AN EXPLORATORY ANO DESCRIPTIVE STUDY IN THE APPLICATION OF A MARKENNC PERSPECTIVE TO THE COLLEGE CHOICE PROCESS: AN INSTITUTIONAL APPROACH Dissertation for the Degree Of Ph. D. MICHIGAN STATE UNIVERSITY ‘ ‘ LEONARD EUGENE SHEFFIELD ' 1975, «I III III IIII III III III III III III III II IIII III 1293 10 This is to certify that the thesis entitled AN EXPLORATORY AND DESCRIPTIVE STUDY IN THE APPLICATION OF A MARKETING PERSPECTIVE TO THE COLLEGE CHOICE PROCESS: AN INSTITUTIONAL APPROACH presented by Leonard Eugene Sheffield has been accepted towards fulfillment of the requirements for Ph. D. degree in Marketing /}/[¢, ajor professor Date if’ /§/: /.5-A 0—7639 "Jmi ' , A 'L‘ "P1: 233‘ AR AOI—I OP ABSTRACT AN EXPLORATORY AND DESCRIPTIVE STUDY IN THE APPLICATION OF A MARKETING PERSPECTIVE TO THE COLLEGE CHOICE PROCESS: AN INSTITUTIONAL APPROACH BY Leonard Eugene Sheffield This study examined certain aspects of buying behavior within the non-profit setting of higher education. The col- lege choice problem was viewed as a purchase problem not sig- nificantly different from the type faced by consumers when purchasing economic goods. The recent trend toward more widespread application of marketing technology by colleges, particularly private colleges, suggested the need to examine the college selection process using a marketing perspective. The study had as its purposes; (1) to provide additional knowledge and under- standing of prospective college students' information search and informational source usage during the choice process, (2) to identify the importance of selected evaluative cri- teria used in the choice process, and (3) to identify seg- mental differences within a set of prospective students who had indicated a prior interest in a specific college. Leonard Eugene Sheffield A longitudinal research design was used which allowed time-dependent comparisons to be made on individual and group bases. Three time-reference points were included in the analysis: (1) pre-application, (2) post-application, and (3) post-enrollment. Within each of these periods, compari- sons were made between identified segments using chi—square analysis, to determine significant differences. Compari- sons were also made over time, using correlation analysis, to determine consistency in the importance attached to selected evaluative criteria by the prospective college students. The following conclusions were drawn with reference to the study's five major hypotheses. Hypothesis I: A buying intention statement in terms of the prospective student's choice rating of a particular college, i.e., first, second, third choice, etc., will serve to predict appli- cation and enrollment more frequently than other data available to the college. A buying intention statement indicating that a col- lege was the prospective student's first choice was found to be the best single predictor of student applications. No difference was found in the predictive quality of a second, third, fourth, etc., choice designation. Of those prospective students who made application, the purchase intention expressed as a first choice prefer- ence did not predict enrollment significantly better than any other choice designation. Leonard Eugene Sheffield Hypothesis II: Identifiable market segments of prospective students interested in a particular college, such as, the ACT1 segment and the SAT2 seg- ment will differ in their characteristics and behavior. Significant differences were found in the character- istics and behavior of the ACT defined and the SAT defined market segments. Other behavior determined classifications also produced significant differences. These identifiable differences between market seg- ments suggest an opportunity for colleges to develop special- ized marketing strategies to more effectively attract stu- dents. Hypothesis III: Purchase patterns as reported for the purchase of economic goods with respect to the level of information and degree of decisiveness will carry over to the college choice process. Some support was found for the carry-over of economic goods purchasing patterns to the college selection process. Those prospective students who considered themselves to be well informed when purchasing economic goods, also appeared to be better informed about colleges. Hypothesis IV: Prospective college students will change their assessment of the relative importance of selected evaluative criteria over time. 1ACT refers to the test of the American College Testing Program used for college admission. 2SAT refers to the Scholastic Aptitude Test of the College Entrance Examination Board and used for college admission. Leonard Eugene Sheffield The relative importance of the selected evaluative criteria for individual prospective students tended to change over the time period studied. This apparent lack of a firmly structured set of evaluative criteria is con- sistent with buying behavior theory, where buyers lack previous purchase experience for the item involved. Hypothesis V: Behavior determined segments of prospective college students will differ in the relative importance of selected evaluative criteria at different points in time. Certain behavior determined segments differed in the importance attached to the evaluative criteria at specific time-reference points and across time. One such example was with the private college enrollees. They were found to be less homogeneous in their evaluative criteria structure than were the public college enrollees. The scope of this study was limited to an analysis involving prospective students identified with one specific college. However, the existence of differences between market segments, as revealed by the methodology used in the study, indicates a need for all colleges to identify the characteristics and buying behavior of their markets, prior to planning their marketing strategy. The findings and conclusions of this exploratory study also suggest the need for additional research, both of a theoretical and an empiri- cal type. AN EXPLORATORYIAND DESCRIPTIVE STUDY IN THE APPLICATION OF A MARKETING PERSPECTIVE TO THE COLLEGE CHOICE PROCESS: AN INSTITUTIONAL APPROACH BY Leonard Eugene Sheffield A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Marketing 1975 @Copyright by LEONARD EUGENE SHEFFIELD 1975 DEDICATION To my wife Barbara and son Jeffrey ii ACKNOWLEDGMENTS I wish to acknowledge and express my appreciation to all those who have contributed to my completion of this research and the doctoral program at Michigan State Uni- versity. The expert assistance given by my dissertation committee helped to make this goal attainable. Dr. William J. E. Crissy, chairman of the committee, was a constant source of encouragement and positively oriented advice. Dr. Gilbert Harrell and Dr. John Fuzak also contributed significantly through their constructive comments and sug- gestions. To the entire committee, I express my sincere gratitude for the needed guidance which they provided. I also want to acknowledge the helpful cooperation given to me by the administration of the subject college and those students who participated in the study. And most importantly, I wish to recognize my wife Barbara for her unselfish sacrifice, dedication, and as- sistance in the accomplishment of our mutual goal, the attainment of the Ph.D. degree. iii TABLE OF CONTENTS DEDICATION o ' o o o o o o o o o o o o 0 AC KNOWLED GMENT S O O O O O O O O O O O 0 LIST OF TABLES . . . . . . . . . . . . . LIST OF FIGURES O O O O O O O O O O O 0 LIST OF APPENDICES . . . . . . . . . . . Chapter I. II. III. RESEARCH FRAMEWORK, INTRODUCTION 0 O O C 0 O O O O O 0 Problem Background . . . . . . Marketing Perspective of Educational Choice Related Consumer Behavior Theory . . . . General Research Purposes . . . . . . Research Approach Statement . . . . . Potential Contribution of the Research . . LITERATURE REVIEW . . . . . . . . . Higher Education Market Demand . . . . Market Share and Institutional Change . Private and Public Sector DevelOpment . . Early Developments . . . . . . . . Educational Competition Favored . . . Nature and Structure of Competition . . Philosophies About Who Should Go To College College Admission Trends . . . . . College Choice Factors . . . Geographic Dimension of Accessibility Student Migration . . . . . . Ability Dimension of Accessibility . Price Dimension of Accessibility . . Other Choice Factors . . . . . . Marketing in Higher Education . . . . Research Problem Statement . . . . . . iv HYPOTHESES, AND METHODOLOGY Page ii iii viii xv xvi OOQbWI—J |-‘ 13 14 16 17 18 19 24 26 30 37 36 37 4O 44 47 52 52 Chapter Market Segmentation: Theory and Schools of Thought . . . . Consumer Choice Behavior . . General Hypothesis . . . Areas for Research Hypotheses . Major Research Hypotheses . Research Design . . . . . VIADE Concept . . . . . Time Period Definitions . . ACT and SAT Group Definitions Research Methodology . . . . Pre-Application Period . Post-Application Period Sample . . . . . Data Collection . . Data Analysis . . . Statistical Methods . Post-Enrollment Period Data Collection . . Data Analysis . . . Data Collection Matrix Limitations of the Study IV. PRE-APPLICATION PERIOD ANALYSIS . Pre-Application Data Analysis . Preliminary Study Group . . Combined Descriptor Match . Relative Importance of Evaluative Criteria Rank Order Classification . Individual Descriptor Match . College Choice Designation . Combination of Significant Choi Summary of Analysis . . . V. POST-APPLICATION PERIOD ANALYSIS Post-Application Data Collection Post-Application Data Analysis Socioeconomic and Demographic Variables Education . . . . . . Income . . . . . . . Residence Value . . . . Mobility . . . . . . Vacation Companion Preference Goods Purchase Pattern . . College Information Level . Information Sources for Goods Purchasing Colleges Visited, Considered, and Applied V Page 54 59 63 71 72 73 74 74 76 76 78 78 79 8O 80 81 81 82 82 82 83 83 Research . . . . . 86 88 88 89 91 . . . . 94 ce Elements . 96 98 . . . . . 100 . . . . . 100 101 101 101 104 105 105 107 109 110 113 113 Chapter Page Decision to Attend College . . . . . . 116 College Information Level: Factor Evaluation . . . . . . . . . . . 118 College Information Level . . . . . . 119 Usefulness of College Information Sources . 121 Specific College and Major Intentions . . 124 Matched Condition Analysis . . . . . . 125 College and Major Intentions . . . . . 125 Early and Late Deciders . . . . . . 125 VI. POST-ENROLLMENT PERIOD ANALYSIS WITH PRIOR PERIOD REFERENCE . . . . . . . . . . 129 Post-Enrollment Data Collection and Analysis . 129 Post-Enrollment Data Analysis Methodology . 130 Longitudinal Analysis: Individual . . . 131 Combined ACT and SAT Groups . . . . . 134 Applied and Non-Applied . . . . . 135 Match Condition and Correlation: Combined ACT and SAT . . . . . . . . . . 136 Early and Late Deciders . . . . . 137 College Decision-Matched Condition . . . 137 Number of Applications . . . . . . . 138 Ability Variable . . . . . . . . . 139 Type of College . . . . . . . . 139 Individual College Level . . . . . . 140 Evaluative Criteria t2: Scaled Values . . 140 Applied and Non-Applied (t2) . . . . . 141 Private and Public Colleges (t2) . . . 142 Within Applied Group: Enrolled and Not Enrolled (t2) . . . . . . . . 142 Evaluative Criteria t3: Scaled Values . . 143 Applied and Non-Applied (t3) . . . . . 143 Private and Public Colleges (t ) . . . 143 Within Applied Group: Enrolle and Not Enrolled (t3) . . . . . . . . . 144 Evaluative Criteria: Within Group Association of Rank Order . . . . . . 145 Coefficient of Concordance W Analysis . . 147 Comparative Group Analysis: Applied Group . . . . . . . . . . . 147 Comparative Group Analysis: Non- Applied Group . . . . . . . . . 152 Enrolled to Private and Public Comparison 158 Selected College Characteristics . . . . 159 Student Aid Characteristics . . . . . 160 vi Chapter Page VII. SUMMARY FINDINGS AND CONCLUSIONS . . . . . 161 Purposes and Approach of the Study . . . . 161 Pre-Application Period Findings . . . . . 162 Post-Application Period Findings . . . . . 163 Socioeconomic Variables . . . . . . . 163 Goods Purchase Pattern . . . . . . . 164 Number of Colleges Visited, Considered, and Applied . . . . . . . . . . 165 Decision To Go To College . . . . . . 165 College Information Level . . . . . . 166 Intentions: College and Major . . . . . 167 Most Informed and Intended College . . . 167 Post-Enrollment Period Findings . . . . . 168 Evaluative Criteria t2: Scaled Values . . 169 Evaluative Criteria t3: Scaled Values . . 170 Applied Group Analysis: Rank Order of Evaluative Criteria . . . . . . . . 170 Non-Applied Group Analysis: Rank Order of Evaluative Criteria . . . . . . . . 171 Selected College Characteristics Comparison 172 Hypotheses and Conclusions . . . . . . . 172 Recommendations for Future Studies . . . . 176 APPENDIX . . . . . . . . . . . . . . . . 178 BIBLIOGRAPHY O O O O O O O O O O O O O O 222 vii Table 1. 10. 11. 12. LIST OF TABLES Page Estimated Tuition and Fees, Room and Board Rates in Institutions of Higher Education, by Type of Control of Institutions: United States, 1962-63 to 1972-73 . . . . . . . . . 41 Application State Comparison of Evaluative Criteria Importance . . . . . . . . . 90 Summary: Difference in Evaluative Criteria Importance by Rank Order Tested Across the Applied and Non-Applied Classification . . 92 Comparative Dependence of Matched Conditions and Application State for Evaluative Criteria Descriptors . . . . . . . . . . . 94 Ratio of Brothers and Sisters with College Experience to Student Sample (n) . . . . 104 Percentage Distribution of Parent's Estimated Income Before Taxes, 1973 . . . . . . . 106 Differences in Socioeconomic Variables Within Groups Between the Applied and Non-Applied Classification . . . . . . . . . . 108 College Informed Classification, Before Senior Year of High School: Combined Group . . . 112 Rank Order and Mean Values of the Importance of Information Sources for Goods Buying DeCiSions O O O O O O O I O I O O 114 Number of Colleges Applied, Frequency Distri- bution: Combined Group . . . . . . . 115 Time the Decision to Attend College was Made: Frequency and Percentage Distribution . . . 117 Reasons Given for Being Most Informed About One College, Before Senior Year of High School: Combined Group . . . . . . . . . . 120 viii Table 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. Degree of Usefulness of College Information Sources: ACT Group . . . . . . . . . Degree of Usefulness of College Information Sources: SAT Group . . . . . . . . . Differences in Selected Variables Within Groups Between the Matched/Not Matched Classification . . . . . . . . . . Post-Enrollment (Follow-up) Response Distribution . . . . . . . . . . ‘. Evaluative Criteria: Degree Correlated t2 - t3 0 O O O O O O O O O O O 0 Significance of Within Group Homogeneity in Rank Order of Evaluative Criteria Based on the Coefficient of Concordance W, Using a Chi-Square Test: Enrolled and Not Enrolled, Within the Applied Group, t2 and t3 . . . Significance of Within Group Homogeneity in Rank Order of Evaluative Criteria Based on the Coefficient of Concordance W, using a Chi-Square Test: Private and Public, Within the Non-Applied Group, t2 and t3 . . . . Rank Order of Importance of the Evaluative Criteria Based on the Coefficient of Concordance W Analysis: Enrolled and Not Enrolled, Within the Applied Group, t2 and t3 . . . . . . . . . . . . . Rank Order of Importance of the Evaluative Criteria Based on the Coefficient of Concordance W Analysis: ACT and SAT, Enrolled and Not Enrolled, Within the Applied Group, t2 . . . . . . . . . Rank Order of Importance of the Evaluative Criteria Based on the Coefficient of Concordance W Analysis: ACT and SAT, Enrolled and Not Enrolled, Within the Applied Group, t3 . . . . . . . . . Rank Order of Importance of the Evaluative Criteria Based on the Coefficient of ix Page 122 123 128 130 135 148 149 150 151 153 Table 24. 25. A-s. A-6. A-7. A-8. A-9. A-lO. A-11. A-12. Concordance W Analysis: Private and Public, Within the Non-Applied Group, t2 and t3 . Rank Order of Importance of the Evaluative Criteria Based on the Coefficient of Concordance W Analysis: ACT and SAT, Private and Public, Within the Non-Applied Group, t2 . . . . . . . . . . . Rank Order of Importance of the Evaluative Criteria Based on the Coefficient of Concordance W Analysis: ACT and SAT, Private and Public, Within the Non-Applied Group, t3 . . . . . . . . . . . Frequency of Match of the Descriptors and the Evaluative Criteria . . . . . . . . Frequency of Match of the Descriptors and the Evaluative Criteria by High and Low Range Sum of Rank Order Values and Mean Values of the Evaluative Criteria Variables . . . Type of College Variable: Frequency by Rank Order Without Regard to the Match or No Match Condition . . . . . . . . . Type of College Variable: Frequency of Match Condition and Rank Order Values . . . . Type of College Variable: Frequency of Match contition O O O O O I O I O O 0 College Choice Preference Rating: lst, 2nd, and 3rd or Below . . . . . . . . . College Choice Preference Rating: lst, 2nd or Below 0 C I I O O O C O I 0 College Choice Preference Rating: lst and 2nd . C C O C C O C O C O . . College Choice Preference Rating: 2nd and 3rd College Choice Preference Rating for Students Matched on the Type of College . . . . Individual Parent's Educational Classification: Combined Group . . . . . . . . . X Page 154 155 156 180 180 181 182 182 183 183 184 184 185 185 186 Table A-13. A-14 o A-ls o A-16. A-17 O A-18. A-lg c A-20. A-21. A-22. A-230 A-24. A-25. A-26 o A-27. Page Individual Parent's Educational Classification: ACT Group . . . . . . . . . . . . 186 Individual Parent's Educational Classification: SAT Group 0 O O O O O O O O O O O 1 8 7 Brother's and Sister's Educational Classification: Combined Group . . . . . 187 Brother's and Sister's Educational Classification: ACT Group . . . . . . 188 Estimated Value of Homes Within the Neighbor- hood of Residence (excluding rural and farm) combined Group C O O O O O O O O 0 18 8 Respondent Classification by the Number of Moves (last seven years): Combined Group . 189 Vacation Trip Companion Preference: Applied Group 0 O O O O O O O O O C O O 189 Vacation Trip Companion Preference: ACT Group 0 O O O O C O O O O O O O 190 Descriptive Accuracy Associated with Purchase Pattern Descriptor Statement "C": Combined Group C O O I O O O O O O O O O 190 College Informed Classification, Before Senior Year of High School: ACT Group . . . . . 191 Information Source Degree of Importance, Frequency Distribution for Sales People: Combined Group . . . . . . . . . . 191 Number of Colleges Visited Before Senior Year of High School, Frequency Distribution: Combined Group . . . . . . . . . . 192 Number of Colleges Applied, Frequency Distribution: ACT Group . . . . . . . 192 Time the Decision to Attend College was Made: Applied Group . . . . . . . . . . . 193 College Information Level Before Senior Year of High School Regarding "Social Opportuni- ties:" Combined Group Based Upon When They Decided to Attend College . . . . . . . 193 xi Table A-28. A-29 o A-300 A-3lo A-32 c A-33. A-34. A-35. A-36. A-37. A-38. A-39. A-400 Page College Information Level Before Senior Year of High School Regarding "Social Opportunities:" Combined Group . . . . . . . . . . 194 College Information Level Before Senior Year of High School Regarding "Fields of Study:" Combined Group . . . . . . . . . . 194 College Most Likely to Attend: Combined Group C O O O O O I O O O C O O 195 College Major Most Likely: Combined Group . . 195 College Major Most Likely: SAT Group . . . 196 Time of Decision To Go To College Across the Matched Condition of College Most Informed and College Most Likely to Attend: Combined Group . . . . . . . . . . . . . 196 Time of Decision To Go To College Across the Matched Condition of College Most Informed and College Most Likely to Attend: ACT and SAT Applied Group . . . . . . . . . 197 Number of Colleges Considered Across the Matched Condition of College Most Informed and College Most Likely to Attend: Combined Group . . . . . . . . . . . . . 197 Evaluative Criteria: Degree Correlated Across Time (ACT Group) . . . . . . . . . . 198 Evaluative Criteria: Degree Correlated (t1 - t3) Across the Applied and Non-Applied Groups (ACT Group) . . . . . . . . . 198 Evaluative Criteria: Degree Correlated (t2 - t3) Across Matched Condition (Combined Group) . . . . . . . . . . . . . 199 Evaluative Criteria: Degree Correlated (t2 - t3) Across College Attendance Decision Classification (Combined Group) . . . . . 199 Evaluative Criteria: Degree Correlated (t - t3) and Matched Condition Across the ColIege Attendance Decision Classification (Combined Group) . . . . . . . . . . . . . 200 xii Table A-41. A-42. A-43. A-44 c A-45. A-46. A-47o A-48. A-49. A-50. A-51. A-52. A-53 O Evaluative Criteria: Degree Correlated(t2 - t3) and the Number of Applications (Combined Group) . . . . . . . . . . . . . Evaluative Criteria: Degree Correlated (t2 - t3) and the Number of Colleges Considered (Combined Group) . . . . . . . . . . Evaluative Criteria: Degree Correlated (t2 - t3) Across High and Low Test Scores (Combined Group ) o o o o o o o o o o o o o Evaluative Criteria: Degree Correlated (t1 - t3) Across the Private and Public College Classification (ACT Group) . . . . . . Cost Criterion (t2): Across the Applied and Non-Applied Classification (Combined Group) . Type Criterion (t2): Across the Enrolled and Not Enrolled Classification (Applied Group) . Field Criterion (t2): Across the Enrolled and Not Enrolled Classification (Applied Group) . Extracurricular Criterion (t2): Across the Enrolled and Not Enrolled Classification (Applied Group) . . . . . . . . . . Location Criterion (t3): Across the Applied and Non-Applied Classification (Combined Group) . . . . . . . . . . . . . Size Criterion (t3): Across the Private and Public College Classification (Combined ' Group) . . . . . . . . . . . . . Cost Criterion (t3): Across the Private and Public College Classification (Combined Group ) o a o o o o o o o o o o 0 Student Body Criterion (t3): Across the Enrolled and Not Enrolled Classification (Applied Group) . . . . . . . . . . Size of College Actually Enrolled (Combined Group) . . . . . . . . . . . . . xiii Page 200 201 201 202 202 203 203 204 204 205 205 206 206 Table Page A-54. Type of College Actually Enrolled (Combined Group) 0 o o o o I o o o o o o o 207 A-55. Cost of College Actually Enrolled (Combined Group) 0 o o o o a o o o o o o o 207 xiv LIST OF FIGURES Figure Page 1. VIADE Concept: College Information and Decision Pattern O O O O O O O O O O O O O 75 2. Longitudinal Time Reference-Student Action Pattern O O O O O C I I O O O I O 77 3. Data Sought with Data Collection Form and Time Reference 0 O O O O O O O O O O I O 84 4. Purchase Pattern Matrix . . . . . . . . . 109 XV LIST OF APPENDICES Appendix Page A. Chi-Square Tables . . . . . . . . . . . 179 B. Questionnaires and Letters . . . . . . . . 208 xvi CHAPTER I INTRODUCTION Problem Background Student college choice (buying) can be viewed as a decision process with: 1. educational cost outlay implications, 2. institutional choice implications, 3. vocational and other future pay out implications accruing from the total product acquisition, 4. acquisition process cost and benefits associated with the educational product, and 5. opportunity cost implications. The potential college student is concerned with making choice decisions from a set of available and known alterna- tives. This set of alternatives, however, is expandable; determination of the breadth of this choice range would ap- pear to be a function of prior information, either solicited or unsolicited. Conditions of social, cultural, family, and peer group exposure; academic ability; and economic means; plus other factors, interact to provide the basis of moti- vation and knowledge associated with the college choice decision. This suggests the need to view college choice as a purchase choice of personal and social significance, and one which results in a major cost; including time, money, effort, and foregone opportunities. Potential college student choice behavior can be viewed in a consumer buying behavior context. This view is not the common view of edu- cators or that of the student (and others such as parents) when the college choice is made. Since World War II and through the 19605, the in- creased demand for a college education served to push for expanded expenditures and facilities in the educational sector, both private and public. The existence of a sellers market focused the attention of colleges on meeting the expanding demand, with less concern for efficiency and virtually no concern for generating selective demand for a particular institution. Selectivity was a matter of establishing entrance requirements to screen from a large number of potential customers (applicants) those who best fit the "image" or "mold" of the institution. Getting into a college was a major concern of many potential students during the early and mid 19603. The strategy of deliberate restriction of supply was not the case. Such a strategy would have been incon- sistent with the philosophy of educational opportunity for all those capable of utilizing it, and the recognition of social benefit accruing from a better educated populace. This public attitude, plus the growth pressure from within many educational institutions and educational systems, particularly public systems, resulted in the expansion of educational facilities. Capacity expansion in both the public and private education sectors, by the late 19608, had brought supply more in line with demand, and for many private colleges space availability in both classrooms and dorms exceeded the demand. By 1970 the evidence of a buyers market, where supply in at least some sectors of higher education exceeded demand, was becoming all too evident. This buyers market trend has continued to the present, with not only private colleges and universities being affected, but also the public colleges and universities. Marketing Perspective of Educational Choice Which prospective students go where to college and for what reasons, in the aggregate, has been studied exten- sively by those in education. The student as the subject of study is nothing new. This research study is designed to apply a new perspective to the conceptualization of student college choice behavior and choice processes. The choice of going or not going to college, and the choice of which college to attend appear similar to the consumer choice problem of selecting an economic good or service. The latter problem situation is the focal point of traditional marketing. The rationale of both marketing and production effort is consumer satisfaction through consumption. This is basically what is meant by the phrase "consumer orientation," a basic tenent of the marketing concept. Is not the purchaser of an educational product with the associated experiences also a consumer, in the sense that monetary outlay is made for something in return--qgig pro 929? The difficulty of defining the educational product (or set of services) is a problem, but the same is true of products in the business sector when a broad view is taken of what a consumer gets for a monetary outlay. Unless we restrict the product definition to tangible or identifible elements generating certain levels of satisfaction, "product" remains a subjectively defined construct. Related Consumer Behavior Theory The Howard and Sheth model of buying behavior postu- 1ates that the buying process begins with the brand choice decision, given that the buyer is motivated to buy a product. The elements of his decision are (l) a set of motives, (2) alternative brands, and (3) choice criteria by which the motives are matched with alternatives.1 The alternative courses of action are the evaluations made of the various brands and their potential to satisfy the buyer's motives.2 1John A. Howard and Jagdish N. Sheth, The Theory of Buyer_Behavior (New York: John Wiley and Sons, Inc., 1969), p. 25. 2Ibid., p. 26. ll 1.] ' I I I l '1‘ I‘ I'll“ The brands that become alternatives to the buyer's choice decision are called the evoked set, and are generally few in number.3 Out of the total number of brands on the market the buyer may be aware of a small portion, and out of this small portion only a few are generally contained in his evoked set. The Engel, Kollat, and Blackwell model holds that the decision process begins with problem recognition and proceeds through four other stages: (1) internal search and alternative evaluation, (2) external search and alter- native evaluation, (3) purchasing processes, and (4) out- comes.4 The internal search is said to occur instantaneously and largely unconsciously. If the buyer has an adequate level of information and experience, well structured evalu- ative criteria, and established attitudes toward the products, an internal search is adequate for a buying decision. This search pattern is associated with habitual decision-process behavior.5 When the internal search proves inadequate for the evaluation of alternatives, an external search is required. 31bid., p. 26. 4James F. Engel, David T. Kollat, and Roger D. Blackwell, Consumer Behavior (2nd ed.; New York: Holt, Rinehart, and Winston, Inc., 1973). P. 439. 