SNH.mm «H on H mm III OHO. H mmm.HN SO0.0m ON mm O.O.0 SO0.0H www.mH m ON Q.m HO III OOH. H OO0.0H OO0.0N SH SN 0.p.0.n omHm OO0.0H OO0.0N OH ON O OO III HOO. H HOn.NH OMO.HN OH HN 0 OmO.mH HOO.HO OH mm O.O.Q.m mm III mON. H OOH.OH HO0.0m OH HO 0.0.Q HmO.HH OOH.ON OH NN m NO . . HON.mH mHH.Nm HN ON O.O mo v mHH H H HHS.mH Hmm.mm m mm o.n.m Hm III mNN.N H OO0.0H NOm.HN SH HN 0.0 NOO.SH Om0.0m HH Hm 0.Q.m Om III OON. H mOO.SN HmH.SH SN OH 0.3 HmH.O mO0.0 O O o.m OH III OON. H OOH.OH HO0.0m SH NO 0.9 HOO.HH OOH.ON OH HN O OH III mSN.H H mS0.0H SNH.ON OH mN O.O.0 SNH.SH mSO.mN ON SN n.m SH III OmO. H OHm.OH OOO.SN SH SN 0.0 OO0.0H OH0.0N OH mN 0.9.m OH O osHm> .m.p HH H HH H Amvomcommom :OHDOODO mx OOHOSOOUOHN OOHUSOOUOHN pou0mmxm Oo>uomno poocHnooo I HH> xHOZSSEH 112 III NOH. H ONO.NH OO0.0N HH NN 0.U OO0.0H ON0.0N SH ON 9.m HO NOO.NH OHO.HN HH ON 0.0 III HOH. N OHO.HH OOH.OH NH OH 0 OmN.S NOS.HH O OH 9.m OO mO0.0 SH0.0H O HH C III HMO. N HHS.OH OON.ON SH ON 0.9 NO0.0 OO0.0H OH OH 0.m NO III . SHO.HH OO0.0N OH SH 0.9 NNH N H mmo.oH NHm.oN mH on o.o Hm III HOO. H OO0.0H NOH.OO ON OO O.O NOH.HH OO0.0N HH HN 0.9 OO III . OO0.0 SOH.OH NH HH O.O.0 HOH H H SmH.NN www.mm 0N HH n.o mm . . WOO.SN SOH.OH HO OH m.m.m.m 0H v moo m H NOH.m nmm.S N HH n mm . omN.m omN.NH O OH m.n.o.n In- H N m H H omN.HN omN.mm NN on H mm .9 umOUxO III omcomwmu Sam Ooxume Bow om O0SHm O Al.mx .Ov mOsHm> OOuUOmxm ON SH 9.m SO OON.O OOS.O O O 0 III Omo. N ONH.OH OS0.0H OH SH 6.0 ONO.SH OS0.0N OH ON 9.m OO O msHm> _ .HHO HH H HH H Amvmmcommmm OOHumOOO mx OOHUOOOUOHN OOHUSOOUOHN Owu0wmxm Ow>nmm9o UOSGHHGOU I HH> XHQmemd OOOOHUOOU .pmnhHmsm OHO3 OOOHummsU Ommnu 309 No mowusssmeo as no“ OO omsm new > XHOSOQQO OOO .SOHummso HO$OSOIOHQH9HOE m OOOOUHOOH * 113 II- HHm. H mmH.HH Nom.0N oH NH n.o.n Nmm.mH mmH.oN NH mN o.o.o o Nmm.HH moo.NH HH mH n.o.n In- mNm. N mmH.HH Nmm.mH 0H NH n.m.o oom.m 00H.HH m mH o.m m . . NHo.HH mmm.HH OH OH o.o.n No v NHN m H mom.HN NHo.Hm 0N mm o H In- How. H www.mH NmH.NN HH HN o.o.n NNH.mH mmm.oN HN mN m n mHm.N HmH.0H N HH mcomoon III OOO. H OHOE no N