5 Ibid., p. 59. This may involve a search for additional information about the alternatives contained in the domain of feasible alterna- tives, but there is no need to procure information about the domain of feasible alternatives. This is referred to as limited decision-process behavior.6 Finally, the external search behavior which seeks information about the domain of feasible alternatives in order to define this domain is associated with extended problem solving.7 Both information about the domain of alternatives and information about the alternatives within the domain are sought and require a greater search effort. Howard and Sheth describe three decision making stages associated with the psychology of simplification in repetitive decision making. This is where the buyer attempts to reduce the complexity of a buying situation with the help of information and experience. The decision making stages are:8 (1) Extensive Problem Solving which refers to the early stages of repetitive decision making, in which the buyer has not yet developed well—defined and structured choice criteria. The buyer has no strong predispositions toward any of the brands he is considering as alternatives. 61bid., p. 59. 71bid., p. 59. 8Howard and Sheth, op. cit., p. 27. (2) Limited Problem Solving is the next stage, in which the choice criteria are well-defined and structured, but the buyer is undecided about which of a set of brands is best for him. The buyer has moderately high predispo- sitions toward a number of brands, but does not have very strong preference for any one brand. (3) Routinized Response Behavior is the last stage, in which the buyer not only has well-defined and structured choice criteria, but also strong predisposition toward one brand. At this stage, although the buyer may consider several brands as possible alternatives, he has, in fact, only one or two brands in mind as the most probable choice alterna- tives. The farther the buyer is along in simplifying his environment, the less is his tendency toward active search behavior.9 Purchasing an education can be viewed as an initial purchase decision when the student first enrolls, and a series of repetitive purchases each time he re-enrolls. The fact that most students do not switch colleges, but rather stay until they graduate suggests a routinized response behavior pattern. The focus of this study is, however, on the initial purchase. This appears to involve a more extensive problem solving type situation and requires an external search process. Individual differences will 91bid., p. 27. ll;ll|‘[.il{{ [ (l. exist based upon the prospective student's characteristics and prior environmental exposure. General Research Purposes Consistent with current theory of buying behavior and a market segmentation perspective, five general research purposes are set out for this study. 1. To gain knowledge and understanding about the decision process, structure, and evaluation-decision criteria used by prospective college students in choosing a specific college. 2. To determine the information needs, information sources, and information processing methods used as well as the level of information and its degree of specificity at various stages in the prospective student's college choice (buying) process. I 3. To explore the application of market segmentation analysis to identify student segments which are more likely to respond favorably to a college's market offering. 4. To contribute to the development of more effi- cient and effective methods of allocating student recruit- ment effort through a better understanding of the prospective student's buying behavior processes. 5. To contribute to the more effective and effi- cient planning of all educational marketing-mix elements based upon a better understanding of the prospective stu- dent's buying behavior and buying process needs. To accomplish these research purposes, the study encompassed the following aspects of student buying behavior across time. 1. Identify patterns of similarity or difference between prospective students who apply and those who do not apply to a specific college after they have indicated some initial degree of interest. 2. Identify patterns of similarity or difference between prospective students who are accepted by the college and enroll, and those who are accepted but do not enroll. 3. Identify patterns of similarity or difference between prospective students who select a private college rather than a public college. 4. Identify patterns of similarity or difference between prospective students across additional behavioral characteristics. Research Approach Statement A longitudinal design was used to explore and describe the buying behavior process of prospective college students, and to identify the associated independent vari- ables which serve to differentiate market segments and affect the prospect's response to a college's marketing effort. Major emphasis was placed upon the examination of search behavior and the evaluative criteria related to college buying intentions and final college choice behavior. 10 All of the respondents who were a part of the study had shown an initial interest in the c00perating college. This micro level institutional approach was considered appropriate in view of the unique set of characteristics associated with a specific college. Such uniqueness provides the basis for differentiation of the total educational prod- uct which the student experiences as a result of his or her college choice. Potential Contribution of the Research The study is expected to contribute to the recogni- tion that marketing, as a discipline, is applicable to other than commercial ventures. The approach taken in this study is to recognize the survival and growth objectives of colleges as motivational forces leading to competition in the pursuit of differential advantage. Marketing as a discipline is vitally concerned with any organization striving for survival and growth. Competition in the educational market between and within the public college sector and the private college sector, the environment of changing attitudes toward the value of a college education, the recent trend of declining birth rates, and the projected decline in the number of college age youth, all suggest that growth and survival may be difficult goals for colleges to attain in the future. This is a very real problem in higher education today, and 11 this research focuses on how a marketing approach can be applied by colleges to better solve this problem. It is hoped that this research will contribute to the development of a methodology which can be utilized by colleges in identi- fying and analyzing their various market segments, thus providing a better base for deve10ping improved marketing strategies. Specific contributions of this research to the field of marketing relate to buying behavior processes and the concept of market segmentation. l. The study provides empirical evidence illus- trating the more general application of buying behavior models to choice situations outside the traditional context of consumer goods purchasing. 2. The study provides empirical evidence of the role played by "weights" or "importance" measures associated with evaluative dimensions (criteria) in predicting "class" or "type” of product choice. 3. The study provides empirical evidence of the stability in the relative "weights" or "importance" associ- ated with a common set of evaluative dimensions for an individual buyer during the purchase process. The extent to which the buyer maintains consistency between his buying intentions, actual purchase behavior, and the relative ”importance" of the evaluative dimensions can be examined with reference to consistency theory. 12 4. The study provides empirical evidence concerning the transfer of previously developed consumer goods purchase patterns (as self-reported by respondents) to a new choice situation. Such a carryover would support learning as an important construct associated with generalized choice process behavior. In summary, the contribution made by this inter- disciplinary research is to examine empirically aspects of marketing theory, as they apply in a non-profit setting of higher education. CHAPTER II LITERATURE REVIEW Higher Education Market Demand The growth in American higher education during the decade of the 19508 can be characterized as explosive. Total enrollment in all institutions of higher education reporting to the U. S. Office of Education rose from 3.8 million in 1960 to 8.5 million in 1970. An additional 1.5 million students not included in the Office of Education statistics were enrolled in vocationally oriented schools of a largely proprietary nature.1 A major factor in the extraordinary growth in enroll- ments during the 19608 was the relatively high birth rate of World War II and the early postwar period. All told, the rise in college-age pOpulation accounted for about 45 per- cent of the increase in undergraduate degree-credit enroll- ment from 1960 to 1970. The remaining increase was attri- butable to a rise in the enrollment rate, i.e., the ratio of undergraduate degree-credit enrollment to the population aged 18-21--from 33.8 percent in 1960 to an estimated 47.5 _ 1A Report and Recommendations by the Carnegie Commission on Higher Education, New Students and New Places (New York: McGraw-Hill Book Company, 1971), p. 11. 13 14 percent in 1970.2 Increases in the enrollment rate in the other age categories are also reflected in the undergraduate enrollment totals. The U. S. bureau of the Census data show that between 1940 and 1970, the proportion of persons ages 18 to 21 en- rolled in college rose from 11 to 34 percent, and the Carnegie Commission staff projects an increase to about 54 percent by the year 2000.3 An additional factor revealed in the analysis of undergraduate enrollment from 1940 to 1970 is that the age range of undergraduates has widened and this is projected to continue. By the year 2000, 54 percent of the population aged 18 to 21 will be enrolled in degree-credit programs; while the undergraduate degree-credit enrollment as a per- centage of those 18 to 21 is projected for the year 2000 at 73 percent.4 Market Share and Institutional Change The founding of Harvard College in 1636 marked the beginning of the growth and development of American colleges and universities. The early years of educational develop- ment were dominated by private colleges, most of which were church related. In 1900 over 60 percent of all college 21bid., p. 11. 31bid., pp. 13-14. 4Ibid.. pp. 14-15. 15 students in the United States attended private colleges and universities. By 1960 the figure had fallen to 40 percent; and in 1970, only 25 percent of all college students were enrolled in private institutions.5 By 1972 the figure declined further to about 20 percent.6 The expansion in higher education was clearly in the public sector, and competition for students between the private and public sectors was developing. The fear that some private colleges would have to close their doors was realized during the early 19708. The trend toward public colleges is usually attri- buted to the large differences in tuition and other fees between the private and public institutions. Also, the growing importance of junior and community colleges, and other two-year institutions has affected private college enrollment. Between 1965 and 1970 the enrollment at public two- year schools doubled, accounting for almost 22 percent of total enrollment by 1970. In the same period, en- rollments at public and private four-year institutions grew by 42 and 8 percent respectively.7 The number of two-year institutions increased from 622 in 1963 to 1,061 in 1970. Two-year institutions accounted 51bid., p. 17. 6Fred M. Hechinger, "Is Common Action Possible?" Change, September 1972, p. 41. 7Richard R. Spies, The Future_of Privgte Colleges (Princeton: Princeton University, 1973), P. 5. 16 for 38 percent of the 2,827 institutions reporting to the Office of Education in 1970 and for 28 percent of all stu- dents in 1970.8 The Carnegie Commission on Higher Education report, New Students and New Places looks to the future and projects the following trend. The next three decades are likely to be a period of substantial innovation and change in the organization and structure of higher education comparable in signifi- cance to two earlier periods of change. The first was the period following the Civil war when many of the leading colleges were transformed into universities. The second was the period since the end of World War II, which was characterized not only by rapid enroll- ment increases and a steady increase in the share of the public institutions in total enrollment, but also by the emergence of planned state systems of public higher education and of the public two-year community college as the most rapidly growing type of institution. Along with the continuation of recent trends, we anticipate a new type of development as perhaps the predominant characteristic of the last three decades of the present century--a movement away from partici- pation in formal institutional higher education in the years immediately following high school toward a more free-flowing pattern of participation spread over a broader span of years, perhaps well into middle age and beyond.9 Private and Public Sector Develgpment During the expansionist period of the 19508 and 19608 the availability of students allowed the private and the public colleges to drop their rivalry. The 19708, so 8A Report and Recommendations by The Carnegie Com- mission on Higher Education, New Students and New Places, _p. cit., pp. 18-21. 91bid., p. 39. 17 .far, has seen a rebirth of competiveness between the public and private sectors of higher education and is attributable to a number of developments: 1. 2. The total higher education enterprise has, at least for the short run, overestimated the extent and the duration of the enrollment boom. The recession and the aftermath of the college rebellion have reduced the amount of money available for the support of higher education. The reduction of Federal research funds left many universities with costly facilities and over—expanded staff. The poor and disadvantaged represent the only major population sector that could account for further enrollment growth. These students require financial support instead of bringing money to the campuses. Neither the public or private sector stands to benefit immediately from this flux.10 The educational pie has shrunk, and the public and private sectors are once again fighting for the slices of the pie. half a million reported classroom vacancies. Before the 1972-73 academic year, there were about 11 Early Development 'Higher education in the United States started largely as a private enterprise, and a deliberately elitist one. Despite early dreams by Madison and Washington of a truly national university, it was Harvard that established the prototype. Although there were exceptions to the rule--the University of Virginia and the City College of New York among them-- the public sector came igto its own only after the Land-Grant Act of 1861.1 10Hechinger, gp. cit., p. 38. llIbid., p. 39. lzIbid., p. 39. 18 Protagonists of public colleges wanted colleges for the people, and those colleges were expected to serve practical ends; study of agriculture and of the mechanic arts should be honored equally with study of the classics, or even take precedence over these traditionally elitist studies.1 The demands for the establishment of the land-grant colleges (Morrill Act) were political demands voiced by public leaders on behalf of their constituents, yet the private demands for places in colleges were not that great. One of the main problems of the new land-grant colleges (as has long been true of private colleges) was to find students, and often that could be done only by first building up a more adequate system of secondary schools in the state. Attracting students to the institutions of higher education is nothing new. Both the private and the public institutions have faced this problem before. The strong demand of the 19608 was the exception, not the historic norm. Current efforts to attract prospective college stu- dents reflect a weak demand situation, but not a completely new situation. Educational Competition Favored Competition between the public and private sectors of higher education produces favorable results by encouraging more students to continue their education and by offering l3C. Arnold Anderson, Mary Jean Bowman, and Vincent Tinto, Where Colleges_Are and Who Attends (New York: McGraw- Hill Book Company, 1973), p. 3. 14Ibid., p. 3. 19 prospective students a choice among alternate educational opportunities (product, brand, etc.). It is desirable to maintain a strong private higher education sector simply because monopolies are intrinsi- cally undesirable. Public systems, whether municipal or statewide, are subject to acroSSethe-board rulings and policy changes from which there is no escape. Even the best of systems have a homogenizing effect. The pressures for standardization on the public universities, moreover, would become much harder to resist if the counterweight of private competition were eliminated. With some slight poetic license, it may even be argued that private higher education before the 1860's enjoyed something of a monopoly and that it would probably have failed to meet the changing demands of a changing nation without the growing competition from the emerging state univeristies. Both the public and the private sectors of higher education are public in their mission. However, the private institutions do have the opportunity to be far more selective and purposeful about their academic mission, disciplines, and services offered. The product offering can be tailored to meet the target market group needs and satisfy the goals of the institution, but both must be well defined prior to developing a strategy. Nature and Structure of Competition While it is convenient to refer to private education as separate and distinct from public education, the distinc- tion is not that clear. There is a considerable amount of variation within the private institutional grouping. Private institutions range from small, church related schools that 15Hechinger, 22: cit., p. 42. 20 depend almost entirely on tuition income, to universities such as Columbia with more than half its budget derived from public funds, and engaging in a variety of public service activities.16 The latter are quasi-public insti- tutions. Alexander Astin and Calvin B. T. Lee make the point that all private colleges are not alike in their study, The Invisible College. Using an index of institutional visi- bility composed of (l) enrollment size and (2) undergraduate selectivity, they distinguished two groups of private col- leges, "invisible" colleges, and the ”elite" college8.l7 The basis of measuring selectivity was the SAT scores, both verbal and mathematics; and the ACT composite score. Col- leges were then classified by selectivity from 1, least selective; to 8, most selective. The colleges were classi- fied also by size from 1 to 8. The most visible colleges were those with the highest degree of selectivity and largest enrollments.18 Well over half of those institutions (524 of 918) have selectivity scores below level 4 (combined SAT Verbal plus Mathematical scores of less than 1,000). If one eliminates from this group of 524 the 30 with enrollments of 2,500 or more, the remaining 494 16"The Crisis of Money and Identity," Change, September 1972, p. 36. 17Alexander W. Astin and Calvin B. T. Lee, The Invisible College (New York: McGraw-Hill Book Company, 1972), pp. 3-40 18 Ibid., p. 4. 21 invisible colleges still represent more than half of all the private four-year colleges in the country, one- third of all institutions offering at least a bachelor's degree, and about 21.5 percent of all institutions of higher learning in this country. They also enroll an estimated 500,000 students, or 15 percent of all stu- dents attending four-year institutions.19 At the other end of the visibility continuum, we find that there are only 44 colleges in the top two levels of selectivity (combined SAT Verbal and Mathe- matical scores above 1235). Although these 44 "elite" private colleges account for nearly two-thirds of all higher educational institutions on the two highest selectivity levels . . . , they represent less than 5 percent of the four-year private colleges. . . . The major finding of this study was, with respect to their student inputs and their environment, invisible col- leges are much more similar to the public colleges than to elite colleges.21 The invisible college and the elite college--except for being privately controlled and rather sma11--have very little in common. The two types of private colleges serve radically different student clienteles, and their social and intellectual environments are highly dis- similar. By the same token, the public college, except for its larger size, closely resembles the invisible college both in its environmental characteristics and in the students that it attempts to serve.22 19Ibid., p. 10. 20Ibid., p. 10. 211bid., p. 79. 221bid.. p. 79. 22 While the invisible and state four-year colleges appear to be appealing to the same student market segment, there is a wide diversity in the tuition and fees of the two institutional types, and the relation of these fees to the cost of educating a student. The tuition at private colleges has been rising at about 7.5 percent per year; and at best it pays for about 75 percent, but usually closer to 50 percent of the costs of educating a student.23 Among the fastest rising costs for independent schools has been recruitment costs, now estimated at as much as $500 per student on the average, compared to $250 in 1967.24 Included in these higher costs are higher recruiter salaries and expenses, and the followvup activity-- for example phoning, brochures, and entertaining of visitors; the cost of fund raising has climbed similarly.25 Is the difference between the public and private college educational experience worth the higher charges paid by the buyer of the private education? If the product is not different, then the fee difference is hard to justify. Paul C. Reinert S. J., comments that the lack of funds at private colleges leads to an economic homogenization process 23Paul C. Reinert S. J., To Turn The Tide (New York: Prentice-Hall, 1972), p. 20. 24 Ibid., p. 20. 25Ibid., p. 20. 23 that dissolves the uniqueness and reduces the diversity potential of the private college.26 If the private sector of higher education is to serve as an alternative to the public sector, in a plural- istic system, it must first survive. The loss of product uniqueness endangers such a survival and offers a weak platform on which to build a marketing strategy. Yet, many private colleges are engaging in a more extensive use of marketing technology in an attempt to survive. Change or die is the command of competition, even in education. The invisible colleges are in a constant state of flux, and their turmoil is not simply a matter of minor revision in curriculum or internal governance but of fundamental change, change that relates to their whole raison d'etre. Unlike the elite colleges, their ability to survive has always been in question. It is difficult to live from day to day in such doubt. The changes that they have undergone reflect their desire to survive and indeed their ability to change in order to survive. The primary concern of all the private colleges both sectarian and independent was, and still is, survival. In view of the competition, and the trend in the United States toward the lower (or even free) tuition in state-supported higher education, survival is becoming more difficult. Can the use of marketing technology and the appli- cation of the marketing philosophy of consumer orientation 261bid., p. 27. 27Astin and Lee, 22- cit., p. 23. 24 and long term profit (income to cost relationship required for survival) be applied to this area? The present study is designed to examine one basic aspect of this question, i.e., to develop a better understanding of a college's perspective on student buying behavior patterns. This fundamental understanding is considered necessary to the development of sound marketing plans by a college. To know the market and its various segments, and the buying behavior pattern of the prospective customer set would allow more effective and efficient planning for differential advantage by the college. The concept of differential advantage and its implication for survival in a competitive environment, as expressed by Wroe Alderson (1957), seems applicable to higher education today. Philosophies About Who Should Go To College The extent and nature of the prospective student market is partially defined by the phi1080phy of who should go to college. Two elements, (1) the opportunity to go to college, and (2) the willingness or desire to go to college, are involved in the definition of the college student market. The first element is affected by the individual's ability, the attitude of society, and the amount of political influ- ence exerted to make educational opportunity available to selected publics. The second element, willingness or desire, 25 is dependent upon the individual as influenced by many factors, both internal and external. K. Patricia Cross identifies three major philosophies related to who should go to college. 1. The Aristocratic philosophy--where students with money and family social status, with or without adequate ability were able to go to college while others were not. These were high-tuition private colleges. The educational system was a closed system. 2. The Meritocracy philosophy--criteria for college admission should be based upon scholastic ability and the willingness to study hard--i.e. upon academic merit. This philoSOphy became evident with the land grant college movement and remains evident today. The talent searches of the 19508 were active campaigns to bring into colleges those who did not meet aristocratic criteria but who were the epitome of meritocratic ideals, i.e. high aptitude test scores. The barriers of the aristocratic period gave way to new barriers of the meritocracy. 3. The Egalitarian philosophy--those who want to go to college should be allowed to go, not just those who desire and have the ability as traditionally measured.28 "Young people who have not considered college in the past but who are newly entering college in the 19708 are distinguished more by low test scores than by any other single measure available, including race, sex, and socio- 29 economic status." These "New Students" are part of a new growth market segment, but with a different set of academic 28K. Patricia Cross, New Students and New Needs in Higher Education (Berkeley: Center fOr Research and Develop- ment in Higher Education, University of California, 1972), pp. 1-5. 291bid., p. 25. 26 needs. Traditional educational programs and teaching methods are not suited to the needs of these students. This market segment represents a currently unfulfilled need which may offer market potential to some colleges if they can develop and implement the correct programs. The cost implications of such a venture would have to be considered against the revenue opportunity for a private college. A market segment cost/revenue analysis would appear to be an appropriate approach which a college could use in assessing such an opportunity. Educational market demand currently differs for men and women when the ability factor is used as a dimension of market segmentation. The largest increases in college attendance for women are now coming from the ranks of the above- average students from all socioeconomic levels as women continue toward the peak of the meritocratic era in college attendance. For men, the meritocratic phase has passed its peak, and in the decade of the 19708 the major increases in college attendance will come from the lower—ability men as the egalitarian phase is entered. . . . for men, at least, low academic ability is keeping more students from continuing their education than is the barrier of lack of financial resources.31 College Admission Trends A study was done on admission trends32 for three groups of colleges; (1) Private I, which includes the most 3°Ibid., p. 17. 311bid., p. 18. 32Spies, gp. cit., pp. 5-17. 27 prestigious and selective private colleges and universities, (2) Private II, which includes roughly the same student charges as group one but are generally less prestigious academically, and (3) a group of the best state universi- ties. The application data used were for 1967 through 1971 academic years. ”In general, the evidence seems to support the hypothesis that there has been no significant trend toward applying to the top state universities rather than the select private colleges and universities."33 This conclusion was drawn from the application pattern of men. For women a somewhat different pattern exists. "The most obvious difference is the relative absence of growth at the most prestigious private schools, where the number of applications has barely changed over the last five years."34 Many of the women's colleges have felt the pressure of the increasing number of women wanting to go to coeducational rather than women's schools. Decisions of schools like Yale and Princeton to become coeducational has altered the structure of competition. Another difference in the application pattern for women is the rise in 1971 of in-state applications to state universities. In part, this may reflect the greater social 33Ibid., p. 11. 34Ibid., p. 11. 