O>MO A=NO HmH.mN mHm.mm ON mm common H o>mo A.HO *N mmm.on .NHH.NH mN OH H no .a.m II- omH.N H .n.o.o.u.o A.NO NHH.m mam.N m m n L.HO *N Hmo.0H NHm.oH OH NH S Inc mo.v NNH.N N HmH.mH NHm.ON m mN o ANS mom.HH NnH.mH NH mH n o n o AHO *N II- NmN.H H HNH.mH oom.mH HH HN m.n.o.o oom.HN HNH.NN HN NN o.n.o H m ODHO> .H.O >H HHH >H HHH Amvmmsommmm OOHumwsO mx OOHOOOOOOHN OOHOSOOOOHO . Omu0mmxm OO>HOm9o 3H s HHH mmpomoc N 203220 mom 5243563 no mHMSHH O24 .mmoqflw mx .mmgmg sozmozHezoo HHH> NHQZMQQN UQJCHU‘COU l HHH> UOHQZMHAHANAV ii OOOSHUSOU 114 III OON. NHN.HN OOS.ON NN ON 0.0 OOS.O NHN.O O OH 9.m HH OO0.0 OO0.0H OH NH m.o III OHO. OH0.0 OOH.O O HH v.0 OOH.HH OH0.0H NH OH 9.m OH III ONO. OO0.0H SO0.0N OH ON 6.0 SO0.0. OOO.HH OH HH 9.m NH HOO.HH OOH.OH NH OH H.0 III HOS.H OOH.S SH0.0 O NH O.0 OO0.0H HH0.0H OH OH 9.m HH OO0.0H HH0.0N ON ON OOSHx III OSH. H um3mm HO O A.NO HH0.0H OO0.0N OH HN 9903 HO OOSHx OHOE no H A.HV *OH HO0.0 OO0.0H S OH M HOn.m.U III . . .H...H.9 HOV OHO H N OOH.OH HO0.0H OH OH H HO .o.s.m ANV OON.HO OO0.0H OO OH E HO .9 . ~m~msus0~nflsm AHV *0.“ III OHN.H ON0.0H OS0.0N SH HO O.p.0.9 OSO.HH ONO.SH HH OH O O III NSH. OOO.SH OHH.SN SH ON O.O.0.9 OHH.OH OO0.0H HH OH O O HOH.OH OOO.HH OH NH O III OOO.H NO0.0H OHH.OH OH ON 0 HO0.0 OH0.0H O OH 9.m S o oSHm> .n.o >H HHH >H. HHH Amvomcommom coHnmooo NX , mOH0GODUOHm OOHUSOOUOHN OOuUOmxm Om>nmm9o HHH ifll JIII oosaanoo . HHH> xHozmooH OmscHucou .0 no A mmmcommmu Omxumfi 3mm 0m moch S O m no A ANV O.Al mx .OV mmsHm> Omuommxm HN HH O no .o.m AHV *HN .mOmHHoo Ucmuum uoc OHO mucmoHHmmm mHuson m3» m0 umoE moch >H Ocm HHH mmzouw How OGHNmHmam nuHOB uoc mum ON dam .NN .HN .ON mGOHummso . Smm.mH mHH.om mH mH m.w Hmo H H nHH.SH Smm.mm mH mm u.n.m mH .mmonmo whoa no N 0H0: ON OH mUHmmo H no O x.NV 115 III mucmoHHmmm mmm£u mo 3mm om muaHm O O mmUHmmo O AT.NX .OV mmsHm> Umuommxm whoa no N A.HV *OH III OHO.N H OON.OH NOS.HO OH OO m ANV NOS.O OON.OH OH OH .U.Q~m “HO *OH HN0.0 OSH.OH S OH ©.u III HOO. N OS0.0 OO0.0 O O n OHH.OH OO0.0N OH SN m SH III OOO. H NOH.HN OOO.NO NN NO m.U.o OO0.0 NOH.HH O OH Q.m OH III OHO.H H HHS.O OON.OH HH HH muofi Ho O AgNO OON.NN HHS.OO ON OO mmHuH>Huom Hmzmm no N A=Hv *OH III SHH.H H NO0.0H OO0.0N OH ON wHoE no N A.NV mom.HH mmm.Hm «H «N muH>Huom H Ho O A.HV *OH III HHO. H NOH.OH OOO.HN OH ON m no ~m~O~o.m ANV www.mm .mmH.mm mm mm H no .n.m.n AHO *OH O msHm> .O.O >H HHH >H HHH Amvmmcommmm COHummao «x muHocmswmum mmHucmsvam Omuummxm Um>ummno J I II CODGHUGOU I HHH> NHQmem¢ 116 UmSCHucoo OOO.S_ OON.OH H SH 0.0 III SHH.H OON.O HHS.OH NH OH 9 HH0.0 OO0.0H OH HH U.m HO III HnH. mmN.m HON.NH O NH non H go o ANV HOS.ON OON.OO ON SO 050: m cH mnofl whoa no N AHV LOO III OHO.H OOO.HH OON.NN NH ON once no S A.NV OON.SH OO0.0N ON ON mmHuH>Huum uwzmm no O A.HV *ON NH0.0 OO0.0 S O x no .3.>.o.a AHV mNN.mN mNN.mm HN mm a no .m.e.H III mdflo NHNSNUsHNmNU AWV Nno.mN www.mm HN NH u no .m~x.H ANO OS0.00 HN0.0H HO OH > no .H.v.u~n.m AHV *ON III NOH. HO0.0H OHH.OH NH OH m.©.u OHH.OH HO0.0N SH OO n.m ON III OON. HH0.0H OO0.0N SH HN Q OOO.NH HH0.0H HH ON m.w~v.o~m SN III OOO. OOO.SH OOO.SN SH ON m OOO.HH OOO.SH NH SH U.o.Q.m ON . . OSH.OH HNO.SN OH HO 0 0H v HON N HNm.m mNH.mH NH NH u.n.m mN III HOO.N HOO.SH OH0.0N HH NO m no .O.U.Q A.Nv OH0.0H HOO.HN OH HN m A.HV *HN m mSHw> .O.O >H HHH >H HHH Amvmmaommmm COHummso NX mmHoaoskum mmHocmsvah Umuommxm Om>ummno mmscHuaou I HHH> an2mmm¢ 117 OmacHucoo III HOH. III OOO. H OH0.0H HO0.0H NNN.NH OSS.ON HO0.0N OH0.0N OSS.SH NNN.OO OH OH HH OH HN ON OH NO .GOHumNHcmOHo HMGOHmmmmonm m on OGOHwn Ho m ANV O no .056 HO 6.0.Q m *mH .NH .mmsommmm UHHQUQ mam mOmE 0>mn .muaoumm GHQ: .mummmm HMUchumu UmSmHHQDQ 0>mn muGOUHHmmm mHudon msu mo 3mm om much >H mam HHH museum How OGHNmHmcm nuuo3 no: mum HH can .OH ~OO .OO OGOHummso III OHH. H OH0.0 OO0.0H OH OH 6.0 OOO.HN oH0.00 HN SO A SO III NHO. H OO0.0 OO0.0 O OH 0.6.0 OO0.0N OO0.00 SN OO O OO III . OOo.OH HOO.SH HH 0N v.0.n HOo H H Ho0.0H OOo.ON OH HN m OO OO0.0 SO0.0H O OH O.0 III OOO.H H NOO.O OHO.O H HH n OOO.oH OHo.mH NH SH m.m HO NNN.NH OSS.SH O HN mOGHx ONCE .HO m... A.HV *WW III SHo. H OOO.SN SHH.NH ON NH O Ho .O.m.n ANV SHH.O OOO.HH O OH O