'I'i'lll‘" 28 activism of women and their desire to share more equally the benefits of our society; or it may be that the choice of a college by women is more affected by purely economic 35 factors. If the latter is true, they are more likely to react to the rising cost differentials between private schools and their own state universities than are the men. The yield on admission pattern (the percentage of those admitted to a particular school who actually enroll) showed a general downward trend for 1967 through 1971 for 36 all types schools. The downward trend, however, is particularly noticeable for private colleges. There are two possible interpretations. First, private schools in general may be losing students to the state universities and other public institutions. The implications of such a trend, if it exists, for the future of private higher education are fairly obvious. Second, private schools may be competing more and more among themselves for the same group of students. Although such competition is much less serious in terms of what it implies about the future of private higher education, it clearly poses a threat to many individual institutions. Unless the pool of qualified applicants grows more rapidly than it has in the last five years, an attempt by any one of these select private colleges to improve the quality of its student body or to expand in size must be accomplished largely at the expense of the other schools in this category. . . . it appears that the problem faced by select private colleges and universities in attracting qualified students have been somewhat exaggerated. Over the last five years, the number of applications 351bid., p. 11. 361bid., p. 14. 37Ibid., p. 15. 29 followed the same general pattern at private and public institutions. The select private colleges have experi- enced a drop in yield, but the proportion of students declining admission who go to public schools has re- mained quite stable. Although individual private insti- tutions are facing increased competition, it does not appear that the private colleges as a group have declined in popularity. The most serious problem seems to be a virtual halt in growth of the pool of qualified appli- cants. The contrast between the selective private colleges (elite group) and the visible private colleges, as described by Astin, present different competitive environments. The most evident difference is associated with the pool of prospective students. Academic ability appears as the major criterion defining the two markets. For either the elite or the invisible college the quantity of prospective students from which to draw is limited. However, the elite college does have the option of drawing (accepting) from the lower ability group. The invisible college has far less Oppor- tunity to tap the high ability group, and it is already drawing from the lower ability group. The invisible col- lege may look to the even less academically qualified (the "new students" as described by Cross) who have previously not considered college. Any specific college is unique in what it offers potential students. To define the market for a particular college and to develop a strategy to attract its clientele, the college must first understand the buying behavior and 381bid., p. 17. 30 characteristics of the prospective students. Academic ability is a convenient criterion to use in defining student market segments, but many other factors are involved in the individual choice of a college. Factors which are not academically related may be of even greater importance in making the actual buying decision. These other variables then are important in defining segments of potential stu- dents. College Choice Factors We may postulate three dimensions of "accessibility" to a particular college with which a prospective student will be concerned. The first is geographical accessibility. This dimension is the geographical distance between the student's residence and the college he attends. Considerable research has been done on this dimension, including the migration pattern studies from state-to—state, and the studies con- sidering the impact of local colleges upon the rate of college attendance among high school graduates. The recent growth in junior colleges and community colleges has re- flected a generally held attitude that convenience of location will stimulate college attendance as well as reduce the cost of attendance. The positive influence of location on the rate of college attendance among high school graduates has recently been refuted by the research of Anderson, Bowman and Tinto (1973). 31 The second dimension of accessibility, the ability of the prospective student, affects the college attendance pattern two ways: (1) some students lack the necessary ability to gain admittance at particular schools and are rejected when they apply and (2) some qualified students lack the motivation to attend particular colleges which are perceived or known to have high level academic requirements. The third dimension is that of price or cost to the student associated with attending particular schools. Here the price difference of public and private colleges; and the price to in-state versus out-of-state students, when attending state schools, is important. These three dimensions of accessibility are only part of the total set of factors or variables which may affect a particular college choice. Further, these three dimensions are not solely independent, but may in combination serve to define and limit the range of alternative choices available to the prospective student. The preference for a particular major or field of study is another extremely important variable which must be considered in the choice process. This is not considered here as a dimension of accessibility. The major or field of study variable, however, may affect or be affected by ‘the dimensions of accessibility. It seems quite likely that a state of conflict could result from the two variable categories. It may be impossible, for instance, for a 32 particular student to attend a local college and get the major he wants. This conflict could be resolved only through a choice involving some compromise. In this section the current literature associated with these accessibility variables, and other variables associated with the college decision will be reviewed. Geographic Dimension of Accessibility Geographic accessibility has been of interest to educational researchers as they study the effect of location upon the rate of college attendance in the aggregate; and across ability categories, social status groups, ethnic or cultural groups, and states or regions. we might expect that geographical accessibility to a college will affect an individual's college decision in .one or more of three main ways: (1) through relationships between immediate geographic access and cost of attending, (2) through effects on preference attitudes, and (3) through diffusion of information or intensity of communication. One recent study, Where Colleges Are and Who Attends,39 addresses the problem of examining the effects of college accessibility (geographical) upon attendance. The accessi- bility approach used looks out to the range of college options available at various distances from the community of residence for a given set of high school graduates. (An alternative 39Anderson, Bowman, and Tinto, gp. cit., pp. 1-293. 33 approach would be the market-area or recruitment approach which identifies the catchment area from which new enrollees in a given college come.) The authors utilized three models in the study.40 Model 1, was a simplified econdmic model of the.stu- dent decisibn maker as an investor in education; This model included the costs of attending each of a number of col- leges (local or nonlocal), the ability to pay, any non- monetary constraints that limit access to some colleges, and the future benefits from choosing one college over another. This model, however, did not consider tastes or adequacy of information. Models 2 and 3, introduced tastes or preferences, and the diffusion of information. In model 2, the variable of tastes or preferences was added to allow a comparison of tastes against the characteristics of specific colleges. The taste variable was treated as an exogenous variable. In model 3, allowance was made for limitations in knowledge among high school graduates about educational options and any possible effects of college location upon subsequent attitudes and tastes for continuing higher edu- cation. These variables (tastes and information) were stipulated as endogenous intermediate variables, thus model 3 was dynamic in nature, while models 1 and 2 were not. 4oIbid., pp. 6-14. 34 While model 3 was the most realistic one, data used in the study were not of a longitudinal type, therefore on 41 going relationships could not be tested. Nevertheless, the 'recognition of a need for such a model to study the influence of college location on the student's decision process was important. The following conclusions of interest were drawn in the study: 1. . . . spatial accessibility to one or more colleges has little effect, for most youth, on whether they will attend college-~be the accessible school a junior college, an open-door four-year college, or a more selective institution. . . . the correlation between a youth's ability and the type of postsecondary activity he chooses (including the type of college attended) is only moderate, and the ability distributions vary less by type of college (though not between pairs of colleges) than most persons would assume. . . . family status and personal ability outweigh accessibility (geographic) in explaining variations in college attendance rates, despite large over- lapping in the ability distributions for college and noncollege youth. Despite many irregularities, the data do indicate that individuals tend to choose the nearer option in attending college, but that this preference is usually weak and for some sets of youths may even be reversed. . . . we find, that the more able youth from the economically most advantaged homes will be the most likely to go to college, not only at a distance but in another state. The much-desired expansion of attendance by able youth from low-status families cannot dependably 411bid., p. 279. 35 be increased through the implanting of colleges closer at hand. 7. Propensities to attend college are spread by many influences, but college proximity is among the least influential factors bringing about the diffusion of college going among members of a community. Both low-cost tuition and the elimination of ability constraints on entry are more relevant than school location to those youth who are at the decision margins. 8. . . . it is important to specify generalizations for an interlocked set of cells characterized by types of schools, by types of communities from which students go to college, by types of colleges to which they go as enrollees, and by characteristics of youth who enter college and those who do not. That statement actually is the most general finding of the study.42 Geographical proximity, as indicated in this study, may not cause a higher attendance rate among high school graduates in a community, but this does not mean location is an unimportant variable in choosing a particular college. Several studies have found the location of a college the second most important reason given by students and their parents for their college choice.43 421bid., pp. 268-288. 43Charles Abbott, "An Investigation of the College Environment Perceptions of Prospective College Freshmen and Their Relationship to the Choice of a College or University" (unpublished Ed. D. dissertation, Michigan State University, 1967), p. 67; Thomas A. Bowers and Richard C. Pugh, "A Comparison of Factors Underlying College Choice by Students and Parents," American EdugetionaliResearch Association Paper and Symposia Abstracts, 1972, p. 97; and Robert V. Hanle, “Freshman College Selection Evaluation," Institutignal Research and Communication in Higher Education, 1970, p. 128. 36 Student Migration.--Since 1938 there has been a steady increase in the absolute number of students attending colleges and universities outside their home states, but the percentage of all students attending institutions outside their home state has been decreasing. In 1968, 16.8 percent of 6,545,363 students enrolled in colleges and universities were reported as out-of—state students.44 The pattern of migration differs for public and private schools. Publicly controlled institutions of higher edu- cation show a steadily decreasing proportion of migrant students; while the private institutions show a steadily increasing proportion, from 28.1 percent of the students enrolled in 1949 to 34.8 percent of those enrolled in 1968.45 The net migration (those entering less those leaving the state to go to college) pattern indicates that certain states are major exporters of college students. Since 1938, the states of New Jersey, New York, Illinois, and Connecticut have remained major exporters of college students; while the District of Columbia, Massachusetts, Indiana, North Carolina, and Tennessee have remained major importers of college stu- dents.46 44Thomas E. Steahr and Calvin F. Schmid, "College Student Migration in the United States," The Journal of Higher Education, Vol. 43} NO. 6.(June 1972), pp. 4444445. 45 Ibid., p. 445. 461bid., p. 450. 37 When a student becomes a migrant there is presumably a decision-making process that is completed prior to his physical relocation. The complex nature of this process is partially documented by existing research on migration and mobility in general, but very little study has been done specifically on college students. For example, it might be expected that a student coming from a highly mobile family would tend to be less geographically constrained in his choice of a college than one coming from a less mobile family. Ability Dimension of Accessibility Higher educational institutions may range from very selective units with rigid entrance requirements to open- door units with no ability screening requirements. For the prospective student, academic ability serves as a factor which influences both his decision to go to college and his choice of a particular college. Using "rank in high school class" as an ability measure, 41.8 percent of the high school graduates entering college in the fall of 1971 were in the upper quarter; while 4.0 percent were in the lowest quarter.47 The ability factor appears to influence college attendance of males differently than it does females. In 1971, 67.1 percent of the males entering college as freshmen 47National Center for Educational Statistics, Digest of Educational Statistics, 1972 ed., Table 91, p. 78. 38 were in the upper 50 percent of their high school classes; while 80.1 percent of the females were in the upper 50 per- cent of their high school classes.48 Cross suggests fear of failure as the explanation for low aspirations (to go to college) on the part of low ability students. Based on SCOPE data collected from high school seniors, students scoring in the lowest third on a test of academic ability were more than twice as likely to want to avoid the possible failure situation of being re- jected by a college of their choice, as students scoring on the top third.49 If these analyses are correct, we would predict that low-achieving fear-threatened high school seniors would apply either to open-door community colleges or to highly selective colleges. They would be sure of ac- ceptance at the open-door colleges, and to be turned down by Harvard is not really very threatening to the student who has no expectation of going there.50 The moderately selective colleges are the ones that prove threatening, thus prospective students who seek to avoid failure will avoid applying to these institutions. Access to a college is not limited by the college's decision alone, but also by the personality of the prospective student. Other research findings have indicated that most students tend to apply mainly to schools that are similar, 481bid., p. 78. 49Cross, 92' cit., pp. 38-39. 5°Ibid., p. 39. I ll ‘ I I I‘ll. I 39 particularly in terms of cost and quality. Richard R. Spies tested the effects of the quality of schools on applications. Classifying sets of schools according to their median SAT scores he found that: . . . an increase in the quality of the school (or its median SAT) will result in a higher proportion of applications from all those students whose SAT scores are above that level. For students more than 175 points below the median, the probability of their applying will fall51 In effect, the school would become too good for them. Applicants, then, attempt to find schools which are commensurate with their academic ability. Academic achievement and socioeconomic status have been considered as interacting variables affecting college attendance. Project Talent, American Institutes of Research, 1966, provided evidence that as the combined variables decreased, so did the probability of college attendance. For the high achievement, high socioeconomic status quartile the probability of college attendance was .92 for males and .87 for females; while for the low achievement, low socio- economic status quartile the probability of college attend- ance was .10 for males and .08 for females. The high socio- economic status, low achievement quartile showed .38 male and .37 female probability of college attendance; while the low socio-economic status, high achievement quartile showed .61 male and .42 female probability of college attendance.52 51Spies, gp. cit., p. 36. 52Seymour E. Harris, A Statistical Portrait of Higher Education (New York: McGraw-Hill Book Company, 1972), p. 61. 4O Ability appears to be a more significant qualifying factor of accessibility than does socioeconomic status. The admissions standards of a college, however, serve to deter- mine this degree of accessibility. The lack of income, as an element of socioeconomic status, can be altered by financial aid, thus improving the accessibility of a college. This strategy of financial aid seems evident from the above pattern of probabilities of college attendance, and is consistent with the "meritocracy" philosophy of higher education. The findings of one study indicated that while curriculum, faculty reputation, location, low costs, and university reputation were all important variables, finan- cial aid was the most important single variable influencing the choice of a university.53 Price Dimension of Accessibiligy How much is it going to cost to go to college? Price may be considered an important variable affecting the college choice of many prospective students. In both the public and private sectors of higher education the price or cost to the student has been increasing, as shown in data from the Office of Education Surveys of Higher Education. 53G. M. Naidu, "Marketing Strategies for Higher Education," Broadening the Concept of Marketing, ed. by David Sparks (Chicago: 'Am8rican Marketing AssoCiation,. 1970), p. 28. 41 .vma .m .moa manna .muma .omsmEH¢ sowumoscm pumpcmum .GOHHMUSUM Hmnmfim ca mucosaaoucm Hash msflsmmo can mmmumcu unwpsum oammm sowumoscm Hmnmwm mo m>m>nsm coaumoapm mo mnemmom .Hmmm anmpmom wuwusm mnu How was mumn .mnmaaoo omumnn081802 “wuoz mvmm momw meow mhvw mam.aw «em» mum>wnm «mew vqmw Nmmm mmvm mama mmmm Deansm mnlmnma mmnmwma mnlmnma molmmma mnlmhma mwlmmma coausuflumcH mcowusuflumsa Had msOHusuHumsH Had msOHusuwumsH Had no mEoom muouweuoo mmumm oumom , mwmm now mmmumnu pmuflsvmm paw coauwsa Houusou m.MhI~hmH ou mmumwma .mmumum Monaco «macausuflumsH mo Houucoo mo mama an .80flumospm Hmnmfim mo msowusufiumsH 8H mmumm pumom cam Boom .mmom pom cowuwsa pmuwfiflummll.a Manda 42 As the data indicate, the price gap between private and public institutions has widened over the period con- sidered. Preliminary government figures indicate that income from student tuition and fees has increased slightly faster than total spending during the 1971, 1972, and 1973 fiscal years.54 Student tuition and fees last year (1972- 73) accounted for 13.6 percent of the income of public colleges and universities, and for 36.0 percent of the income of private institutions.55 G. Richard wynn, in a study of pricing at liberal arts colleges,56 found that for the 425 sample colleges, the total percentage growth of tuition and fees from 1964- 65 to 1971-72 was 81.4 percent (8.9 percent compounded annually), while total student charges (including room and board) increased 60.4 percent (7.0 percent compounded annually). When the data were deflated by the Consumer Price Index, the growth was, 38.0 percent for tuition and fees, and 22.0 percent for total student charges, over the 1964-65 to 1971-72 period. Price increases by liberal arts colleges exceeded the general price inflation of the economy, during the study period. 54"Tuition, Fees Rising Faster Than Colleges' Spending," The Chronicle of Higher Education, Vol. 7, No. 36 (June 24, 1974), p. 6. 55Ibid., p. 6. 566. Richard wynn, "Liberal Arts College Pricing: Has the Market Taken Over?" Liberal Education, Vol. 58, ' No. 3 (October 1972), pp. 422-432. 43 A further price comparison was made by Wynn between the 425 liberal arts colleges and 42 universities. The net difference in 1971-72 was $1,242 in total student charges (the 425 colleges were higher by this amount). The pro- jection of the difference to 1978-79 was $2,264.57 The increasing absolute price gap between the private and public sectors, coupled with a narrowing of product difference may result in large numbers of potential regis- trants dropping out of the private education market. This price level impact is most likely to affect the lower and middle income strata of prospective private college students. When only the most selective and prestigious col- leges, both private and public are considered, the rates of cost increase were roughly the same, and were not increasing much faster than the general price level, as measured by the GNP deflator, during the period from 1967 to 1971.58 For this group of elite schools, Richard R. Spies found that students try to find schools that closely match their own academic ability, and that financial consider- ations (both income and costs) are of only secondary sig- nificance. However, students are less likely to apply to a school as it gets more expensive, all other things being 57Ibid., p. 427. 58Spies, gp. cit., p. 17. 44 equal; and high-income students are less affected by costs than their low-income counterparts.59 Other Choice Factors Havinghurst and Rodgers60 have drawn a multiplicity of psychological and situational factors together into a probability equation to describe whether a given high school graduate will go to college. The probability depends on the following factors: mental ability; social expectation, or what the family and society expects of him; individual motivation, or his own life goals; financial ability in relation to the cost of continued education; propinquity to an educational insti- tution. The resulting equation is stated: P==a.amamalatdlfl30 +k>(sxfial¢aqecumfion)4-c Chtfivflmmd motivation) + d (financial ability) + e (propihquity): Beezer ahd Hjelm61 in a summary of research on what influences college attendance cite: (l) parental 59Ibid., pp. 34-37. 60Robert J. Havighurst and Robert R. Rodgers, "The Role of Motivation in Attendance at Post-High School Edu- cational Institutions," Who Should Go To College, ed. by Byron S. Hollingshead (New York: Columbia University Press, 1953), p. 137. 61Robert H. Beezer and Howard F. Hjelm, Factors Related to College Attendance, Cooperative Research Monograph, No. 8, U. S. Department of Health, Education, and Welfare (washington, D.C.: U. S. Printing Office, 1963), pp. 35-37. 45 characteristics, i.e., occupation, education, attitudes, and ethnic origin; (2) high school characteristics, i.e., size, peer influence, teacher and guidance personnel influ- ence, and curriculum; and (3) community characteristics, i.e., socioeconomic levels and proximity to a college, as important influences, but in varying degrees. The college environment and its impact on present and prospective students also has been the subject of con- siderable research. Pace and Stern62 developed the College Characteristic Index (CCI) as an instrument to measure environmental forces, called presses, and thus describe the college. Astin and Holland63 developed the Environmental As- sessment Technique (EAT). EAT is based on the belief that the Characteristics of the college environment are largely depend- ent on the characteristics of the student body. Specifically, EAT is defined in terms of eight variables: size of the stu- dent body; the mean intelligence level of the students; and the personal orientation of the student body as reflected in 62George G. Stern, Preliminary Manual: Activities Index--College Characteristics Index (Syracuse: Syracuse Psychological Research Center, 1958); C. Robert Pace, "Evalu- ating the Total Climate or Profile of a Campus," Current Issues in Higher Education 1961, ed. by Kerry G. Smith (Wash- ington, D.C.: National Education Association, 1961), pp. 171- 175; and C. Robert Pace, "Diversity of College Environments," Journal of the National Association of Women Deans and Counselors, Vol. 25 (1961), 21-26. 63Alexander W. Astin and John L. Holland, "The Environ- mental Assessment Technique: A Way to Measure College Environ- ments," Journal of Educational Psychology, Vol. 52 (1961), 308- 316. 46 the percentage of baccalaureate degrees awarded to students in each of six classes of major fields--Rea1istic, Scientific, Social, Conventional, Enterprising, and Artistic.64 A third instrument, entitled College and University Environmental Scales (C.U.E.S.) was developed to sample the general atmosphere of a college. It consists of five scales: (1) Practicality, (2) Community, (3) Awareness, (4) Propriety and (5) Scholarship.65 66 67 C.U.E.S. was applied by Pace and Abott, and both found that incoming students and presently enrolled students had different perceptions of the college environ- ment. In the Pace study the incoming students' statements about their ideal college and what they expected from their chosen college were nearly identical, but both differed substantially from the actual profile of the college they hoped to enter. Such an information discrepancy reflects the inaccurate or uninformed state of prospective students during their college decision process. In another study of institutional images as a factor in college choice it was found that the images held by 64Alexander W. Astin, Who Goes Where to lelege (Chicago: Science Research Associates Inc., 1965), P. 22. 65C. Robert Pace, College and University Environment Scales (Princeton: Educational Testifig Service, 1963); and C. Robert Pace, "Five College Environments," College Board Review, No. 41 (1960). pp. 24-28. 66C. Robert Pace, "When Students Judge Their College," College Board Review, No. 58 (Winter 1966), PP. 26-27. 67Abbott, 9p. cit., pp. 104-113. 47 students differed markedly. Moreover, the images held by entering freshmen were different from those held by sopho- mores. The reasons reported for selecting each campus differed from one another in a direction congruent with the different images held of the campuses by the freshmen.68 The image of an institution is apparently one critical element in understanding the complexities of the student's college choice, just as product image is critical in most consumer buying decisions. Marketing in Higher Education The recent conditions of competition and financial difficulty in higher education have led to more open refer- ence to and interest in the application of marketing tech- nology to the field, particularly among private colleges. Krachenberg69 has suggested the use of the McCarthy, 4P's model--price, place, promotion, and product--as a suitable framework for marketing strategy planning in higher education. 70 O'Brian states, "Private institutions in the long run have no alternative but to satisfy their customers." He 68A. I. Morey, "Institutional Images: Importance to Student Choice of College," American Educational Research Association Paper and Symposia Abstracts, 1972, p. 97. 69A. R. Krachenberg, "Bringing the Concept of Marketing to Higher Education," The Journal of Higher Edu- cation, Vol 43, NO. 5 (May 1972)} 369- '380. A 70Edward J. 0' Brian, "Marketing Higher Education," College and Universitngournal, Vol. 12, No. 4 (September 48 recommends the application of the marketing philosophy to orient the organization's total operation toward meeting the wants of its student customers. With the realization that higher education has lost its vaunted position in the eyes of the public, administrators of colleges and universities must be prepared to enter into competition with all other suppliers of products and service8--educational and non-educational.71 Sutton72 calls for college admission directors to construct a written marketing plan for college admissions. The development of such a plan would include six essential steps: diagnosis (market research), prognosis (projection of where the college is going), objectives (planned across time for geographic areas, majors, and quality of students), strategy (personnel, training, budgeting, and communication), tactics (specifics of how the school is presented to the students), and control (measures to evaluate the strategy and tactics). The need to move to a more open marketing approach in admissions has resulted from the students making their own choice. Mr. Ted S. Cooper of The National Association of College Admissions Counselors, commented: No less than 10 years ago most of the exclusive and influential institutions in the country felt little obligation to inform potential students about the 7lIbid., p. 22. 72David 8. Sutton, "Marketing Tactics Put System Into Recruiting," College and University Business, February 1972, pp. 52-53. 