Ho .o.m AHV *OO III HOO.H H OO0.0H OOO.SN OH OO m.O.o OOO.HH OOO.HN SH OH Q.m NO m msz> .m.O >H HHH >H HHH Amvmmcommmm coHummsc Nx . mmHoamsvmum mmHoamsvmnm Umuummxm Ow>nmmno GmDGHuSOU I HHH> NHQmem¢ UmscHucoo 118 .m ammoxm mmcommmu ham vmxHME 30m om mUCHm O O 0‘9 0 ml mx .Ov 332, @38me ON OH O. NO OSH. NH0.0H OOH.SN SH ON m.U OOH.HH NHO.HN OH ON 0.9.m HO NSO. HON.S OOS.OH O O m.© OOS.HN HON.OO ON OO U~Q.m OO HOH. OS0.0N HN0.00 ON OH 0.9 HN0.0 OS0.0 S O m.m OH HOH. NNH.ON OS0.0N HN ON v.0.n OSO.NH NNH.OH NH ON O OH OOO. ON0.0H SSO.SN OH ON w.©.o SS0.0H ON0.0N HH ON Q.m SH OSO.H OSH.NH HN0.0H OH OH m.U HN0.0N OSH.OO OH OO o.n.m OH SOO.H SHN.NH OOS.OH OH OH 0.U OOS.OH SHN.OO SH OO U~Q.m OH OOO. HHH.O OO0.0H HH NH 6.0 OOO.HN HHH.OO ON OO Q.m HH .mmHuH>Huom OO OH H no O A.Nv wuoE Ho waHME 3mm 0m moch N O mmHuH>Huom O.Al Nx .OV mmsHm> Umuommxm muofi uo N A.HV *OH O 09Hm> .m.O >H HHH >H HHH xmvmmcommmm QOHummDO Nx mmHucmsUmnm mmHocmsvmnh Uwuummxm Uw>nwmno Ill L OmscHucoo I HHH> anzmmm¢ 119 OmscHucou III . OO0.0H OO0.0N OH ON 0.9 000 o H ooo.O ooo.OH O OH m.m NO III HOO. H OOO.S OOO.HH O OH 0.3 OO0.0H OO0.0N O HN m.m HO III HSS. H HO0.0N OHO.SO SN OO 0.0 OHO.S HOO.HH O OH U.n OO III . HHN.S OOS.OH OH O 0.6.0 OHN N H OOS.ON HHN.OO ON HH n.m OO III OOS. H NO0.0 OOO.NH S HH O.0.O.o.n OOO.HN NOO.HO ON OO O OO III HNO. H OO0.0N OOO.SO ON OO U.n OO0.0 OOO.S O S m SO OH0.0H OO0.0N OH ON 6 III HOO. N HOH.O OH0.0 O OH v.0 HON.O OOS.HH O OH Q.m OO III OOO. H OO0.0H OOH.ON OH ON v.0.n OOH.OH OOO.HN OH ON O OO III . SOO.HH OH0.0H NH OH m.©.o 000 o H OHO.SH SOO.ON SH SN n.m HO OOH.O SO0.0H O O m.w.n III HON.N N OH0.0H OOH.SH NH OH 0 NOO.NH OH0.0H O ON O OO O $5Hm> .m.© >H HHH >H HHH Amvmmcommmm GOHummsO Nx wwHocmfivah moHUcmsvmum Omuummxm ©w>nmmno Omschcoo I HHH> xHQZOOOO 120 OO0.0 HOO.HH N OH m OO.v OH0.0 N SSO.S ONO.NH OH OH O HO0.0 OHH.SH HH OH Q.m HO . . OOS.O OON.OH OH OH 0 OH v OON O H OON.HH OOS.OH O NN n.m OO O mSHm> .m.© >H HHH >H HHH Amvwmaommmm COHummso NX mmHUcmsvmum mmHoamsvam Umuummxm Um>ummno OmscHucoo I HHH> xHQZOOOO "7'11”1111111'11711011 ITS