49 selection process. . . . a goodly portion of their game plan was to make the student feel that he was about to enter into a secret society, shrouded in mystery and promises of financial success and intel- lectual enlightenment.7 The view that the student is a consumer with alter- native choices available to him, and the need for colleges to develop marketing plans is widely suggested in the literature. However, these plans, based on principles of sound business management, are not always accepted by col- lege and university administrators. "In fact, there are a few academicians who automatically reject any proposal that uses the terminology of business."74 The current educational environment, as one article suggested, rejects the "order taker" role of college admission offices; rather the admissions director must be a combination marketing analyst, manager by objectives, communication/graphics image broker, and a sales-oriented planner.75 While these new roles are important to the college's vested interest, the utilization of a more open admission process and marketing techniques will also help the student 73Stanford Erickson, "Marketing Is Only a Part of Admission," College and University Business, February 1972, p. 56. 74Luther H. Hoopes, "Your Recruiting is Showing," College and University Journal, November 1973: p. 31. 75"Looks Like an Art, Acts Like a Science," College and University Business, February 1972, p. 47. 50 select the college that best meets his needs. If properly used, they will efficiently and genuinely differentiate 76 The measurement of the choices for student and school. efficiency and effectiveness remain germane questions in the analysis of marketing application to this field. The bandwagon approach to the adoption of some marketing tech- niques, by some schools, does suggest doubt about the appropriateness of some marketing techniques. Practices such as paying finder fees to free-lance recruiters, multiple-college recruiting, discounting with no-need scholarships, guarantees of advanced credit, and other practices have been considered by some as ethically ques- tionable.77 The main mechanism for controlling such abuses is the National Association of College Admissions Counselors and the College Board. So far these groups have relied solely on sending out cease-and-desist letters to stop reported abuses.78 "The recruiter and public relations practitioner should share the conviction that the recruiting program of any institution of higher learning must be related to pur- poses and goals that have basic integrity as regards the 76Ibid., p. 47. 77Larry Van Dyne, "Quest for Students Leads Many Colleges to Adopt Sales Techniques Once Shunned on Campuses, The Chronicle of Higher Education, May 13, 1974, p. 7. 78 Ibid., p. 9. 51 larger interests of human society."79 Yet, the need for colleges to communicate with their various publics is basic. "In a democratic society, every idea is competing with every other idea in the marketplace for public knowledge, public interest, and public support."80 Reference to marketing practices and applications are quite frequent in the current literature of higher education. In most instances the focus is on the communi- cation or promotion elements of marketing, particularly as they relate to student recruitment. Antecedent under- standing of the prospective student's buying behavior appears to be lacking. As this literature review has indicated, educational research studies of the student are numerous, but the perspective of marketing, i.e., viewing the student as a buyer of the educational product, is generally lacking. The perspective of consumer behavior research applied to prospective students is felt to be an appropriate approach to develop a foundation for planning legitimate marketing strategies by a college. This research study is designed with this basic premise in mind. 79Wesley Sheffield and v. P. Meskill, "The Ethics of College Recruiting," gellege and University Journal, Vol. 13, No. 2 (March 1974), pp. 26-27. 80Edward L. Bernays, "Parity for Public Relations in Higher Education," College and UniversityAJournal, Vol. 11, No. 4 (September 1972), p. 7. CHAPTER III RESEARCH FRAMEWORK, HYPOTHESES, AND METHODOLOGY Research Problem Statement The premise of this research is that the use of a market segmentation approach based upon buying behavior theory can be used by colleges to identify different market segments and to plan their marketing effort. Planning based upon the recognition of segmental differences within the prospective student population will thus produce more ef- fective and efficient college marketing programs. An indi- vidual college, particularly a private one, is highly depend- ent upon the revenue flow via fees and tuition to provide operating revenue to maintain its operation. While profit is not the objective, survival is, and this requires revenue adequate to meet costs in the long term. Since students are the source of as much as 60 to 90 percent of the private college's revenue, the emphasis is placed on attracting students. The similarity in revenue needs of the private college and the private business firm suggests a similarity in need for the application of marketing technology and a marketing philosophy. Students are customers. They make institutional and educational product choices, i.e., 52 53 buying decisions. Hey these decisions are made and why_a certain educational institution is selected while others are rejected is the subject of this research. The most direct association of marketing technology and college effort appears to be in the area of student recruitment. This is the focal area for current revenue planning, and the area where most marketing activity is being applied. While the various tactics of colleges are quite evident, e.g., commercial advertising and professional recruiting, there is little evidence of strategic planning or measurement and evaluation of the effort. The specific research purposes of this study are to provide additional knowledge and understanding relating to: 1. the prospective student's information seeking and search processes used in making the college choice, 2. the identification of the relative importance of selected evaluative criteria used in making the college choice, and 3. the identification of segmental differences and similarities within a group of prospective students, who have indicated an interest in a specific college, at selected time reference points and across time. Just as consumers must (1) identify a buying problem, (2) decide upon a class or type of product, and (3) choose a brand and/or a source; the prospective college student must decide (1) whether or not to attend college, (2) what type of 54 college to attend, and (3) which specific college to attend. The brand and source decision of the consumer is collapsed into the institutional choice decision of the prospective college student, since the producer and source are combined. This is the same for most situations involving the marketing of services. The purchase process view of a college choice de- cision and the likelihood of differences in purchasing be- havior and personal characteristics, within the prospective college student market, suggests the potential applicability of market segmentation and consumer buying behavior theory to the research problem. Market Segmentation: Theory and Research The recognition of a need to identify and know the market group or groups to be served by a firm has led to the development of considerable research and resulting theory on ways to segment markets. From the time Wendell Smithl introduced his concept of market segmentation until the present time, correlates have been sought to divide the mass market into segments which have within group homogeneity and between group heterogeneity. These differences between the segments of the total market become useful in market 1Wendell R. Smith, "Product Differentiation and Market Segmentation a8 Alternative Marketing Strategies," Journal of Marketing, Vol. 21 (July 1956), 3-8. 55 planning only when the conditions of measurability, ac- cessibility, and substantiality2 are met. The real benefit of segmentation to the firm and consumers results from the opportunity to develop more specifically tailored marketing programs. While all market- ing variables may be adjusted to the specific character— istics of the various segments, the promotional variables have drawn a disproportionate amount of emphasis. Variation in the response of consumer segments to differentiated promotional programs is frequently experienced by firms. Frank, Massy, and Wind, suggest that strategies for market segmentation can be broken down in terms of a number of dimensions: two of the most important ones are (1) mar— keting tool variables (components of the marketing mix) which are used to exploit the differences between market segments, and (2) methods of targeting marketing effort, i.e., directing it to one segment as opposed to another.3 Where market segments can be identified by a col- lege, either of the two strategies cited above would be appropriate. A purpose of this research study is to illus- trate the application of the philosophy of market segmentation, 2Philip Kotler, Marketing Management: Analysis, Planning, and Control (2nd ed.; Englewood Cliffs, N.J.: Prentice-Hall, 1972), pp. 167-168. 3Ronald E. Frank, William F. Massy, and Yoram Wind, Market Segmentation (Englewood Cliffs, N.J.: Prentice-Hall, 1972) I pp. 6—70 56 as it has been developed in marketing, to the education sector. The specific problem of identifying differences between the characteristics of variously defined market segments will be addressed. This study does not attempt to directly explore the differences in response to specific marketing variables, which is surely a part of a total market segmentation analysis. Rather, it focuses upon identifying the differences which exist between segments selected upon an 3 priori basis, i.e., ACT and SAT segments; and between segments determined by behavioral classification. The differences found between and within segments, will serve as a potential base for controlled coverage of marketing effort. Determination of the response differences of the various segments to specifically directed marketing effort is left to future studies. This study is only a first stage effort. It is recognized, however, that one critical cri— terion for determining the desirability of segmenting a market is whether or not the submarkets have different elasticities with respect to the marketing policies of the firm.4 This would be equally true for a college and its marketing policies. Bieda and Kassarjian5 in their search of the market segmentation literature concluded that two approaches to 4Ibid., pp. 133-134. 5John C. Bieda and Harold H. Kassarjian, "An Overview of Market Segmentation," Marketing in a Changing WOrld, ed. by Bernard A. Morin (Chicago: American Marketing Association, 1969), pp. 249-253. 57 segmentation seemed to emerge. One approach is where the researcher starts with an existing product and studies the customers of that generic product to determine if there are differences between buyers of different brands. The other type of segmentation research starts with preconceived notions of what the critical segmentation variables are-- social class, personality, cultural variables, etc., then members of each segment are isolated, and product usage, brand loyalty, media exposure, etc., are then collected and analyzed. In general, the consistency of the results tend to indicate that the research in market segmentation has been either unsuccessful or if a relationship is shown, quite weak. The poor results of these studies are mainly attri- buted to unrealistic assumptions made in developing the methodology used, and the attempts to use demographic and psychological type variables to predict product choice. Frank7 found the most frequently used bases for defining market segments to be considered targets for pro- motion were: (1) demographic and socioeconomic character- istics, occasionally together with personality traits; and 6Ibid., pp. 249-253. 7Ronald E. Frank, "Market Segmentation Research: Findings and Implications," The Application of the Sciences to Marketing Management, ed. by Frank M. Bass, Charles W. King, and Edgar H. Pessemier (New York: John Wiley and Sons, Inc., 1968), Pp. 39-68. 58 (2) purchasing characteristics, especially the total con- sumption of a product, i.e., heavy versus light buyers and brand loyalty. In his evaluation of the effectiveness of these bases for market segmentation he expressed doubt about their usefulness. Nondemographic market segmentation bases including; personal values, susceptibility to change, purpose, aesthetic concepts, attitudes, individualized needs, and self-confidence were found to be more useful than demographic bases by Yankelovich.8 Volume segmentation (the so-called "heavy 10 have also been half" theory)9 and benefit segmentation used successfully. Other research which focuses on consumers' activi- ties, interests, prejudices, and opinions; and variously called "psychographic" research, "life-style" research, and even "attitude" research attempts to draw recognizably human portraits of consumers which can be utilized in segmenting a market.11 8Daniel Yankelovich, "New Criteria for Market Seg- mentation," Harvard Business Review, Vol. 42 (March-April 1964), pp. 83-90. 9Dik Warren Twedt, "Some Practical Applications of the 'Heavy Half' Theory" (New York: Advertising Research Foundation 10th Annual Conference, October 1964). 10Russell I. Haley, FBenefit Segmentation: A Decision- oriented Research Tool,” Journal of Marketing, Vol. 32 (July 1968), 30-35. 11William D. Wells and Douglas J. Tigert, "Activities, Interests, and Opinions," Journal of Advertisigg Research (The Advertising Research Foundation, Inc., 1971). 59 Schools of Thought Two schools of research on marketing segmentation, (l) the "Behaviorally" oriented school, and (2) the "Decision" oriented school appear to exist.12 The "Behaviorally" oriented school is concerned with the identification and documentation of generalizable differ- ences among consumer groups which can lead to insight into the basic processes of consumer behavior. Behavioral science theories and accumulated empirical research findings from both inside and outside the marketing field provide the guidelines and hypotheses for behavioral market segmentation research. The "Decision" oriented school is also concerned with the existence of group differences in consumption and the prediction of such differences from customer characteristics. However, this school places greater emphasis on "how" to use the findings to improve the efficiency of the firm's marketing program and less emphasis on "why" such differences occur. The expanded attention given to the development of buying behavior models and theory, such as those by Nicosia,13 12Frank, et a1., 92. cit., pp. ll-13. 13Francesco M. Nicosia, Consumer Decision Processes: Marketing and Advertising Implication (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1966). 60 Howard and Sheth,14 and Engel, Kollat, and Blackwell15 contribute insights and perspectives which can be used to structure market segmentation research. The emphasis of these models upon the individual's buying behavior processes is consistent with the aggregative approach of developing market segments useful to the firm in its marketing planning. Claycamp and Massy16 point out that market segmen- tation should be considered a process of aggregation rather than disaggregation. Because of the diseconomies usually associated with develOping separate marketing strategies for each individual, consumers must be aggregated into larger groups. The best level of market segmentation and combi- nation of marketing strategies will be determined by the profit maximization rule. The ideal method of aggregating consumers into market segments would be based upon their similarity of response to marketing stimuli. To operationalize the aggregative approach to market segmentation would appear to require considerable under- standing of buying behavior processes. The current research in the buying behavior area may provide a foundation for 14John A. Howard and Jagdish N. Sheth, The Theory of Buyer Behavior (New York: John Wiley and Sons, Inc., 1969). 15James F. Engel, David T. Kollat, and Roger D. Blackwell, Consumer Behavior (2nd ed.; New York: Holt, Rinehart, and Winston, Inc., 1973). 16Henry J. Claycamp and William Massy, "A Theory of Market Segmentation," Journal of Marketing Research, Vol. 5 (November 1968), 388-394.' 61 market segmentation studies which are more productive than those-using demographic or other general bases.‘ Decision process theories of consumer behavior lead one to select measures which differ substantially from those used in previous studies. These theories emphasize the process which generates buying behavior. Inferences from these theories suggest that: (1) relationships probably exist between a consumer's personal character- istics and his purchase decision process: and (2) indi- viduals who have similarly structured purchase decision processes are likely to exhibit over-all similarity in buying behavior}.7 Empirical tests by Lessig and Tollefson provided support for these relationships. Blattberg and Sen18 evaluated (1) Customer character- istic segmentation, (2) Attribute segmentation, (3) Purchase behavior segmentation, (4) Consumer characteristic-Purchasing behavior segmentation, and (5) Perceptual mapping segmen- tation, and concluded that all five major segmentation approaches had some disadvantages. They, in turn, recommend a multi-stage segmentation approach as an improved method- ology. While a large number of different segmentation approaches have been taken by researchers, the number of l7V. Parker Lessig and John O. Tollefson, "Market Segment Identification through Consumer Buying Behavior and Personal Characteristics," Marketing Segmentation: Concepts and Applications, ed. by James Engel, Henry F. Fiorillo, and Murray A. Cayley (New York: Holt, Rinehart, and Winston, Inc., 1972), p. 436. 18Robert C. Blattberg and Subrata K. Sen, "Market Segmentation Using Models of Multidimensional Purchase Behavior," Journal of Marketing, Vol. 38 (October 1974), 17-28 0 62 different variables used has been even larger. Hisrich and Peters19 examined the significance of each of four commonly used segmentation variables (income, social class, age, and family life cycle) as a correlate of two aspects of purchase behavior associated with various entertainment activities. They concluded: (1) a variable significant in one market/ product context may not be significant in another; and (2) the practitioner, at least in the instance of many consumer goods and services, should be concerned with the frequency of use of his product or service when determining the segmentation variable(s).20 21 found that of seven independent variable Wiseman sets the most important set in predicting automobile pur- chasing behavior was the "shopping patterns and usage expectation" set. Attempts have also been made to track the consumer 22 used a self- through his decision making process. O'Brien reportive consumer panel to provide data for an operational model based upon hierarchy-of-effects theory. 19Robert D. Hisrich and Michael P. Peter, "Selecting the Superior Segmentation Correlate," Journal of Marketing, Vol. 38 (July 1974), 61. 20 Ibid., p. 63. 21Frederick Wiseman, "A Segmentation Analysis on Automobile Buyers During the New Model Year Transition . Period," Journal of Marketing, Vol. 35 (April 1971), 46. 22Terrence V. O'Brien, "Tracking Consumer Decision Making," Journal of Marketing, Vol. 35 (January 1971), 34-40. ll! [III II. [111“ 'll‘ ‘11. Ill III 1‘ [‘1 Ill! [1 I II. I." I ‘|‘ ll 63 These two findings-~no attitude-purchase effect and no attitude-intention effect--contradict many of the findings in behavioral marketing. Attitude is apparently a genuine aspect of thinking, but it is a prodggt of purchase decision making not a determinant of it. (In this study attitude represented affect, and intentions represented the likelihood of purchasing a brand.) Consumer Choice Behavior Consumer choice behavior has drawn considerable recent attention in marketing. Two approaches have been apparent. One approach involves the investigation of constructs rigorously and in great detail. "Multi-attribute models of attitude provide an example of this type of re- search which, though model oriented, is limited in its scope (Bass, Pessemier, and Lehmann, 1972; Lehmann, 1971, 1973; Fishbein, 1967; Rosenberg, 1956; Wilkie and Pessemier, 1973)."24 Another approach is the development of larger-scale behavioral models which are more complex, but show less detailed concern with constructs and greater concern with relationships among the constructs. Models of this type 23Ibid., p. 40. 24Donald R. Lehmann, Terrence V. O'Brien, John U. Farley, and John A. Howard, "Some Empirical Contributions to Buyer Behavior Theory," Journal of Consumer Research, Vol. 1 (December 1974), p. 43. 64 would include those of Nicosia (1966), Engel, Kollat, and Blackwell (1973), and Howard and Sheth (1969).25 The Fishbein attitude model in particular has stimulated research interest into its application to con- sumer choice. However, several controversies have arisen regarding the Fishbein model as it has been adapted to consumer choice problems.26 Attitude according to Fishbein's theory is comprised of two components: (1) the strength of a belief about an object, which is defined as the probability that the atti- tude object is related to some other object, and (2) the evaluative aspect of a person's belief, i.e., its ”goodness" or "badness."27 The application of Fishbein's theory to marketing has generally been to predict relative preference for similar objects, e.g., brands of products. Bass and Talarzyk28 predicted brand preference for six product categories by measuring beliefs about salient 25Ibid., p. 43. 26Masao Nakanishi and James R. Bettman, "Attitude Models Revisited: An Individual Level Analysis," Journal of Consumer Research, Vol. 1 (December 1974), 16. 27Martin Fishbein, "A Consideration of Beliefs, Attitudes, and Their Relationship," Current Studies in Social Psychology, ed. by Ivan D. Steiner and Martin Fishbein (New York: Holt, Rinehart and Winston, Inc., 1965), P. 117. 28Frank M. Bass and W. Wayne Talarzyk, "A Study of Attitude Theory and Brand Preference," Marketing Involvement in Societygand the Economy, ed. by Philip R. McDonald (American Marketing Association, Fall Conference, 1969), pp. 272-279. 65 attributes of competing brands and evaluative aspects of those beliefs. Their research strongly supported the hy- pothesis that brand preference is related to attitude measurements based on product attributes. 29 found that some attributes, Meyers and Alpert while they are very important to consumers, are taken for granted. They concluded that attitudes toward features which are most closely related to preference or to actual purchase decisions are determinant; and the other features or attitudes, no matter how favorable, are not determinant. However, Sheth and Talarzyk applying a Rosenberg type model to product specific attributes failed to find any improve- ment in the prediction of affect when the value of importance 30 component was included. Further support was found by Scott 31 for the conclusion that it is not necessary to and Bennett scale attribute importance so long as only important attri- butes are included in the study. They too used the Rosenberg *— 29James H. Meyers and Mark I. Alpert, "Determinant Buying Attitudes: Meaning and Measurement," Journal of Marketing, Vol. 32 (October 1968), 13-20. 30Jagdish N. Sheth and W. Wayne Talarzyk, "Relative Contribution of Perceived Instrumentality and Value Importance in Determining Attitudes Toward Brands," Broadening the Con- cept of Magketing (Chicago: American Marketing Association, 1970), p. 35. 31Jerome E. Scott and Peter D. Bennett, "Cognitive Models of Attitude Structure: 'Value Importance' is Im- portant," American Marketing Association Combined Proceedings, 1971, ed. by Fred C. Allvine (Chicago: American Marketing Association, 1972), pp. 348-349. 66 model, which postulates consistency between the affective and the cognitive components of attitudes.32 Scott and Bennett's research also showed, however, that different product attributes were ordering appeal in the different market segments. "It is evident, then, that prior clustering of participants on the importance of attributes may be necessary to avoid errors in ranking determinant attributes or to avoid missing attributes peculiar to individual segments."33 A comprehensive review of the research on the ef- fects of "weights" in the weighted, additive utility (WAU) models is provided by Moinpour and Wiley. They concluded: The results of these studies generally suggest that "weights” incorporated into the WAU model contribute little to its predictive power. All aspects of the "weighting" hypothesis, however, have not been thor- oughly investigated. The issue remains an important area in consumer attitude research.34 In Moinpour and Wiley's research on the predictive qualities of "important" attributes compared with "unim- portant" attributes, they found that higher quality predic- tions can be made from the "important" attributes, however, 32Ibid., pp. 346-347. 33Ibid., p. 350. 34Reza Moinpour and James B. Wiley, "An Empirical Investigation of Weighted, Additive Models of Attitude in Marketing," American Marketing Association Combined Pro- ceedings, 1972, ed. by Boris W. Becker and Helmut Becker (Chicago: American Marketing Association, 1973), p. 388. 67 respectable predictions can also be made from the "unim- portant" attributes.35 While the major research focus of these cited studies was on the quality of the prediction of attitude, and the relevance of the weighting element in the quality of the resulting predictions, the inference can be drawn that the importance (weighting) structure of attributes may serve as a basis for segmentation. Attitude measures, however, may not be effective predictors of actual purchase behavior, and market segmentation theory does emphasize the actual purchase of the product in the evaluation of the effective- ness of a particular segmentation scheme. The need for a better understanding of the linkage between attitude and actual purchase behavior seems apparent. The role of an intervening variable in this process has been summarized by Engel, et a1. . . . it thus may be concluded that attitude usually will not predict behavior accurately unless intention is utilized as an intervening variable. Intentions, in turn, predict behavior to the extent that outside moderating influences are absent or at a minimum. When these environmental constraints are operative, their influence also must be accounted for if behavior is to be predicted. Therefore, attitude change is a valid marketing goal, because a change in attitude is reflected by a change in behavior as expressed through changed intentions.36 35Ibid., p. 389. 36James F. Engel, et al., Consumer Behavior, 2nd ed., pp. cit., p. 274. I '1" I‘ll) l. llll.‘ I '1‘ 1| I! 11 II I III 1" (II I! IIII' I I I'll 'I‘ I '1 II. III I I!» ‘l l l l | {I 68 Most of the previous research using product-specific attitude measures to predict preference have been static analyses. In an experimental study of attitude change and the choice of a new brand Ginter37 evaluated attitudes, preference, and previous choice as predictors of choice measured at several points in time. Analysis of choice indicate that preference was a better predictor than the multi-attribute measure of affect. The attitude measure was a better predictor than previous choice, the new brand, or the brand to which the subject was previously most loyal. . . .33 Hi8 results also indicated that attitude change does occur both preceding and following behavior change, and that 39 He did not postchoiCe attitude change was greater. investigate the specific cause of the postchoice attitude change (whether it was caused by additional information or cognitive dissonance). Rosenberg40 in reference to attitude change and attitude organization asserts that most individuals cannot long tolerate inconsistency, and they are motivated to 37James L. Ginter, "An Experimental Investigation of Attitude Change and Choice of a New Brand," Journal of Marketing Research, Vol. 11 (February 1974), 30-40. 38 Ibid., p. 39. 39Ihid., p. 39. 40Milton J. Rosenberg, "Inconsistency Arousal and Reduction in Attitude Change," Current Studies in Social Psycholggy, ed. by Ivan D. Steiner and Martin Fishbein (New York: Holt, Rinehart and Winston, Inc., 1965), PP. 122- 125. l.ll|ll|ll|ulu.|ll',ll|ll III III II Ill" ‘| 'll‘ll‘l‘.’|| II 'II' [III-ll '1 ['1' l l ll 69 maintain internal consistency between the affect and cogni- tive components of attitude. The personality constructs of cognitive clarity and cognitive style derived from the work of Kelman and Cohler as cited by Sweeney, Mathews and Wilson41 are also related to attitude change and the person's reaction to persuasive communication. Two types of indi- viduals were identified in terms of their cognitive styles (i.e., an individual's way of dealing with situations involving ambiguity and incongruity), "clarifiers" and "simplifiers." The "clarifiers" were found to be more likely than "simplifiers" to manifest an attitude change following persuasive communications. Resistence to attitude change is said to be less when the attitudes are peripheral to the self-concept, basic values, and other significant focal objects.42 Conversely, those attitudes which have psychological centrality, personal goal revelance, and are anchored by other attitudes in the system will tend to be more difficult to change, as compen- satory attitude changes must follow to restore balance.43 41Timothy W. Sweeney, H. Lee Mathews, and David T. Wilson, "An Analysis of Industrial Buyers' Risk Reducing Behavior: Some Personality Correlates," American Marketing Association Combined Proceedings, ed. by Thomas V. Green (Chicago: American Marketing Association, 1974), pp. 217-218. 42W. J. McGuire, "The Current Studies of Cognitive Consistency Theories," Cognitive Consistency, ed. by S. Feldman (New York: Academic Press, 1966). 43T. M. Newcomb, R. H. Turner, and P. E. Converse, Social Psychology (New York: Holt, Rinehart and Winston, 1965)] p0 1360 ”I All 7' I'll I‘ll ll I I I I'll I lll‘fi’l I I III. I l 'lllllll.| I II It 70 A person's degree of commitment to a position has also been found to influence his message perception (assimilation effect and contrast effect).44 In summary, this literature review reflects the current interest and research in market segmentation, and the various approaches taken to identify homogeneous cus- tomer groups. It is apparent that previous research on market segmentation has not determined any single set of variables which is universally applicable in dividing a total market into market segments. Socioeconomic, demo- graphic, personality, life-style, and perceived benefit variables, to name a few, have been used to segment markets, with a varying degree of success. Current literature also suggests an association between aspects of buying behavior theory and market seg- mentation analysis. The linkage of these two areas lies in the contribution buying behavior research has made to the identification of new bases for aggregating individuals into more responsive submarkets. The decision process theories of consumer behavior suggest that relationships probably exist between a con- sumer's personal characteristics and his purchase decision process. Individuals who have similarly structured purchase 44Carolyn W. Sherif, M. Sherif, and R. Nebergall, Attitude and_Attitude Change (Philadelphia: W. B. Saunders Company, 1965). 71 decision processes may also exhibit over-all similarity in buying behavior. Such patterns of similarity, if they can be identified, could be used to segment the college market just as they are used to segment a consumer good market. General Hypgthesis Buying behavior theory postulates that buyers be- ginning to purchase a product class where they lack prior purchase experience, will not have a well-defined set of evaluative criteria or a high level of knowledge about the various products or brands in that product class. This condition leads to an active search for information necessary to make a product choice. Along with the active search for information, the buyer may, to a considerable extent, gener- alize from similar experiences in the past.45 The general hypothesis of this study is that pro- spective college students have a weakly structured set of evaluative criteria, which is subject to change during the college buying process. The degree of structuring and the stability, i.e., resistence to change, of the evaluative criteria will vary with the prospective student based upon his or her exposure to, knowledge of, and experience with college associated information and decision making processes. Social, economic, and other demographic factors may affect the prospective student's level of information about 45Howard and Sheth, 9p. cit., p. 26. 72 college attendance in general and the characteristics of specific colleges. Family influence in particular is expected to play a major role in providing information, influencing the evaluative criteria structure, and affect- ing the college evaluation process. Peer group and other reference groups also serve as potential sources of infor- mation and influence. The basic decision to attend college, the choice of the type of college, and the choice of the specific college, all involve a cost in terms of time, money, foregone oppor- tunities, effort, and associated psychological and social risk. A purchase decision with this degree of importance would seem to require considerable information and thought, and would be expected to require an external search for information, consistent with buying behavior theory. An exception might be where the prospective student is a member of a college educated family, capable of providing the necessary information. Areas for Research Hypotheses The exploratory nature of this study did not permit a complete enumeration of hypotheses to be tested prior to data collection. Throughout the data analysis phase addi- tional hypotheses were formulated and tested as they were suggested by previous findings. To guide the determination of needed data, and the develOpment of apprOpriate data collection instruments, five 73 basic areas of exploration were set out and five major hypotheses were stated. It was anticipated that each area would allow the generation of additional specific hypotheses during the course of the research. The five basic areas of research were: 1. The identification of the importance of selected evaluative criteria and the change in these criteria over time, within and across identified market segments 2. The association of prior economic goods pur- chase patterns with the college choice patterns 3. The association of socioeconomic and demographic variables with the purchase behavior patterns and preference statements of the prospective students 4. The change over time in the purchase intentions (college and major) of the prospective students 5. The level of college associated information and the usefulness of various information sources in the search process. These five designated areas of research suggested the following major hypotheses. Major Research Hypotheses Hypothesis I: A buying intention statement in terms of the prospective student's choice rating of a particular college, i.e., first, second, third choice, etc., will serve to predict application and enrollment more frequently than other data available to the college. .III‘ I. A D ... it'll“ 74 Hypothesis II: Identifiable market segments of prospective students interested in a particular college, such as, the ACT segment and the SAT seg- ment will differ in their characteristics and behavior. Hypothesis III: Purchase patterns as reported for the pur- chase of economic goods with respect to the level of information and degree of decisive- ness will carry over to the college choice process. Hypothesis IV: Prospective college students will change their assessment of the relative importance of selected evaluative criteria over time. Hypothesis V: Behavior determined segments of prospective college students will differ in the relative importance of selected evaluative criteria at different points in time. Research Design The critical phase of the research study examines the college choice (buying) process of the prospective students by using a longitudinal type research design. VIADE Concept To visualize the longitudinal nature of the choice process the information state, and decision state of the prospective student was conceptualized as proceeding from a Vague (V), to an Informed (I), to an Application (A), to a Decision (D), to an Enrollment (E) condition. The as- sociated conditions of time, number of colleges considered, depth of information, decision structure definition, and action taken are shown in Figure l. ||llll|l|1llll 75 .8888 8888 p8 88808 8H8 3.898 88>... . a 888 g .8888 as mm m» 88888802 8888 m . Ho Ham-8:05 83 En. ”886 882 no one ugaonfim 1. . 88:00 “W. gang?“ :83 R: ...L.. Nu gm» 88.38 m 038me m §u¢ Hofiwmfimm n. . B 8888 88 888 8oz 8 80 8888808 m . Amy $8.300 / As 08888 88 p88 8 >88 8 88802 8 Bug 38 8 888888 w . 889nm H gag can fl Xfi Ad "W p 888 88888.5 8888 8mg 88 38 8888 m. 8.350 8 ages #83 B “838:4 B 8:898 88 EH88 8888 88 >8: 38 88> :98 8888 8883 8888898 88: 88m 8888 88 868m 8888 8 88 A 88880 .8 .8982 8 888th 76 Time Period Definitions l. Pre-application period (t1). This period is defined as the time prior to the submission of an appli- cation for admittance. 2. Post-application period (t2). This period is defined as the time period after the submission of appli- cations for admittance, but prior to the actual enrollment (the beginning of classes) Fall, 1974. 3. Post-enrollment period (t3). This period is defined as the time period after the beginning of classes in which the student has enrolled, and continues until the student discontinues his attendance at the college. ACT and SAT Group Definitions 1. ACT Group. Those prospective college students who had taken the American College Testing Program (ACT) examination and submitted their test scores to the college under study were defined as the ACT group (segment). 2. SAT Group. Those prospective college students who had taken the Scholastic Aptitude Test (SAT), sponsored by the College Entrance Examination Board, and submitted their test scores to the college under study were defined as the SAT Group (segment). Data were collected and analyzed from each of the three time periods to identify changes and differences in behavior patterns, as shown in Figure 2. lll' nil {[1 'I: Al I! III‘ l[[ l I {f {I .I [I (II I 4" [ll '1 'l 77 lACTChxmp I I SHTChxmp] Prefigplflxnflon “8) EafiurInUEHion] [Eanhzrnuaufion I Rxnnqzfliamfionlfié) lfiifiil Dortm. ,Nxflyl Dorkm Apply Apply Post-enrollnent (t3) [Enroll Do Not Enroll Do Not ' """' mefll mefll Figure 2.--Longitudina1 Time Reference-Student Action Pattern. The pre-application data came from the ACT profile data and the SAT data available on prospective students having indicated interest in a specific college. During the post-application period, additional data were collected from a sample of prospective students from both the ACT and the SAT groups. During the post-enrollment period those prospective students responding to the post-application questionnaire were surveyed to determine their actual enrollment decision and why they made this choice. It was expected that this design would allow the development of a pattern of buying behavior which could be used to determine: (1) why some interested students applied, 78 while others did not apply; (2) why some who applied enrolled, and others who applied did not enroll. It was also expected that the design would improve the overall understanding of prospective students information search and decision process behavior. Research Methodology The longitudinal nature of the research design allowed the analysis of data within and between time periods. Separate research methodology was required for each time period. Pre-Application Period Pre-application data in the form of the ACT student profiles and SAT profiles were made available by the col- lege participating in the study. Both data sources were used, but the ACT data were more complete and more consistent with the data needs of the study. These data (ACT) included early college preference (intention) statements, the rank order importance of seven evaluative criteria, and descriptive statements about preferred characteristics of colleges made by the prospective students when taking the ACT examination, in either their junior or senior year of high school. The fact that the student had requested the testing service to send his profile to the school indicated some degree of interest in the college. 79 A total population of ACT profiles received by the school was 193 by July 15, 1974. This included only poten- tial freshman students for Fall, 1974. It was also known at that time which students had applied and which had not applied. A preliminary study was conducted to determine differences between the applied and non-applied groups. A sample of approximately one-third was randomly drawn for study. Tests of significance were made using chi-square analysis. Where comparisons were possible the contingency coefficients were calculated. In addition to the preliminary analysis of pre- application data, these data were used as the initial state- ments (t1) against which later statements were compared to determine change on both an individual and group basis. Post-Application Period The population definition remained the same for this period of the study, i.e., all prospective students who had submitted ACT and SAT profiles to the college prior to July 15, 1974. The population was limited to prospective freshman students for Fall, 1974 enrollment, with no prior college. Foreign students were excluded. Those students who had not been tested or did not submit their profiles to the college were not included. III-'11 80 Sample.--The total sample for the questionnaire mailing was composed of (1) all members of the ACT profile group, and (2) a randomly drawn sample approximating twenty percent of the total SAT profile group. This resulted in an initial mailing list sample size of 357 prospective students, 194 from the ACT profile group and 163 from the SAT profile group. The complete population of ACT members was used because of the greater pre-application period (t1) data available for them. These data were vital to the comparison over time of the evaluative criteria, and the intention statement comparison to actual behavior. Consideration was given to the expected reduction in the sample size caused by failure to respond to the mail questionnaire. Where a student had taken both tests he was placed in the ACT sample only. Data collection.--The data collection was by means of a mail questionnaire sent to the total sample of 357 prospective students. The mailing was made during late July, 1974. And a follow-up letter was sent two weeks later. The data collection instrument was a six page questionnaire which was coded to identify the respondent for analytical purposes, as required by the nature of the study design. The data collection instrument was designed to provide information in the following areas: 81 1. Family background, particularly family edu- cational patterns, income, occupations, and mobility 2. College information search pattern and infor- mation source importance 3. College application pattern 4. Evaluative criteria data, including a ranking and scale measure of importance of those criteria used 5. Prior purchasing pattern profile 6. Degree of preference for a specific college and major Data analysis.--The responding sample group was divided by behavioral classifications for comparative analysis. The applied and non-applied, and later the en- rolled and non-enrolled were the two basic classifications used. Time dependent analysis was also used to determine change in the evaluative criteria and intentions. Statistical methods.--Associative statistical analy- sis to evaluate dependency and difference of populations was made using the chi-square statistical method. Com- parison of change in rank data on an individual basis over time was made by using the Spearman rank correlation coefficient. The Kendall coefficient of concordance was used in the analysis of similarity of judgments by groups. 82 Where comparisons in data could be made, given the require- ment of equal size contingency tables, a contingency coeffi- cient was used. Additional descriptive statistics were used when appropriate in presenting the data. Post-Enrollment Period The third time period, post-enrollment, is defined as the period following Fall enrollment 1974. Since col- leges begin classes at different times, the arbitrary date of October 1, 1974, was used to Operationally define the beginning of the period. Data collection.--A second data collection instrument was developed as a follow-up to the post-application period questionnaire. The major purpose of this follow-up was to determine the following: 1. The importance of the various evaluative criteria 2. Where the students in the sample were actually attending college and their major field of study 3. The profile of their chosen college 4. Why they made their particular college choice 5. What if anything they would do differently in making the college choice. The sample used in the follow-up questionnaire mailing was restricted to those students who responded to the first questionnaire. Data analysis.--Comparisons were made across time. The analysis of the ACT group involved all three data points. llil‘l'l‘l‘rit I’lltll‘lllllll l:llllul¢|li[[f|ll[f"t[’llll‘l 83 For the SAT and combined groups only t2 and t3 period analysis was possible. The statistical methods employed were the same as those described for the post-application period. Data Collection Matrix The data collection matrix shown in Figure 3, summarizes the data sought and the time periods in which the data were collected. Copies of the post-application and post-enrollment questionnaires appear in Appendix B. Limitations of the Study This study was limited by the research design to an exploratory type study. The set of prospective students was narrowly defined as those who had shown a known interest in one specific college. However, these students were not limited in their interest to only one college, as they all considered a number of other colleges. The array of different colleges considered and applied to, as a total, was quite large. This wide variety of colleges and the fact that there were few common colleges in the sets considered from one student to another, limited the opportunity for direct institutional comparison. A further limitation resulted from considering only one component of the currently used attitude or behavior intention models. The "weighting" or "importance" component (evaluative criteria) was utilized, but an evaluation of l l 'u it‘ll-I11! .n’llllrt,[l[.{l[il 'I.‘[-[a[¢’le|‘[['ll[["l‘1(.lllll|'| 84 .88m8 88. 8 88 880888 88 88 88 38.1.8 88 >4><><><><>< ><><><><>< X x 803803 08:00 H085 3339... 838331880.”me 80.82 888888888 888 . 8988 8803 6888 888884 3888 89888388 983 0868081088 0393ng 88:128qu 858 888 388 3.83.8 9,3888 18958881881 HH mgflmg I888888uu8v H 8.888033 1888884881 :88 88 8 8» .HOflamlHOghv 8888 8 888m 88 85 individual colleges with respect to these criteria was not made. This methodology did not permit the determination of a specific attitude measure toward the various colleges. The study was also limited by the nominal and ordinal nature of the data. Statistical analysis was necessarily limited to the use of nonparametric methods, somewhat reducing the power of the significance tests. The sample size, when testing multiple variables by cross classification, was considerably reduced. This proved to be a limitation as certain statistical tests were not possible, while in other cases matrix cells had to be combined to allow hypothesis testing. This had an effect upon the completeness of the analysis in some instances. The fact that all of the prospective students in the study had shown an interest in the college under study (a private four year college) anchored the study to the uniqueness of that institution. This institutional approach parallels the approach taken in many market segmentation studies. However, caution must be taken in generalizing the findings to other situations. CHAPTER IV PRE-APPLICATION PERIOD ANALYSIS Pre-Application Data Analysis Early interest in a college is expressed by a pro- spective student when he directs an academic testing service to submit his test scores and other data to that college. Both the ACT and SAT data are of this type. Data concerning the relative importance of seven college selection factors is furnished by the prospective student when taking the ACT test (usually during his junior or senior year in high school). These seven factors; (1) type of college, (2) student body composition, (3) location, (4) cost, (5) size, (6) field of study, and (7) extracurri- cular activities are ranked in their order of importance. This ranking indicates the relative importance of each factor, as perceived by the student. Consistent with a buying behavior framework, these selection factors can be defined as evaluative criteria used in the college choice process, and the ranking reflects the prospective student's attitude toward their relative importance at a point in time (t1). The ACT student profile also provides a college with data on specific characteristics which the prospective student 86 87 desires in a college (e.g., private college, coeducational, costing over $3,000, located in Michigan, with a business major, and less than 2,000 students). These dimensional statements are more specific than the evaluative criteria, and can be matched against the actual characteristics of a particular college. For example, a student may have ranked location (an evaluative criterion) fourth in im- portance, and specifically indicated the state of Michigan (a specific dimension of location) as the preferred state in which to attend college. An expression of the prospective student's college purchase preference is also provided by the ACT student profile. The specific college to which the test scores are sent is indicated as either a first, second, third, etc. choice (purchase preference). This preference statement reflects an evaluation of specific colleges, presumably consistent with the importance of the evaluative criteria, and based upon the information possessed at that time. These early statements about the importance of evaluative criteria, specific characteristics desired in a college, and the order of preference of particular col- leges provide a college sOme insight into its future appli- cation and enrollment pattern. However, it is not clear if any of the information will serve to predict students' applications and enrollments. 88 Preliminary Studngroup A preliminary study was made to assess the factors which were most frequently associated with the prospective student's decision to apply. The data were from a randomly drawn sample of 68 prospective students from the total population (193 total) of prospective students having sub- mitted their ACT profile reports to the cooperating college, as of July 15, 1974. All prospective students in the popu- lation indicated they were planning to enroll in some col— lege as freshmen in the Fall of 1974. Of the 68 prospective students in the sample, 55 had complete data forms. Twenty of the 55 students had applied to the college under study, and 35 of the 55 had not applied as of the sampling data. Combined Descriptor Match A comparison was made between the prospective students' preferred college characteristics and the actual characteristics of the subject college. The summed frequency of the Match/No Match condition between the preferred college characteristics and the college's actual characteristics was tested across the applied and non-applied groups. The null hypothesis of no difference between the two groups could not be rejected at the alpha = .05 level (Appendix: Table A-l). However, when the frequency of agreement (match) between the preferred college characteristics and the Ilfl.[!ll'\.llll‘ ll' 89 college's actual characteristics was used to classify indi- viduals into two groups; (1) those with a high number of matches (five or more), and (2) those with a low number of matches (less than five), a significant difference was found at alpha = .05 (Appendix: Table A-2). Significantly more prospective students who had applied to the college had a high number of matches. Those not applying to the college more frequently had a low number of matches. The greater number of matching characteristics, without regard to the importance of the associated evaluative criterion, tended to predict application (contingency coefficient, C = .26). Relative Importance of Evaluative Criteria The matched condition considered in the preceding section indicated significant differences between groups where "high" or "low" frequencies of match were identified. However, the analysis did not focus on the difference in the relative importance of the evaluative criteria. The difference in the overall importance ranking of these criteria given by the applied and non-applied groups was determined. Tests of differences could not be made due to the nature of the data, but the absolute data are shown in Appendix, Table A-3. A further descriptive comparison of the rank order values is provided in Table 2. The median value shown is defined as the mid-point value in the array of rank values ll l!!! 'l ‘t I 90 TABLE 2.--Application State Comparison of Evaluative Criteria Importance. Amlied (n = 20) Non—Applied (n = 35) S i 113' S 3 H 1' Onkn:of lumk On urof Radc BquUmxe vanuflfles (nfler nmxntamxa Vhrflflfles Onku' 1. Fiekioftfimdy 1. 1. Piekiofsfimdy 1 2 (rat 2 2 Cam: 2 3 EmmaeofCXHJege 3 3 ‘nnxeof(1fllege 3 4 Ikxztflx: 4 4 loamfion 4 5 Sims 5 5 Extnmnnrhmflar 5 6 somkmmnaxyr 6 6 suxbmtlxdy 5 7 Emtnmxmrflmflam' 7 7 Efize 6 given the variable by the prospective students. Half or more of the prospective students ranked the variable equal to or higher than the median value. For example, "location" had a median value of four, thus, half or more of the students ranked it first, second, third or fourth in importance. The difference in the position of "size" and "extra- curricular activities" in the rank order of evaluative criteria for the two groups is the most evident variation. "Extracurricular activities" was not a well defined de- scriptor, as all students in both the applied and non- applied groups matched on the associated college character- istic. While there was no difference in the match condition, the difference in ranking suggests a difference in importance associated with this variable (see Table 3). 91 Rank Order Classification The frequency of classification by rank order of the evaluative criteria was tested separately for each evaluative criterion, against the classification applied or non-applied. The rank order cells were grouped where necessary to provide an adequate expected cell value to meet the requirements of chi-square analysis. The null hypothesis of independence was tested at the alpha = .05 level of significance (Appendix: Table A-4). A 2 x 2 contingency table was used where the cells required grouping. Where either a 2 x 2 or a 2 x 3 contingency table could be used, both chi-square values were calculated. This allowed the calculation of a comparable contingency coefficient for all variables, which indicates the strength of the association between the applied and non-applied classifications. As indicated in Table 3, the frequency classification by rank order revealed significant differences between the applied and non-applied groups for the variables; (1) extra- curricular activities, (2) type of college, and (3) cost of college. An additional test was made combining the match and no match descriptor classification with the rank order value classification, and testing it across the applied and non- applied classification (Appendix: Table A-5). 92 TABLE 3.--Summary: Difference in Evaluative Criteria Importance by Rank Order Tested Across the Applied and Non-Applied Classification. (kflcuknndcflfirammme criukza Chitkzd Cbnthmmmcy llxueecfif vahxa Vahxa Cbeflfikfient Variables Dependence x2 = 5.99 x2 = 3.84 (2 x 2) Extracurricular Significant 4.84 c = .285 Type of College Significant 9.63 4.32 c = .270 Cost Significant 3.86 c = .257 Student Body Not Significant 2.65 c = .214 location Not Significant 1.84 1.80 c = .179 Size Not Significant .55 c = .100 Field of Study Not Significant .50 c = .095 Critical Value: Alpha = .05; 2 x 3, )3 = 5.99; 2 x 2, x2 = 3.84. The null hypothesis was tested for the "type of college" and the "cost of college" variables. The third variable, "extracurricular activities" was perfectly matched and could not be tested. With the critical value X2 = 7.81, d.f. = 3, alpha = .05, the null hypothesis was not rejected for the "cost of college" variable. For the "type of college" variable the null hypothesis was rejected. The combination of the match condition and the im- portance rank of the "type of college" variable affects the application state. It is difficult, however, to determine the direction of the association. Of those applying, 16 out of 20, whether matched or not matched, ranked the "type 93 of college" third or below in importance. Those not applying were approximately equally divided between the two ranges of rank, when the match condition is not considered. The students that were matched, more frequently applied when they ranked the "type of college" variable from third to seventh in importance, than when they ranked it first or second. The lower ranking also produced a greater proportion of applications among the no match group than did the higher ranking. The opposite was true for the non-applied group. Individual Descriptor Match Further analysis was made of the matched condition of preferred college characteristics and the actual charac- teristics of the subject college, on an individual charac- teristic basis. The characteristics associated with the evaluative criteria (1) student body and (2) extracurricular activities, were found to be matched in all cases, and to be of relatively low importance. Therefore, these were not considered in this phase of the analysis. The characteristics associated with the other five evaluative criteria; (1) type of college, (2) location, (3) cost, (4) size, and (5) field of study, were indi- vidually tested based upon the match or no match condition. A test of the null hypothesis of no difference between the applied and non-applied groups was made using 94 chi-square analysis (Appendix: Table A-6). Of the charac- teristics tested, only the "type of college" characteristic showed a significant difference in the matched condition. Those prospective students matching on the characteristic tended to apply, while those not matching tended not to apply. TABLE 4.--Comparative Dependence of Matched Conditions and Application State for Evaluative Criteria Descriptors. (arminmamy ENahxflfiveo on ooo.o~m on ooo.mam on ooo.oaw dose noon ozone .mnma .moxma OHOMOm mEoosH Umumfiwuwm m.usmumm mo coauonfluumwn mmouamonmmll.m mam¢9 107 Vacation Companion Preference.--To determine social and family ties the respondents were asked: If you had the choice, would you prefer taking a vacation trip, with your family, by yourself, or with friends? Testing the combined ACT and SAT group, and within the groups, no differences were found across the applied and non-applied classification, using a 2 x 3 contingency table. When the data were regrouped into a 2 x 2 contingency table using the data dichotomy Family or Self/Friends, and testing the applied group across the ACT and SAT classi- fication, a significant difference was found at alpha = .05 (Appendix: Table A-l9). The ACT group appeared relatively more "family or self" oriented and the SAT group more "friends" oriented. No difference appeared when the non-applied group was tested, using the same methodology. Within the ACT group and the SAT group, using the Family or Self/Friends data dichotomy and testing across the applied and non-applied classification, no significant difference was found at the alpha = .05 level (Appendix: Table A-20). Only the ACT applied group showed a significantly greater orientation toward the "family or self" classi- fication. All other groups tended to be more "friend" or socially oriented. An overall summary of differences in the socioeconomic variables are shown in Table 7. 108 TABLE 7.--Differences in Socioeconomic Variables Within Groups Between the Applied and Non-Applied Classification. prLnamflmlCLmflfifflxnjon:ApkafiVNmrAggumd Sockncammfic vanflnfles Oamflnei ACT SAT Efluuflion: Parents Significant Not Significant Significant Ekoflxzs‘ssflsuns lku:Sfifififfl2mt Shfififflxum. ‘Nm:8kyfifflamm Drums: Panama ‘Nm:Shyfiiiamu: Ikk:snmdfflxm¢ Bbtéfignifikzmt Rasfliame: thisuyfihnnfl. Net Significant th Significant ‘NOt Significant value th Significant Net Significant NOt Significant bbbitflyz Number of.Moves ‘Not Significant th Significant NOt Significant Distance of Mbves NOt Significant Not Significant Not Significant ‘memnmanakmemxn Emfilycm:SeLfi/ a Enkmds lkn:8flyfifiamuz Ikn:Sflyfifiamm; IKELSflyfifiGmfl: Note: Tests were made at the alpha = .05 level of Significance. éA.significant difference was found*within the applied classi- fflxuion,vmentwquiacnxm'UxeACTamd:flflfgnmms. 109 Goods Purchase Pattern To determine the carryover affect of goods purchase patterns to college purchase patterns the respondents were ask to rank four statements from the most accurate to the least accurate. These statements were: A. I usually buy whatever is most conveniently available, so I don't have to spend much time looking around or thinking about it. B. I usually decide exactly what I want to buy, and then I go out and buy it. C. I usually know what I want to buy, but I like to look around before I make the final decision. D. I usually look around a lot, and based upon what is available, I decide which item to buy. Information Purchase Selection Pattern Statement State Decision State Descriptor A Uninformed Decisive Limited Shopping (Convenience) B Informed Decisive Limited Shopping (Fulfillment) C Informed Indecisive Extensive Shopping (Confirmation) D Uninformed Indecisive Extensive Shopping (Informative) Figure 4.--Purchase Pattern Matrix. 110 Figure 4 describes the intended meaning to be associated with each of the four purchase pattern statements, as determined in a pre-test. The ranking given a statement by the respondent was used to classify the individual with reference to his purchase pattern information state and decision state. For example, a person indicating statement "B" most accurately describes his purchase pattern would then be described as normally informed and decisive in his purchasing pattern. The responses given to the purchase pattern question were tested for each of the four statements to determine any differences in the frequency of rank across the applied and non-applied groups. No difference was found for any of the statements at alpha = .05 (Appendix: Table A-Zl). Testing all four statements within the ACT and SAT groups across the applied and non-applied classification revealed no difference. College Information Level The purchase pattern statements were used for classification in association with the college information level of prospective students, before their senior year in high school. Those classified as "informed" had ranked either statement "B" or "C" first. Those classified "un- informed" had ranked either statement "A" or "D" first. 111 The areas of information about colleges are listed below. These were each scaled from 1, very well informed, to 6, very uninformed, by the respondents. Cost of the colleges Fields of study offered Specific majors offered Reputation of the colleges Quality of the students Quality of the faculty Quality of the facilities Social opportunities Recreational opportunities Admittance requirements The respondents were then classified as either "informed" about colleges, i.e., six or more of the varia- bles were rated 3 or below on the rating scale; or "un- informed" about colleges, i.e., five or less of the varia- bles were rated 3 or below on the rating scale. A test was made across the purchase pattern de- scriptors, and a significant difference was found at the alpha = .05 level. Those respondents who had indicated their goods purchase pattern was one of being informed also showed the informed pattern with regard to college information. The opposite was true for the uninformed purchase pattern group. The fact that some respondents 112 TABLE 8.--College Informed Classification, Before Senior Year of High School: Combined Group. Number of Variables More Informed Purchase Pattern than Uninformed about Colleges lst Ranked Descriptors 5 or less 6 or more Total Informed (B or C) 38 (43.6) ' 92 (86.4) 130 Uninformed (A or D) 13 ( 7.4) 9 (14.6) 22 Total 51 101 152 Critical Value: Alpha = .05, d.f. = 1, x2 = 3.84. 2 Calculated X = 7.47. were more informed when purchasing consumer goods appears to carry over to the educational purchasing situation. This same type of analysis was used within the ACT group (the SAT group cell values were too low for a test), and a significant difference was found (Appendix: Table A-22). Using the first ranked purchase pattern descriptor statement, but classifying the respondents as either "de- cisive" (statement A or B) or "indecisive" (statement C or D), a test of the college application pattern was made. This, however, did not reveal any differences. The "de- cisive" group did not apply to any fewer schools than did the "indecisive" group. 113 Information Sources for Goods Purchasing The study also inquired into the degree of importance prospective students placed on selected sources of infor- mation when making a purchase decision. The mean (i) rating and rank order of these sources by mean rating are shown in Table 9. The greatest difference between the ACT and SAT groups, in rank order, was with the "sales peOple" source of information. The value of "advertising" as a source showed less difference between the groups, and ranked higher in importance than did the "sales people" source. These differences, however, could not be tested for statistical significance. Each information source was independently tested across the applied and non-applied classification to deter- mine significant differences (Appendix: Table A-23). No differences were found at the alpha = .05 level. Colleges Visitedg Considered, and Applied One source of information about colleges is the actual exposure to the colleges via a visit. The differ- ence in the "number of colleges visited," before the stu- dent's senior year of high school, was tested for the com- bined ACT and SAT group across the applied and non-applied classification (Appendix: Table A-24). There was no sig- nificant difference found at the alpha = .05 level. Further TABLE 9.--Rank Order and Mean Values of the Importance of 114 Information Sources for Goods Buying Decisions.a Cdmbined Group ACT Group SAT Group Buying Information _' _ _ Sources Used X Rank X Rank X Rank Parents 2.42 1 2.55 l 2.25 1 Product Testing Svc. 2.88 2 2.94 3 2.79 3 Friends 2.89 3 2.99 4 2.75 2 Advertising 2.97 4 2.92 2 3.03 4 Brothers or Sisters 3.12 5 3.16 5 3.07 5 Special Counselors 3.28 6 3.29 6 3.25 7 Government Sources 3.44 7 3.64 9 3.18 6 Teachers 3.52 8 3.63 8 3.39 8 Other Relatives 3.58 9 3.68 10 3.45 9 Sales People 3.68 10 3.57 7 3.82 11 Strangers Familiar with the Item 3.70 11 3.77 11 3.62 10 Note: Ranking is from most important to least important. aNo statistical test was made because of the ordinal nature of the data. 115 tests within the ACT and SAT groups also indicated no significant differences. The same pattern of analysis was used to compare the frequency distribution for "the number of colleges considered" prior to making final applications. No differ- ence was found testing, (1) the combined ACT and SAT group, or (2) within the ACT and SAT groups. In Table 10, the results are given in the test of difference with respect to "the number of colleges to which applications were made.” TABLE 10.--Number of Colleges Applied, Frequency Distribution: Combined Group. Implflxuion Nuder<flf€b£kgesl¥mflflx1 same Chfixfifhxnfion . l 2 3 4eorntne immal Applied 17(26.8) l6(l4.2) 16(10.5) 11( 8.5) 60 NaneApplied 49(39.2) l9(20.8) 10(15.5) 10(12.5) 88 * Total 66 35 26 21 148 Mine: onhgthaxanaqudaNB(mxflyhxytocxe«afllemaorrmuewmme inchxkd.hathe:§m@de. Critical Value: Alpha = .05, d.f. = 3, X2 = 7.82. Calculated x2 = 12.49. These results indicate a significant difference between the applied and non-applied groups at alpha = .05 level. The members of the applied group tended to apply 116 at more colleges than did the non-applied group. This seems very important for the strategy planning of a college. A trade-off appears between recruitment effort directed toward generating more applications and the effort to enroll more of those who have already applied. The greater the number of applications submitted by a prospective student the greater the potential competition for his actual attendance, post-application. Recruitment effort spent on those who have already applied may produce a higher payout than effort directed toward generating more applications. This would depend, however, upon whether the college was a first, second, or third preference among those applying. The within group tests indicated a difference in the ACT group at alpha = .05 level, with the applied group making more applications (Appendix: Table A-25). No difference was found in the SAT group. The difference in the behavior of the applied and non-applied groups within the two segments (ACT and SAT) suggests a need for different approaches to the recruitment of these segments. Decision to Attend College Another aspect of the buying process is the initial decision to make a purchase. In the college choice process, this would be the initial decision "to go to college." The actual choice of a specific college would come later. To examine the timing of the decision "to go to college" 117 respondents were asked to indicate when they made this decision. These responses are shown in Table 11. TABLE ll.--Time the Decision to Attend College was Made: Frequency and Percentage Distribution. ACT . SAT Combined High School Time Reference (f) (%) (f) (8) (f) (%) Before Sophomore 57 64 41 61 98 63 Sophomore Year 6 7 10 15 16 10 Junior Year 12 13 ll 16 23 15 Senior Year 8 9 4 6 12 8 After Graduation 2 2 l l 3 2 Not Yet 4 4 0 0 4 2 Total 89 100 67 100 156 100 To test for differences in the timing patterns, the combined group and the individual ACT and SAT groups were analyzed across the applied and non-applied classification. No differences were found. No significant difference was found when the data were grouped into the Before Sophomore/Sophomore or After classification and the applied and non-applied groups were individually tested across the ACT and SAT segments (Ap- pendix: Table A-26). 118 College Information Level: Factor Evaluation The association between the level of college infor- mation and the timing of the decision to go to college was evaluated. The dichotomy Before Sophomore year (early deciders) and Sophomore year or After (late deciders) was again used to classify the respondents. They were asked to recall their level of information about ten college factors, before their senior year in high school. Each factor was tested separately for differences. Only one factor, "social opportunities,” produced a difference. The late deciders were significantly more informed about college "social opportunities" than were the early deciders (Appendix: Table A-27). The factors: (1) cost of college, (2) fields of study, (3) specific majors, (4) reputation of colleges, (5) quality of students, (6) quality of faculty, (7) quality of facilities, (8) recreational opportunity, and (9) admit- tance requirements were all tested at alpha = .05 and indicated no difference. These same factors were then tested across the applied and non-applied classification. Differences were indicated for (1) social opportunities, and (2) fields of study, when the combined group was tested at the alpha = .05 level of significance (Appendix: Table A-28 and Table A-29). 119 Within group tests on the ACT and SAT groups across the applied and non-applied classification produced these results: 1. Social opportunities factor: ACT group no difference at alpha = .05. SAT group a difference at alpha = .05. The SAT applied group was more informed about college "social opportunities" than was the SAT non-applied group. 2. Fields of study factor: ACT group a difference at alpha = .05. SAT group no difference at alpha = .05. The ACT applied group was more informed about "fields of study" than was the ACT non-applied group. College Information Level The reasons for being more informed about one specific college and the degree of usefulness of various college information sources were examined. Respondents were asked to indicate the name of the college about which they were most informed, before their senior year of high school; and explain why they were most informed about that college. These responses are summarized in Table 12. The array of reasons given for being more informed about a certain college can be further summarized into five major categories: 120 TABLE 12.--Reasons Given for Being Most Informed About One College, Before Senior Year of High School: Combined Group. % of Total Classification of Reasons Given Responses A. Materials and information sent by college 16 B. Brother or Sister attended or now attending 14 C. Location of college, near home 11 D. Friends attended or now attending 9 E. Had Visited the college 8 . I requested information from the college 7 F G. College representatives provided the information 5 H . High School Teachers or Counselors 4 I. Preferred or more interested in the college 4 J. Unspecified relatives attended or now attending 3 K. Parent(s) attended 3 L. I researched the school 3 M. Talked with students or graduates from there 3 N. Parents, relatives, or friends knew of the school (but not necessarily attended) 3 0. Attended a conference or meeting there 2 9. Church affiliated college 1 Q. Others (each a single response) 4 100 121 1. Information received from friends and relatives: 32 percent of the responses. 2. Information received from the college and its representatives: 29 percent of the responses. 3. Location of the college relative to the student's home: 11 percent of the responses. 4. Campus visitation and direct campus exposure: 10 percent of the responses. 5. Miscellaneous other reasons: 18 percent of the responses. Usefulness of College Information Sources Additional analysis of the college information flow process involved asking the respondents to evaluate the usefulness of selected college information sources. Each source was tested across the, (1) Before Sophomore (early deciders)/Sophomore or After (later deciders), and (2) Applied/Non-Applied classifications. The results of these tests are given in Tables 13 and 14. The ACT and SAT market segment groups were tested separately. A difference in the usefulness of "high school classmates" as a source of information was found in both the ACT and SAT groups. The difference reflected a lower degree of usefulness among those deciding early to go to college, than those deciding late. A difference in the "high school classmates" as a source of information also existed within the ACT group 122 TABLE 13.--Degree of Usefulness of College Information Sources: ACT Group. Classifications Tested Across Applied/ Non-Applied Before Sophomore/ Sources Used Sophomore or After Father Mother Other Family Members Friends in College High School Classmates College Counselors Other College Representatives Radio Television Newspaper College Provided Material College Visits High School Teachers High School Counselors No difference No difference No difference No difference Difference, alpha=.01 No difference No difference No difference No difference No difference No difference No difference No difference No difference No difference No difference No difference No difference Difference, alpha=.05 No difference No difference No difference No difference No difference No difference Difference, alpha=.05 No difference No difference Note: All "No difference" findings were at the alpha = .05 level of significance. 123 TABLE l4.--Degree of Usefulness of College Information Sources: SAT Group. Sources Used Classifications Tested Across Before Sophomore/ Sophomore or After Applied/ Non-Applied Father Mother Other Family Members Friends in College High School Classmates College Counselors Other College Representatives Radio Television Newspaper College Provided Materials College Visits High School Teachers High School Counselors No difference No difference No difference No difference Difference, alpha=.01 No difference No difference No difference No difference No difference No difference No difference No difference No difference NO NO NO NC No NO NO NO NO NO NO NO NO NO difference difference difference difference difference difference difference difference difference difference difference difference difference difference Note: level of significance. All "No difference" findings were at the alpha = .05 124 across the applied and non-applied classification, but the direction of difference was not well defined. No differ- ence was found within the SAT group. "College visits" as a source of information differed within the ACT group, and followed a pattern of greater usefulness among the applied group than the non-applied group. This may reflect the increased use of planned col- lege visits in the recruitment programs of many private colleges. Specific Collgge and Major Intentions An analysis was made of the intention statements given by the prospective students with regard to their specific college choice and major field of study. The strength of their intentions was measured by a probability statement. Almost all of the respondents were found to be firmly committed to a specific college, and there was no significant difference in the frequency of attendance probabilities within any group across the applied and non- applied classification (Appendix: Table A-30). The strength of commitment to a specific major was generally less, but no significant difference was found in the pattern of commitment between the applied and non- applied groups within the combined ACT and SAT group (Appendix: Table A-31). 125 The tests within the ACT and SAT groups did indicate a significant difference in the SAT group, but no differ- ence in the ACT group. The applied SAT group indicated moderate commitment to a major, while the non-applied group polarized, i.e., either strongly committed or weakly com- mitted to a major (Appendix: Table A-32). Matched Condition Analypis A matched condition system was used to classify respondents. The matched condition was determined by comparing (1) the college about which the respondent was most informed (prior to the senior year of high school) and (2) the college where he intended to enroll. If the two conditions were the same (i.e., the same college), the respondent was included in the "Matched" group; if they were not the same, he was included in the "Not Matched" group. College and majpr intentions.--The Matched/Not Matched classification was then used to test the difference in the prospective students' (1) college intention state- ments and (2) college major intention statements. Separate tests were made within the combined group, and within the ACT and SAT groups. No significant difference was found in the frequency distribution of intentions (by probability range) for either college intentions or college major intentions. Early and late deciders.--The Matched/Not Matched classification was also used to test for differences against 126 the Before Sophomore/Sophomore or After dichotomy of when the decision was first made to go to college (Appendix: Table A-33). For the combined ACT and SAT group a significant difference was found at the alpha = .05 level. Those that were not matched on the "college most informed" and the "college most likely to attend" dimensions were more fre- quently those deciding to go to college before their sopho- more year, early deciders. A greater proportion of those matched were late deciders. When testing within the combined ACT and SAT applied group, a significant difference was found. The pattern of difference was again that the early deciders were not matched, while the late deciders were matched more frequently than expected. The pattern was even more pronounced than with the overall combined group (Appendix: Table A-34). A test of the non-applied group resulted in no significant difference. The same methodology (Matched/Not Matched) was used to determine differences for the variable, "number of col- leges considered" prior to applying. A significant differ- ence for the ACT and SAT combined group was found at the alpha = .05 level. Those expecting to attend the same college about which they were most informed, before their senior year in high school, tended to consider fewer col- leges than did those not matched (Appendix: Table A-35). 127 Testing the "number of colleges considered" within the ACT and SAT groups, no difference was found at alpha = .05 for the ACT group; a difference was found at alpha = .05 for the SAT group. Testing the same variable within the applied and non-applied groups of the combined ACT and SAT group; (1) the applied group showed no difference at alpha = .05, (2) the non-applied group showed a difference at alpha = .05. The pattern of difference in all cases followed the general pattern found in the combined ACT and SAT case for the variable, i.e., those prospective students who were matched considered fewer colleges than did those who were not matched. The variable, "number of applications" made to col- leges was tested in the same manner, but produced no sig- nificant differences. Apparently the matched group, although not considering as many colleges as the not matched group (variation in the size of the evoked sets) still considered it important to apply to several colleges. The not matched group tended to have a larger evoked set than did the matched group; but the end result of the evoked set reduction process, in terms of the number of applications made, was the same for both groups. A summary of the tests made using the matched condition system is given in Table 15. 128 TABLE 15.--Differences in Selected Variables Within Groups Between the Matched/Not Matched Classification. Difference Across Matched/Not Matched Classificationa Variable and Group Tested (Alpha = .05) Early Deciders/Late Deciders: Combined ACT and SAT (Applied and Non-Applied) Significant Combined ACT and Sat Applied Only Significant Non-Applied Only Not Significant Number of Colleges Considered: Combined ACT and SAT (Applied and Non-Applied) Significant SAT Only (Applied and Non-Applied) Significant ACT Only (Applied and Non-Applied) Not Significant Combined ACT and SAT Applied Only Not Significant Non-Applied Only Significant Number of College Applications: All Groups Not Significant College Intention Probability: All Groups Not Significant Major Intention Probability: All Groups Not Significant aA matched classification is where the prospective student intends to enroll in the college about which he was most informed, before his senior year in high school. CHAPTER VI POST-ENROLLMENT PERIOD ANALYSIS WITH PRIOR PERIOD REFERENCE Post-Enrollment Data Collection and Analysis The third time period (t3) in the longitudinal time reference of the study is the post-enrollment period. This period is defined as the period following the actual enroll- ment (or non-enrollment) of the student in college, Fall, 1974. At this point in time the college choice process is complete, and the actual college purchase decision executed. The data received in this period, however, may reflect to some degree post purchase evaluation since some experience with the product (college) had been attained. In time period t2 a mailing was made to 357 pro- spective college students, all of whom had indicated an initial interest in a specific college. From this sample, 159 completed questionnaires were returned. All question- naires were coded allowing a follow-up questionnaire mailing to all those having responded. The follow-up questionnaire was designed to collect the needed post-enrollment period data for across time analysis. The mailing was made September 24, 1974. It was assumed all prospective students would be enrolled in their 129 130 chosen colleges at that time. Of the 159 questionnaires mailed, 121 were returned as of the last of October, 1974 (the cutoff date). This was a 76 percent return rate. The return rate of the ACT segment was, 67 of 91 questionnaires (73.6 percent). The return rate of the SAT segment was 54 of 68 questionnaires (79.4 percent). TABLE 16.--Post-Enrollment (Follow-up) Response Distribution. ACI‘ SAT N = 121 Applied Non-Applied Applied Non-Applied Not Enrolled College 2 A 4 1 6 Enrolled College 23 38 21 26 Total 25 42 22 32 Those not enrolling in any college totaled 13 (10.7 percent) of the respondents; while those enrolling in some college totaled 108 (89.3 percent) of the respondents. These sub-classifications of respondents provide the basic units for comparative analysis in this chapter. Post-Enrollment Data Analysis Methodology In this chapter three statistical methods; (1) chi- square, (2) Kendall coefficient of concordance, and (3) Spearman rank correlation were used to make group comparisons within and across time periods, and to compare individual's responses across time. 131 Chi-square analysis was used to test the null hypothe- sis of no difference between groups in their reSponse fre- quency distributions. In these tests an alpha level of .05 was set as the basis for critical value determination for minimum rejection. The Kendall coefficient of concordance (W) was used to determine the degree of agreement among respondents within a class with respect to a set of college evaluative criteria. The significance of agreement was tested using a chi-square statistic and critical value. The Spearman rank correlation coefficient (rs) was used to determine agreement between groups with respect to their rank ordering of college evaluative criteria at a point in time. Across time comparisons were also made using the rS coefficient for both groups of individuals and single individuals. This allowed the degree of change over time to be tested for significance, and provided a basis for further classification into correlated and not correlated groups. Longitudinal Analysis: Individual Within the ACT group, analysis of evaluative criteria change across tl (pre-application), t2 (post-application) and t (post-enrollment) periods was possible. The evaluative 3 criteria were considered relevant to the purchase (choice) decision of prospective college students. In period t1, 132 data were available on the rank order of importance of the following seven evaluative criteria; (1) size of college, (2) cost of college, (3) type of college, (4) student body composition, (5) location of college, (6) extracurricular and t 2 3 the same seven criteria, plus (8) specific major, were activities, and (7) field of study. In periods t ranked by the respondents. This allowed a time period comparison of the degree of correlation by individual respondent, using t3 as a common time period. Only those respondents having ranked the criteria in all three periods were used in the analysis. This constraint reduced the sample size to 30. To be classified as correlated, the respondent's calculated value of rS was equal to or greater than a criti- cal value of rS at the alpha = .05 level. Where rS was below the .05 alpha level, the individual was classified as not correlated. The absolute number of prospective students cor- related at t2 - t3 was greater than at tl - t3. The direction of this change in the number of correlated students suggests that the consistency of relative importance of the evaluative criteria increases for some students over the decision period. However, testing the hypothesis of no difference in the fre- quency of students correlated at t1 - t3 and t2 - t3, produced no statistically significant difference at the alpha = .05 level. Neither were significantly more students 133 - t or t - t correlated than not correlated at either t1 3 2 3 (Appendix: Table A-36). These findings tend to support the general hypothe- sis that the evaluative criteria used in making the college (buying) choice vary in importance as the prospective student moves through the buying process. College choice is not merely a matter of comparing institutions against a firm set of evaluative criteria, but considerable reevaluation of the importance of the evaluative criteria also takes place. This appears consistent with consumer buying be- havior where the buyer lacks prior purchase experience, as is the situation with a prospective freshman college student and the purchase (selection) of a college. Further analysis within the 30 individual cases revealed a varying pattern of consistency in the relative importance of the evaluative criteria. Nine of the 30 cases (30 percent) were not correlated in t1 - t3 or t2 - t3. This indicates instability of the criteria over time. Ten of the 30 cases (33 percent) showed stability in the evalu- ative criteria over time. These 10 cases indicated basically the same degree of importance in the evaluative criteria from pre-application (t1) through post-application (t2), to post-enrollment (t3). Nine of the 30 cases (30 percent) reflected a mixed pattern of stability, with insignificant correlation, t1 - t3, but significant correlation, t2 - t3. This pattern suggests the reevaluation of criteria up to 134 the time the students decided upon the college they would most likely attend. Almost all of the respondents in t2 were 90 percent or more certain about the college they were going to attend. The remaining 2 of the 30 cases were correlated at t - t3, but were not correlated at t - t . They tended 2 3 period importance pattern after they 1 to revert to the t 1 had actually enrolled. In both cases only one application was made. In summary, only 30 percent of the ACT cases were stable on the relative importance of the evaluative criteria through the purchase period (t1, t2, and t3). This further supports the general hypothesis of evaluative criteria instability over time among those making the college pur- chase decision. Combined ACT and SAT groups.--For period t2 - t3 both ACT and SAT responses to the rank order of importance of evaluative criteria were analyzed. Nine additional responses from the ACT group were available, since the constraint of responding in both time periods (t1 and t2) could be dropped. A test of difference between the ACT and SAT groups was made using chi-square. No significant difference (alpha = .05) was found in the frequency of correlation at t2 - t3 for the two groups. 135 TABLE 17.--Evaluative Criteria: Degree Correlated t2 - t3. Correlated ' Not Correlated ACT 23 of 39 59.0% 16 of 39 41.0% SAT 23 of 35 65.7% 12 of 35 34.3% Total 46 of 74 62.2% 28 of 74 37.8% Applied and non-applied.--A test of difference was made using the applied and non-applied classification, and the correlated and not correlated classification at t1 - t3. The distribution indicated a significant difference in the applied and non-applied groups within the ACT segment at t1 - t3. Significantly fewer respondents in the applied group were correlated (i.e., consistent ordering of the evaluative criteria) (Appendix: Table A-37). When the same groups were tested correlating t2 - t3 evaluative criteria statements, there was no significant difference. This indicates the applied group between t1 and t reevaluated the criteria to produce more correlations 2 with the t structure. The distribution of correlations 3 across the applied and non-applied classification was exactly proportional (actual value equal to expected cell values in all cells) for the t2 - t3 comparison. (This does not imply the applied and non-applied ranked the evaluative criteria the same.) 136 Match Condition and Correlation: Combined ACT and SAT In the post-application period (t2), respondents were asked to list the college about which they were most informed, before their senior year in high school. They were also asked to list the college they would most likely attend. These two statements were then compared to produce either a (l) matched (same college), or (2) a not matched (not same college) condition. The matching was done to associate the highest level of information with the enroll- ment intention to determine consistency. Using the classification correlated and not corre- lated for evaluative criteria, t2 - t3, a test was made across the Matched and Not Matched classification (Appendix: Table A-38). Of the combined ACT and SAT group the corre- lated group was more frequently associated with the matched condition, while the not correlated group was more frequently associated with the not matched condition. These differences were significant at the alpha = .05 level. The same type comparison using the t1 - t3 correlation condition could not be tested because of the small sample size (one cell value below five). Descriptively stated, 9 of 18 matched respondents were correlated, while 1 of 9 not matched respondents were correlated. Those students with a more consistent set of evalu- ative criteria (correlated group) tended to have a higher .llll'l'lI" 137 early level of information about the college they planned to attend, than they had about other colleges. Early and late deciders.--The correlation state of the individual respondent, t - t3, was tested across the 2 Before Sophomore (early decider) and Sophomore or After (late decider) classification. (The reference is to when they first decided they would attend college.) Though more of the early deciders were classified as correlated, there was no significant associative relationship between when the decision to attend college was made and consistency of the relative importance of the evaluative criteria (Appendix: Table A-39). College decision--matched condition.--The combined classification of a matched (same) state, and the correlated and not correlated state tested across the Before Sophomore/ Sophomore or After dichotomy produced a significant differ- ence at alpha = .05 (Appendix: Table A-40). These findings indicate that those students planning (t2) to attend the college about which they were most in- formed, prior to their senior year in high school, were generally more stable in their ranking of the eight evalu- ative criteria from t2 to t3. The matched and correlated set of students also had decided earlier that they were going to attend college (before their sophomore year) than did the matched but not correlated group. The early decision to go to college appears to be much more associated with those who 138 had developed a more stable evaluative structure, and who had acquired the necessary level of purchase information to make the college purchase (choice) decision. Number of applications.--The significantly correlated (t2 - t3) students tended to apply to fewer colleges than the not correlated students. This difference was statis- tically significant for the combined ACT and SAT group at alpha = .05 (Appendix: Table A-4l). The correlated group with the more stable set of evaluative criteria appeared more decisive and made fewer college applications. Those in the correlated group ap- peared to know more about what they want in a college and were able to screen their evoked set to fewer applied colleges. They appeared to be further along in their actual choice of a particular college at t2 than the not correlated group. As with the purchase of consumer goods, knowing what to consider in a product (college) allows greater decisiveness during the buying process. A pattern of difference was not found, however, in the number of colleges considered (Appendix: Table A-42). The almost complete lack of difference in the number of colleges considered between the correlated and not correlated groups further supports the importance of the association between a stable set of evaluative criteria and the buying behavior process. The process of reducing the considered set (evoked set) of colleges to an applied set, and then to 139 the actual college choice was executed differently by the correlated and not correlated groups. Ability variable.--Ability of the prospective student was considered as a variable which might influence the degree of evaluative criteria stability. Testing the null hypothe- sis (no difference), a rejection was not possible (Appendix: Table A-43). The ability variable has no apparent association with the correlated or not correlated state of the students. The higher ability students did not have a more consistent or stable set of evaluative criteria than did the lower ability students. Type of college.--Whether a student was enrolled in a public or private college was used to test the "type of college" association with the correlated state. No signifi- cant difference was found for the combined group when the 2 - t3 correlation was used. The test was made at the alpha = .05 level, with a critical value of X2 = 3.84. The calculated X2 = 1.03. t When the ACT group was used and the comparison made with t - t data, a significant difference was found (Ap- l 3 pendix: Table A-44). The t1 - t3 correlated condition was significantly more associated with the public college student. The t2 - t3 correlated condition did not indicate such a difference. The greater degree of change was clearly in the private college student category. This may reflect a 140 greater responsiveness to the recruitment effort directed toward these students. Individual college level.--A test was made to deter- mine if the correlated condition would discriminate between those who applied to a particular college and later enrolled, and those who applied and did not enroll. No significant difference was found using the t2 - t3 correlations and testing at the alpha = .05 level. Evaluative Criteria t2: Scaled Values The previous portion of this chapter considered the rank order of evaluative criteria as given by respondents at t and t . The Spearman coefficient of correla- 1' t2' 3 tion (rs) was used to determine the degree of correlation of an individual respondent's ranking of the evaluative criteria across time. In the ACT segment for which t1 rankings were available, correlation was between t1 and t rankings for each individual. Respondents in both the 3 ACT and SAT groups were used for the correlation analysis in periods t2 and t3, with the rs statistic calculated on an individual basis. This analysis used rank order data and no comparison of individual criteria was made. In this section, scale data were used to compare individual evaluative criteria across groups of respondents. The scale used was from 1, very important, to 6, very un- important. These data were taken in the post-application 141 period (t2) and the post-enrollment period (t3). Scale data from the pre-application period (t1) were not available. The analysis with time considered was across three major respondent classifications, (1) applied and non-applied, (2) private college enrollee and public college enrollee, and (3) enrolled and not enrolled. Classifications (l) and (3) were in reference to the respondent's behavior associ- ated with one specific college (the college cooperating with this study). The following eight evaluative criteria were used (these are the same as in the first part of the chapter): (1) size of college, (2) cost of college, (3) type of col- lege, (4) student body composition, (5) location of college, (6) field of study, (7) extracurricular activities, and (8) specific major. Applied and non-applied (t2).--This classification was used for a chi-square analysis of the frequency distri- bution of responses by scale value across two groups. The format of analysis is illustrated in Appendix Table A-45, for the "cost of college" criterion. Cells were grouped where necessary to meet the requirements of chi-square. All eight of the evaluative criteria were tested at the alpha = .05 level. In all eight cases no significant difference was found between the applied and non-applied groups. Both sets of respondents tended to evaluate the 142 importance of the individual criteria the same in the post- application (t2) period. Private and public collegps (t2).--A classification system based upon whether the respondent had enrolled in a public or private college (as reported at t3) was used to determine differences. None of the eight evaluative criteria appeared as significantly different in their scaled importance across the two groups at alpha = .05. Within gpplied group: enrolled and not enrolled (t2).-- The within the applied group analysis refers to one specific (private) college. Three of the evaluative criteria appeared significantly different; (1) type of college, (2) field of study, and (3) extracurricular activities (Appendix: Tables A-46, A-47, and A-48). The enrolled group placed significantly more emphasis upon the importance of the "type of college" than did the not enrolled group. The enrolled group placed significantly less emphasis upon the importance of the "field of study," how- ever, it was generally considered to be important. The enrolled group placed significantly less emphasis upon the importance of "extracurricular activities" than did the not enrolled group. All three differences were highly signifi- cant, at either the alpha = .01 or alpha = .02 level. The value of knowing these attitudinal differences in advance, to a college, would be an improved prediction of Fall enrollment from the applied set. 143 Evaluative Criteria t3: Scaled Values The same pattern of analysis was used for the post- enrollment (t3) data on the importance of the eight evalu- ative criteria. The scale used was the same as in t2, i.e., 1, very important, to 6, very unimportant. Applied and non-epplied (t3).--In period t no 2 difference was found between the applied and non-applied groups. In t however, one of the eight evaluative cri- 3: teria, "location of the college,“ was significantly differ- ent (Appendix: Table A-49). The applied group was significantly less concerned with the importance of the "location of the college" than was the non-applied group. The lower degree of importance attached to the "location of the college" allows a wider geographical range of choice, which places more colleges in competition for these prospective students. The smaller degree of locational constraint seems consistent with the t2 period finding that the applied group had made application to significantly more colleges than the non-applied group. Private and public colleges (t3).--In period t2 none of the eight evaluative criteria were found to be sig- nificantly different in importance across the private and public college groups. In this period (t3) two evaluative criteria, (1) size of college, and (2) cost of college were found to be significantly different at alpha = .05 (Appendix: Tables A-50, A-51). 144 The "size of college" criterion appeared signifi- cantly more important to the private college enrollee than it did to the public college enrollee. The Opposite pattern was observed with the "cost of college" criterion. The public college enrollee considered cost to be significantly more important than did the private college enrollee. These are, of course, the two most obvious differences between most private and public colleges. The difference in the importance of the two criteria was not evident at t2, but tended to emerge between the post- application (t2) and post-enrollment (t3) periods. These two criteria may have weighed heavily in the final choice of a college where applications had been made to both public and private colleges. Within applied_group: enrolled and not enrolled (t3).--Of those respondents who had applied to the subject college, the enrolled and not enrolled groups differed in t2 on, (1) type of college, (2) field of study, and (3) extracurricular activities. In period t the two groups 3 differed on the importance of the "student body composition" (Appendix: Table A-52). No difference was found in the other seven variables. Again there is an apparent shift in the assessment of the degree of importance of evaluative criteria across time. The "student body composition" criterion was sig- nificantly less important for the enrolled group than the 145 not enrolled group. The change from period t2 was for the criterion to become more important for the not enrolled group and less important for the enrolled group. The ”type of college" criterion became less important for the enrolled group from t2 to t3. The ”field of study" criterion became more important for the enrolled group and less important for the not enrolled group. The "extra- curricular activities" criterion became less important to the not enrolled group, with little change in the enrolled group. Change in importance of the evaluative criteria was to t within evident with four of the eight criteria from t2 3 the applied group. Evaluative Criteria: Within Group Association of Rank Order In the first part of this chapter the Spearman rank correlation coefficient (rs) was used to determine the association of two rankings of the evaluative criteria by an individual respondent across time. This allowed a test of significance at alpha = .05, and permitted a classi- fication of the individual respondent as either correlated or not correlated. In this section the ranking of the evaluative cri- teria by each individual respondent in a defined grouping of respondents was used to calculate a measure of the relation among the several rankings. This was accomplished using the Kendall coefficient of concordance W. The value 146 of the coefficient of concordance W was then tested where N (the number of evaluative criteria) was larger than 7 using a chi-square statistic and critical value.1 The rejection of the no difference hypothesis was set at an alpha = .05 level. For example, the individual rankings of the evaluative criteria of all respondents in the enrolled group were placed in a k x N matrix. The number of individuals ranking the eight (N) evaluative criteria is equal to k. The W coefficient was determined and converted into a chi-square (X2) value. X2 was then tested at an alpha = .05, d.f. = N-l, critical value of X2. If a statistically significant relationship was found in the ranking given by the individuals, it was possible to determine the ordering of the evaluative criteria by using the order of summed ranks of the individual criteria.2 The lowest summed value would indicate the criterion which was most important of the set of evaluative criteria, and so on, through the order. After having determined an order of importance for a defined group (e.g., enrolled respondents) a measure of association was determined, (1) for the same group at two points in time, or (2) for different groups at the same lSidney Siegel, Nonparametric Statistics for the Behavioral Sciences (New York: McGraw-Hill’Book Company, 1956), p. 236. 2 Ibid., p. 238. 147 point in time. The Spearman rank correlation coefficient (rs) was used as the measure of association. Tests of significance were then made. Coefficient of concordance W analysis.--The group classifications tested with the chi-square statistic based upon the calculated value of W were all found to have a sig- nificant relationship (similarity of ranking by the member respondents). The specific groups tested, the X2 and W values, and the alpha levels of significance are shown in Tables 18 and 19. For each of the group classifications at time periods t2 and t3 shown in Tables 18 and 19, a rank order array of the eight evaluative criteria was determined from the concordance matrix. Comparative group analysis: applied group.--The applied group was divided into enrolled and not enrolled groups. As shown in Table 20, the between group and across time comparisons indicated a significant degree of corre- lation. The direction of change in the enrolled and not enrolled groups from t2 to t3 was from very highly corre- lated (rs = .923) to less highly correlated (rS = .789). The degree of correlation was relatively greater in the not enrolled group, t2 - t3, than in the enrolled group, t -t. 2 3 A within time period comparison of the ACT and SAT enrolled and not enrolled groups was made at t2. This is shown in Table 21. 148 a He. em.m~ are. He. oe.- mme. m mace saw ea Ho. mm.ae mum. Hoe. NH.Hm emu. a Sago you he see. ee.em com. Hoe. me.ae mom. ea omeeonom uoz a He. mm.o~ mmm. Hoe. H~.em mee. «a Sago Saw a He. eH.o~ mam. Ho. mm.ae mam. a memo you we Hoe. HH.Ne cam. Hoe. H~.me mom. Hm omeeonom when .auaomam moem> moen> .aueooam moen> moau> when coaucoamauuneo x no H0>0q Nx 3 no e0>mq Nx 3 x ooono madam mndad Amuv ucoaaaousmlumom Amuv sofiuooaammflumsam>m mo Hmouo xcmm Ga muwmsmmoeom macho cflnuwz mo mocoowmwcmflmnl.ma mnmde 149 we see. cm.a~ one. see. cm.m~ can. me Sago emm Hm Hoe. mv.N¢ mam. Hoo. Nm.hm Hmw. mH NHGO 80¢ mm Hoo. HQ.Oh hem. Hoo. HN.Nm, 5mm. NM OHHQDQ ca Ho. mo.- can. He. mo.e~ eem. ca Sago sew m Ho. mh.mm hhm. Hoo. 0N.mN Gov. m hHGO BUG ma Hoo. NH.N¢ ham. Hoo. mv.N¢ mHm. mH mum>flum when .auecoam moec> moec> .auecoem moac> moem> gnaw oceanoameunceo x no H0>0A x 3 no em>mq x 3 x ozone mamad m mamas N Amuv ucoEHHOHCMIumom Amuv sowuoowammmlumom Am n.zv.©mwammdlcoz .mu can my .msouw cowammdlsoz on» sflnuwa .owansm cam oum>wum "puma mumswmlwso o mnema .3 monocuoosoo mo usmwowmmmoo on» no woman mflumuwuu m>wumsam>m mo HOUHO xcmm cw huwmcmmoeom msouw :Hnuflz mo monoowmecmwmll.ma mqmda 150 .mmm. u mu .Ho. u mnmflfi umwm. H mm .mo. 0 MSQH< ”mu MO mOSHm> HMOflUflHU ..mo. 0 madam um Dancemwcmwm .mmh. u mu «mu .omaaoucm #02 can omaaoucm Rev .Ho. 0 mamas um unmowmasmwm .mmm. u my «Nu .omaaousm 902 can oueeoncm Ame .me. u madam uc neuoaoaooan .ece. u mu .muuau oueeouom uoz Ame .mo. u mamas um usmowmwsmwm .mnm. n ma “mulmu .pmaaoucm Adv ”musowowmmmoo cowumHmHHoo smasofluusomuuxm smasofiunsomuuxm hoom unmosum Hmasowunsomuuxm m upon atom unoccum sowumooq\ucoosum Hmasowuusomuuxm moom ucoooum 5 atom mama unocoum\coflumooq umoo coaumooq m mNflm Guam coaumooq\omma muflm m cowumooq umou omme\cowumooq Homo: owmwommm e unoo mama onem unoo m Hons: owmwommm Mono: camaowmm Mono: owmwommm mama m pamflm pamwm oaoem caowm a lane lane lane lane nacho usosaaousmnuwom sowumowammfilumom ucoEHHousmlumom cowumowammflumaam>m on» no moamuHomEH mo Hmcuo xcmmll.om mqmde 151 .mmm. u an .He. u ceded “nee. u an .me. u cedeg "mu mo nmoac> ecoauano .Ho. 0 mamao um vaguenecowm .mvm. H mm “mu .vwaaoucm uoz can coeeoncm "eon lee .mo. u madam no nomoaeacoen .mee. u nu “Nu .omeeonom uoz com oweeonom "Boa Ame .Ho. u named on domoemeomeu .mom. u we “mml.©maaoucm uoz Bdm 6cm pmaaoucm uoz 90¢ Amy .mo. 0 mamas .HGMOHMflcmwm no: .emm. u we “Nu .omHHonom saw one omeaonom sue 1H1 “nuomaoemumoo oceunemnnoo Hoasowuusomuuxm HMHsOAHHsooHuNM “masowuusomnuxm atom ucoosum m coauoooq zoom ucmcoum moon nomoonm Home: oamaooom a zoom ucoosum sowuoooq ouwm\sowumooq “masowuusowuuxm m Guam onam cowumooq\mufim onwm m mama umou umou umoo v umou mums HOnoz owmwommm sowumooq m Hens: Camem\0flmeommm HOnmz vamwommm meme mama m Homo: owmwoomm\caowm oaowm pamwm oaowm H ooaaoncm uoz ooaaoucm Umaaonsm uoz omaaousm Hwouo Away adm Amuv sum muonw omflammd .mu .ooono ooeeoma 0:» Cannes .cmeaonom uoz can owaaoncm .Bdm can 90¢ "mammamsé 3 mocmouoocoo mo usowowmmooo mg“ no ”mmmm MHHwHflHU m>flflmsflm>m 05“ MO mUCMUHOQEH HO HmUHO vmcmmlloHN Wflmfie 152 At t2 the ACT enrolled and the SAT enrolled groups were not significantly correlated, indicating a difference in the importance ranking of the evaluative criteria. Differences in the importance of "location" and "specific major" were most evident. The ACT not enrolled and the SAT not enrolled groups were significantly alike in their ranking of the evaluative criteria. Overall, the SAT group was more in agreement on the evaluative criteria ranking than was the ACT group. At t3, as shown in Table 22, the ACT enrolled and the SAT enrolled groups were very highly correlated, rS = .929. The groups had not been significantly correlated at t2. From t2 to t3 the ACT enrolled and not enrolled groups moved to a much higher degree of agreement on the ranking (rs = .673 to r8 = .929). The SAT enrolled and not enrolled groups moved from a high degree of agreement to a lesser degree of agreement (rS = .946 to rs = .738). Comperative group analysis: non-applied group.--The non-applied group was divided into those enrolling in a private college and those enrolling in a public college. Comparisons were then made using the rs coefficient. It was possible to further divide each group by the ACT and SAT segment classification. These comparisons of the evalu- ative criteria rankings are shown in Tables 23, 24, and 25. 153 .mmm. u un .He. ended “mac. n nu .me. u noose "nu «0 umoec> euoauano .mo. 0 mamam us unmowmwcmfim .mmn. u on “mu .omaaoucm #02 can UmHHoucm “aflm on .Ho. 0 mamas no psmowmwcmwm .mmm. n on «mu .omaaoucm #02 new Umaaousm "BU< Amy .mo. 0 madam um unwofimwcmwm .mmn. u on Am» .cmeeoncm noz Sam can omeeonom noz Bog Ame .Ho. u ceded no screened Imam .mmm. u on “mu .anaoucm amm was ooaaoucm 80¢ Adv "wucmwowmmoou COHUMHOHHQU “masowuusomuuxm atom usoosum HMHDOfluusomuuxm moon usoozum m muwm HMHDOAHHDUCHuxM moom unwosuw HMHSOflMHsomnuxm n soom nomoaum unoo dame unoo o wmwe cowuoooq umou mama m cowuoooq mama :oflumooq sowumooq v umoo muwm swam Mono: cameoomm m Homo: newwommm Honmz owmeowmm Homo: owmwowmm mNHm N oawwm Camflm oamwm paoflm H omaaoucm uoz omaaonsm ooaaoncm uoz omaaousm Hmono Amuv 94m Amue Boa maono cowammd .mn .doono omeedoe we» zeroes .omeeondm noz can Omaaoucm .Bdm was 90¢ umflm3HMC4 3 mosmwuoocou mo “cmfiufimmwou 03...— GO GOWMM MfiHOu—HHU 0>fiQM9HM>N OS». MO OOQMHHOQEH MO HOGHO MCGMIlomm Wen—”mama... 154 .mmm. n my .Ho. u onaa< “mom. u mu .mo. u mnaam "Mn mo mmSHm> Hmowuwuu .mo. u mnaam um uCMOflmacmwm no: .maw. H mm “mu .oflansa can muo>anm lee .Ho. u cameo no peooemaooem .mma. n on “mu .oaenom ooo ouo>anm Ame .ec. u ocean no nocoamacoen .Nma. u no amuumu .oaeoom Ame .mc. n osaam um ucmowmwcmwm .mmz. n mu «mplmu .mum>aua Adv “mucmfl0fimmmoo sowumamuuou zoom oomooum noHsoaHHsomuuxM undamauusomnuxm Hoaoowuuooouuxm \umasoauusomuuxm m umasoflunsomnuxm zoom nomooum zoom ucoooum zoom ucoooum \zoom nooooum z mmwm mama umoo\cowumooq newumooq m maaa mmwm soeumooq\umou maha m sowumooq COHHMOOA Mono: Ofimwomam mNHm o umoo Homo: owmwomam mama HOnmz oamwomam m echo: camaoomm uuoo when unoo m oamwa vamwa onHa camea a lane . lane lane lane noono usmEHHOHGMIumoa coaumowaaawua anouw Umflaaamlcoz .mu cam my .asonw omwaaamlsoz may swsuw3 .owansa com muo>wua ”mammamcm 3 mocoouoosou mo usmwowmmmoo map so ommmm mwumufluu m>wumsam>m may no moamuuanH mo Hmouo xcomln.mm mamfle 155 m m .mmm. u H .Ho. n mnaad «mom. H H .mo. u mnaad "mu MO mmDHm> Hmoeueuu .mo. u mnaam um unmowmecmwm mmm .zmo. on an» .0aenom ooo muo>anm “emm Ave .mo. 0 madam um homoemeomeu .eao. u on am» .uaedom one mun>anm ”sum Ame .Ho. u cameo no nonoameooau .mmm. u on Am» .oeanom emm ooo oaanom eom Ame .mo. u women .uoooameomau uom .zmo. H mm «mu .mum>wua emm osm muo>eum eUm AHV "mpcmeoemmmoo sceumamnuou coeumooq Hoaooeuusomuuxm umasownusomuuxm zoom usmooum \zoom ucmosum m zoom unmooum maze zoom ucmosum Hmasoeuusomnuxm \soHumooq 5 zoom ucmosum mmwm\coHumooq maze Hoaooeuusomnuxm m mmem coepmooq\mmem mmem maze m COAHMOOA Homo: memeomam coaumooq “ohms oemwomam o HOnoz memeomam umou Homo: memeomam umoo m umou maze umou mmem N oamea oamwm oamwm oamwm H owHQsa mum>eua Deanna muo>wua Hmono xcmm Ammo emm maul eom asouw omwaaamnsoz .mu .moono omaammmuooz 0:» oeeuaz .oeaodm onm mum>eua .emm ocm eo< “memzamcm 3 mosmouoosou onm unmeowmmmou @5 GO Ummmm MflHmn—HHU O>HHM5HM>W 05“... MO OUGMHHOQEH MO HOGHO viewfinllowm maHmANR. 156 w m .mme. n n .ee. u comma ance. u A .me. 0 cream ”on mo umoeo> Hooauano .mo. 0 mnaao um ucmoemefimwm .moz. u mm mm» .oeansa osm mum>eua "emm Ave .mo. 0 msaao um usmoemesmem no: .zmo. u u «m» .oaanom ooo muo>aum "eom Ame .Ho. u madam on noooemaoman .mmm. u on an» .oamnom emm ooo oeaoom eom Ame .mo. u oemao no udooameooem pom .maw. u on .my .mum>eua emm osm mum>ena eom Adv "mucmeoemmmou cowHMHmHHOU Hmasowuusoouuxm Hoasowunsomuuxm Hoasoeuusoouuxm zoom usmooum m zoom usmooum zoom usmosum zoom unmosum Hoasoeuusomuuxm e maze mmem \Honmz oeweomam maze soflumooq o Mono: maze memeomam\maze mmem umou m mmem coeuoooa umoo coeuoooq \Honmz oemwomam o Homo: Homo: memeomam coHumooq\mNem umoo memeomam\mmwm m umou mmem\s0eumooq Homo: oemwomam maze m onem oamea onem oamea H meansa mum>eua Deanna mum>eua Hmouo xcmm Ammo emm Amue eom anonw omwaaamlsoz .mu .moouo omeemmmuooz one canoes .0aenom ocm mum>eua .emm ocm end “memzamnm 3 moamouoocoo mo unmeoemmmoo ms» no ommmm weumueuu m>eumsam>m may no mosmuuanH mo “mono xsmm11.m~ mamme 157 The across time (t2 - t3) comparison of the private college group indicated a significant correlation of the evaluative criteria, but the correlation was less than for the public college group (rS = .738 and r8 = .952 respec- tively). At t the public and private college group comparison 2 indicated a higher degree of similarity between the groups than was indicated at t3 (rS = .923, significant at t2; rs = .613, not significant at t3). The greater change in the relative importance of the evaluative criteria was with— in the private college group. Tables 24 and 25 show the private and public college groups within the ACT and SAT segments at t and t respec- 2 3 tively. At t the ACT and SAT private college groups were 2 not significantly correlated, while the ACT and SAT public college groups were significantly correlated. Within the ACT segment (t2), the private college and public college groups were significantly correlated; but within the SAT segment (t2), they were not significantly correlated. The greatest difference in the ranking of the evaluative cri- teria was with the SAT private college group. At t the ACT and SAT private college groups were 3 not significantly correlated (the same as in t2). The ACT and SAT public college groups were significantly correlated at t3 (the same as in t2). Thus, the ACT and SAT segments 158 within the private and public college classification showed no significant change over time. The private and public college classification within the ACT segment did show a change from significantly corre- lated to not significantly correlated at t2 and t3 respec- tively. The private and public college classification with- in the SAT segment also changed, but from not significantly Thus, correlated at t to significantly correlated at t 2 3‘ the ACT private and public college students became less alike in their evaluation of the relative importance of the evaluative criteria; while the SAT private and public college students became more alike in their evaluation of the cri- teria over time. Enrolled to private and public comparison.--A final comparison at t2 and t3 was made by correlating the enrolled with the (1) private (non-applied) and (2) public (non- applied) college groups. The enrolled and the private college group were significantly correlated (rs = .851) at t2. The enrolled and the public college group were also significantly corre- Att lated (rS = .738), at t the same pattern was found. 2' 3 Both the private and public college groups were significantly correlated with the enrolled group (rS = .875 and rS = .756 respectively). Those students enrolling at the subject college did not differ significantly from those students enrolling at 159 other private colleges or public colleges in their evalu- ation of the relative importance of the evaluative criteria at either t2 or t3. Selected College Characteristics Data were collected in the post-enrollment (t3) period on the characteristics of the colleges selected by the respondents for enrollment (brand characteristics). These characteristics, (1) size of college, (2) type of college, and (3) cost of college, proved to be significantly different for the applied and non-applied segments of the total respondent group. Testing the combined ACT and SAT segments, the applied group (which included those enrolling at the col- lege under study) tended to attend smaller colleges than did the non-applied group. This difference was found to be significant in the ACT segment only, when separate group tests were made (Appendix: Table A-53). Significantly more of the applied students, of the combined ACT and SAT segments, were attending private col- leges, while more of the non—applied students were attending public colleges. This difference was also significant with- in the separate ACT and SAT segments (Appendix: Table A-54). The "cost of college" within the combined group differed significantly. More of the applied group were attending colleges costing $3,000 or over per year than 160 were the non-applied group. Over fifty percent of the non- applied were paying under $2,000 per year in college costs. The separate ACT and SAT segments had the same significantly different pattern (Appendix: Table A-SS). Student aid characteristics.--Associated with cost is the source of financial aid, if any, used by the student. These sources were compared across the applied and non- applied groups to determine different source usage. The respondents were asked to indicate if they were receiving financial aid: (1) from parents, (2) from college, (3) from other sources, or (4) receiving no financial aid. The findings were: (1) no more of the applied group than the non-applied group were receiving financial aid from their "parents" (alpha = .05), (2) significantly more of the applied group than the non-applied group were receiving financial aid from the "college" (alpha = .01), (3) sig- nificantly more of the non-applied group than the applied group were receiving "no" financial aid (alpha = .05). The category "other sources" of financial aid was not significantly different between the groups (alpha = .05). No difference was found within the applied group across the enrolled or not enrolled classifications for any of the financial aid sources (alpha = .05). CHAPTER VII SUMMARY FINDINGS AND CONCLUSIONS Purposes and Approach of the Study The major purposes of this study were: (1) to provide additional knowledge and understanding of the pro- spective college student's information search process and source usage when making the college choice decision, (2) to identify the importance of selected evaluative criteria used in the choice process, and (3) to identify segmental differences within a group of prospective students indicating interest in a specific college, at selected time reference points and across time. The college choice problem was viewed as a purchase problem not significantly different than the type faced by consumers when purchasing economic goods. A marketing per- spective was used to define and structure the research approach, with emphasis placed on the application of market segmentation, buyer intention, and buyer behavior theory. The research design was longitudinal in nature, with three time periods: Pro-application (t1), Post- application (t2), and Post-enrollment (t3). This design allowed both within and across time analysis of the data for defined student segments. Of particular research 161 162 interest was the change in the importance of selected evaluative criteria of the individual student. These evalu- ative criteria, based upon the previous research done on college choice, were considered major dimensions used in evaluating specific colleges. Pre-Application Period Findings 1. Those prospective students with a "high" number of matches on what they wanted in a college and the actual characteristics of the college tended to apply more frequently. 2. The applied who matched on the descriptor, "type of college" generally ranked the criterion moderate to low in relative importance. The non-applied, who matched, considered the type of college relatively more important than did the non-applied who did not match. 3. Only a matched descriptor condition on the "type of college" was significantly different, with the applied matching more frequently. 4. The applied differed from the non-applied on the importance of the "type of college," "cost of college," and "extracurricular activities." 5. Those students indicating the college was their first choice tended to apply more frequently than those indicating the college as a second or lesser choice. Any rating other than first did not serve to predict application. 6. The best predictor of a prospective student's application was the combination of (a) the college designated 163 as his first choice, and (b) a stated preference for the "type of college" (a descriptor match with the college's characteristic). This combination incorporates two market choice elements. First, the match on "type of college" differenti- ates the private and public college market classification. Second, the first choice designation indicates a preference within the private college market classification. 7. Enrollment was not predicted significantly better by the first choice designation compared to other choice designations, within the applied group. Post-Application Period Findings Socioeconomic Variables 1. Parents of the students applying to the college under study more frequently had had some college than did the parents of those not applying. This difference was most apparent within the SAT segment. More of the SAT applied students' parents had had some college, while more of the ACT applied students' parents had had no college. 2. The educational experience of the brothers and sisters of the ACT segment was different for the applied and the non-applied. The applied had relatively fewer ' brothers or sisters who had graduated from college. There was no difference within the SAT segment. nI-"ll‘l'lillullr 164 In general, the SAT segment appeared to have better educated parents and more brothers and sisters with college background than did the ACT segment. 3. Those prospective students from the ACT segment who applied to the college were more family or self oriented than socially (friend) oriented. This orientation pattern was not found in the ACT non-applied or the SAT segment. 4. No significant difference was found for the variables, (a) income, (b) value of home, and (c) mobility, between the ACT and SAT segments, or within the segments (applied and non-applied). Goods Purchase Pattern 1. No significant difference was found between the ACT and SAT market segments or across the applied and non-applied behavior classifications with respect to their reported economic goods purchase behavior. 2. Those prospective students who were identified as "informed" in their economic goods purchase behavior also tended to be more informed about colleges, before their senior year of high school. The carryover of goods purchase behavior, with reference to the degree informed, was most significant in the ACT market segment. 3. The "decisiveness" dimension of economic goods purchase behavior did not carry over to college choice behavior. 165 4. There was no difference in the importance of the information sources used for economic goods purchases between the ACT and SAT market segments. Number of Colleges Visited, Considered, and Applied 1. There was no significant difference in the number of colleges visited or considered between the ACT and SAT segments, or across the applied and non-applied classifications within either segment. 2. The ACT applied compared with the ACT non- applied made application to a significantly greater number of colleges. There was no difference within the SAT seg- ment 0 Decision To Go To College 1. Over 60 percent of the prospective college students decided they would go to college before their sophomore year in high school (early deciders); and ap- proximately 40 percent decided during or after their sopho- more year in high school (late deciders). 2. No significant difference was found between the ACT and SAT segments, or across the applied and non- applied classification within segments on when they first decided to go to college. 166 College Information Level l. The early deciders were less informed about "social opportunities" at various colleges than were the late deciders, prior to their senior year of high school. On other informational dimensions there was no difference in the two time-dependent classifications. 2. The applied were less informed about college "social opportunities," before their senior year in high school, than were the non-applied. 3. The applied were more informed about "fields of study," before their senior year in high school, than were the non-applied. 4. There was no difference between the applied and the non-applied on any other dimensions of college information. 5. The applied of the SAT segment were more informed about the "social opportunities" at colleges, than were the non-applied. There was no difference within the ACT segment. 6. The applied within the ACT segment were more informed about "fields of study" than were the non-applied. There was no difference within the SAT segment. 7. "Relatives and friends," "college sources," "location," and "campus visits" were reported as the major sources of information about colleges (72 percent of the responses). 167 8. In both the ACT and SAT segments, late deciders considered "high school classmates" as a more useful college information source than did the early deciders. 9. Within the ACT segment the applied found "high school classmates" only slightly useful as an information source; the non-applied found this source either very use- ful or not useful. 10. The ACT applied found college Visits more use- ful than did the non-applied. Intentions: College and Major 1. There was no difference in the degree of cer- tainty about attending a specific college between or within the ACT and SAT segments. 2. The SAT segment non-applied were more certain about their college major than were the applied. Most Informed and Intended College 1. Those prospective students who decided early to go to college less frequently expected to attend the college about which they were most informed, prior to their senior year of high school. The difference was more pro- nounced within the applied group. No difference was apparent between the ACT and SAT segments. 2. Those prospective students who were planning to attend a college other than the one about which they had been most informed considered more colleges before applying. 168 This difference was significant in the SAT segment, and among the non-applied of both the ACT and SAT segments. Post-Enrollment Period Findings The stability of the student's evaluative criteria structure was determined by correlating the rank order of importance of the criteria at two time periods. If there was significant correlation (alpha = .05) the evaluative criteria structure was considered to be stable. If there was not significant correlation, the structure was con- sidered unstable. 1. There was no significant difference in the number of students with a stable evaluative criteria struc- ture, within the ACT segment at t2 - t compared with 3 t1 - t3. There was, however, an absolute increase in the number with a stable structure at t2 - t3. (a) The evaluative criteria structure was less stable for the ACT applied, t - t than 1 3' for the non-applied. (b) There was no difference in the structural stability between the ACT and SAT segments, or within these segments, t2 - t3. 2. Those prospective students indicating they would most likely attend the college about which they were most informed, before their senior year in high school (matched) had more stable evaluative criteria than those not matched. 169 3. Whether the decision to go to college was made before their sophomore year (early deciders) or later, did not appear associated with the stable group more frequently than with the unstable group at t2 - t3. 4. Those early deciders who were also matched on the college most informed and college most likely to attend, had more stable evaluative criteria. 5. Those prospective students with stable evalu- ative criteria tended to apply to fewer colleges than did the unstable group. However, no difference was found in the number of colleges considered. The stable group screened the considered college set more closely, resulting in fewer applications. This suggests they were better able to make alternative reducing decisions, thus approaching the actual college choice earlier in the decision period. 6. Of those students actually enrolling in college, those who enrolled in a public college had more stable evaluative criteria, t than those who enrolled in l't3' a private college. This difference was not found in either the ACT or SAT segments at the t2 - t3 comparison. Evaluative Criteria t2: Scaled Values 1. The scaled values of evaluative criteria im- portance were not significantly different for the applied and non-applied, or the private and public college classi- fications. 170 2. Of those students applying to the college under study, the enrolled group considered: (a) “type of college" more important; (b) "field of study" less important; and (c) "extracurricular activities" less important, than did the not enrolled group. Evaluative Criteria t3; Scaled Values l. The applied group considered "location" less important than the non-applied group. 2. The private college enrollees considered "size of college" more important and "cost of college" less im- portant, than did the public college enrollees. 3. Those enrolling in the college under study considered the "student body composition" less important than those not enrolling. Applied Group Analysis: Rank Order of Evaluative Criteria In this section and the following section, sig- nificant correlation indicates the groups were alike in the rank order of importance of the evaluative criteria. A lack of correlation indicates the groups were not alike. 1. There was significant correlation across time periods, t2 - t3, for both the enrolled and the not enrolled groups' evaluative criteria rank order of importance. 2. The enrolled and not enrolled groups were significantly correlated on evaluative criteria at both t2 and t3. 171 3. The ACT enrolled group was not correlated with the SAT enrolled group at t at t the groups were corre- 2’ 3 lated. 4. The ACT not enrolled and the SAT not enrolled 2 and t3. 5. The ACT enrolled and not enrolled groups were groups were significantly correlated at both t significantly correlated at t2 and t3. 6. The SAT enrolled and not enrolled groups were significantly correlated at t2 and t3. Non-Applied Gropp Analysis: Rank Order of Evaluative Criteria 1. There was significant correlation for both the private college and public college groups across time period, t2 - t3. 2. The private college and public college groups at t the groups were were significantly correlated at t2; 3 not significantly correlated. 3. The ACT private college and SAT private college groups were not significantly correlated at t2; at t3 the groups were not significantly correlated. 4. The ACT public college and SAT public college groups were significantly correlated at both t2 and t3. 5. The ACT private college and public college groups were significantly correlated at t2; at t3 the groups were not significantly correlated. 172 6. The SAT private college and public college groups were not significantly correlated at t at t the groups 2’ 3 were significantly correlated. Selected College Characteristics Comparison 1. Within the ACT segment, those applying to the college under study enrolled in smaller colleges than did the non-applied group. The SAT segment showed no signifi- cant difference across the applied and non-applied groups on the size characteristic. 2. Significantly more of the applied of both the ACT and SAT segments received financial aid from their en- rolled college than did the non-applied. Significantly more of the non-applied received no financial aid from any source. 3. The enrolled group of the college under study was not significantly different from the not enrolled group on any source of financial aid. 4. Within both the ACT and SAT segments the applied differed from the non-applied on "type of college" and "cost of college" characteristics. Hypotheses and Conclusions Hypothesis I: A buying intention statement in terms of the prospective student's choice rating of a particular college, i.e., first, second, third choice, etc., will serve to predict application and enrollment more frequently than other data available to the college. 173 The prospective student's choice preference rating, first choice, was found to be the best single predictor of student applications. Preference ratings below first choice did not discriminate, i.e., a second choice not more likely to apply than a third choice, etc. While the first choice preference rating more effectively predicted applications, it did not predict enrollment within the applied group significantly better than any other choice designation. Hypothesis II: Identifiable market segments of prospective students interested in a particular college, such as, the ACT segment and the SAT segment will differ in their characteristics and behavior. The two identifiable market segments, (1) the ACT segment, and (2) the SAT segment displayed significant associated differences both between segments and within segments across applied and non-applied, and other behavior determined classifications. These segmental differences suggest the opportunity to develop specialized communication and recruitment strate- gies better oriented to meet student needs. Hypothesis III: Purchase patterns as reported for the pur- chase of economic goods with respect to the level of information and degree of decisiveness will carry over to the college choice process. Some support was found for the carryover of economic goods purchase patterns to the college selection process. Those prospective students who tended to be more informed 174 when purchasing economic goods were also more informed about colleges, relative to the uninformed economic goods purchase pattern group. There was, however, no evidence that the more decisive economic goods purchasers were also more decisive in their college selection, relative to the indecisive group. This may be the result of no consumption advantage accruing to those who behave decisively, since college enrollment is available only at a preset time. Decisiveness may in- crease risk as fewer options would be available, e.g. apply to one rather than several colleges, with no additional payout. The lack of incentive for decisive behavior results in overt behavior which appears indecisive. Hypothesis IV: Prospective college students will change their assessment of the relative importance of selected evaluative criteria over time. Prospective college students do not have a structured set of evaluative criteria which remains constant over time. The evaluative criteria used in this study varied in rank order of importance for most of the individual respondents over time. The tendency was to become more highly corre- lated as the enrollment period (actual purchase) neared. However, the number of students correlated at t2 - t3 was not significantly different than the number at tl - t3. Many students never did develop consistency in the importance of the evaluative criteria used in the study. This appears to be consistent with the lack of a firm set 175 of evaluative criteria associated with extensive problem solving behavior. Hypothesis V: Behavior determined segments of prospective college students will differ in the relative importance of selected evaluative criteria at different points in time. Of the students applying to a college, the enrolled group and not enrolled group were generally in agreement on the rank order importance of evaluative criteria. The actual enrollment choice appears to be based on individual college differences, as assessed by the student, across the evaluative criteria. More differences exist, with respect to the im- portance of the evaluative criteria, within the non-applied group when it was divided into private college enrolled and public college enrolled segments. 1. Over time the private college enrollees and the public college enrollees become less similar in the ordering of their evaluative criteria. 2. The private college enrollees were less homo— geneous in their ordering of the evaluative criteria than were the public college enrollees. The least change in the evaluative criteria struc- ture was with that segment of students which was going to attend the college about which they were most informed, before their senior year in high school. This segment also applied to fewer colleges, but was not significantly 176 different in the number of colleges considered. Stability in the evaluative criteria structure and greater decisive- ness were associated in this group. However, it cannot be concluded that the more consistent evaluative structure resulted in greater decisiveness. The opposite relation- ship could also have existed. Recommendations for Future Studies The exploratory nature of this study has generated findings and conclusions which suggest the need for addi- tional research, both of a theoretical and empirical type. For instance, the direction of affect associated with the stable (correlated) evaluative criteria structure found in the matched and early decider group was not deter- mined in this study. This would seem to be a fruitful area for additional research. From a specific college's View point, the methodology used in this study can reveal significant differences associ- ated with various student segments. These differences can serve as a foundation for planning different tactical and strategic programs to more effectively serve the student group and the college's purpose. Exposed segmental differ- ences also offer the opportunity for additional in-depth research. For instance, in this study one student segment (ACT applied) tended to apply at significantly more col- leges than did other segments. Research directed to 177 explaining this phenomenon might be advisable. Special post-application recruitment effort might be required for this segment to aid them in their actual college choice. There is evidence that college marketing effort is becoming much more common and overt. This is particularly true among private colleges as they struggle for survival in a highly competitive market place. Additional research that will contribute to more effective and beneficial application of marketing technology to aid both students and colleges seems advisable. Marketing techniques, when applied to areas other than business, may be misunderstood and misapplied endangering the institutions and customers involved, as well as, the reputation of the discipline. The basis for effective marketing planning is knowl- edge. and understanding of the customer group to be served. This is no less true for educational marketing than other types of marketing, and in this study an attempt has been made to add to such knowledge and understanding with both the reported findings and the methodology employed. APPENDICES 178 APPENDIX A CHI SQUARE TABLES 179 180 TABLE A-l.--Frequency of Match of the Descriptors and the Evaluative Criteria. Application State Classification Matched Not Matched Total Applied 97(90.9) 43(49.l) 140 Non-Applied 153(159.l) 92(85.9) 245 Total 250 135 385 Critical Value: Alpha = .05, d.f. = l, X2 = 3.84. Calculated X2 = 1.83. TABLE A—2.--Frequency of Match of the Descriptors and the Evaluative Criteria by High and Low Range. High Range Low Range Application State Classification 7 - 5 4 - 0 Total Applied 13(9.5) 7(10.5) 20 Non-Applied 13(16.5) 22(18.5) 35 Total 26 29 55 Critical Value: Alpha = .05, d.f. = l, X2 = 3.84. Calculated X2 = 3.90. 181 .cowumuwuo m>eumsam>m may no Amosmuuanev umouo xsmu may nmsmen mnu msam> ms» Hmzoa mae "muoz mo.m oe.a oo.m oo.m Hm.o mo.m em.m Ame com: meH Hm awe moa Ame wee mHH Hmouo xcmm no How immune omeammmuooz mo.m mm.H mm.o mm.~ mm.m mm.m oo.m Amy com: HNH em mm am he eoa mm Hmono xcmm mo Bum remade mmmmmmm Hmasoeuuso zosum mmem umoo coeumooq zoom mmmaaoo coaumoememmoau uonuxm mo oemam ucoooum mo maze muoum demuooaammm .mmanmeuo> menmueuo m>eumsam>m map No mmsHm> moms ocm mmsam> umouo Msmm mo Edmll.m|¢ mamme 182 TABLE A-4.--Type of College Variable: Frequency by Rank Order Without Regard to the Match or No Match Condition. Rank Order Application State Classification 1 - 2 3 - 4 S - 7 Total Applied 4( 7.6) l3( 7.6) 3(4.7) 20 Non-Applied 17(13.4) 8(13.4) 10(8.3) 35 Total 21 21 13 55 Critical Value: Alpha = .05, d.f. = 2, X2 = 5.99. Calculated x2 = 9.63. TABLE A-5.--Type of College Variable: Frequency of Match Condition and Rank Order Values. Mahfll Nonknch Emmzankm’ Emm:0nkm' Applkxuionffiate Chmxfifflxnion .l-2 3-7 l-2 3-7 fkfial .mgflied 2CL3) 100L7) 2UL4) 6(1L6) 20 NCnoApplied 7(5.7) 3(8.3) 10(7.6) 15(13.4) 35 Tbtal 9 13 12 21 55 Note: Three cells have expected frequencies below five, violating the 20 percent rule of chi-square. However, because two of the cells were only slightly below five and the high calculated value of X2, it was decided to use this value and make the test. Critical Value: Alpha = .05, d.f. = 3, X2 = 7.81. Calculated X2 = 12.77. 183 TABLE A-6.--Type of College Variable: Frequency of Match Condition. Descriptor Condition Application State Classification Match No Match Total Applied 12( 8.0) 8(12.0) 20 Non-Applied lO(l4.0) 25(21.0) 35 Total 22 33 55 Note: This format was used for testing each of the variables. Critical Value: Alpha = .05, d.f. = l, X2 = 3.84. 2 Calculated X = 5.02. TABLE A-7.--College Choice Preference Rating: lst, 2nd, and 3rd or Below. Application State lst 2nd 3rd Choice Classification Choice Choice or Below Total Applied 15( 7.1) 6( 8.5) 2( 7.4) 23 Non-Applied 6(13.9) l9(16.5) 20(14.6) 45 * Total 21 25 22 68 *Note: Includes students who did not rank the evaluative criteria, but who did indicate a college choice preference rating. Critical Value: Alpha = .05, d.f. = 2, x2 Calculated X2 = 20.32. = 5.99. 184 TABLE A-8.--College Choice Preference Rating: lst, and 2nd or Below. Application State lst 2nd Choice Classification Choice or Below Total Applied 15( 7.1) 8(15.9) 23 Non-Applied 6(13.9) 39(31.l) 45 * Total 21 47 68 *Note: Includes students who did not rank the evaluative criteria, but who did indicate a college preference rating. Critical Value: Alpha = .05, d.f. = 1, x2 = 3.84. Calculated x2 = 19.21. TABLE A-9.--College Choice Preference Rating: lst and 2nd. Application State lst 2nd Classification Choice Choice Total Applied 15( 9.6) 6(ll.4) 21 Non-Applied 6(ll.4) 19(13.6) 25 * Total 21 25 46 *Note: Includes students who did not rank the evaluative criteria, but who did indicate a college preference rating. Critical Value: Alpha = .05, d.f. = 1, x2 = 3.84. Calculated X2 = 10.30. 185 TABLE A-10.--College Choice Preference Rating: 2nd and 3rd. Application State 2nd 3rd Classification Choice Choice Total Applied 6( 4.3) 2( 3.7) 8 Non-Applied l9(20.7) 20(18.3) 39 * Total 25 22 47 *Note: Includes students who did not rank the evaluative criteria, but who did indicate a college preference rating. Critical Value: Alpha = .05, d.f. = 1, x2 = 3.84. Calculated X2 = 1.75. TABLE A-ll.--College Choice Preference Rating for Students Matched on the Type of College. Matched on Type of College Application State lst 2nd Choice Classification Choice or Below Total Applied 9(5.5) 3(6.5) 10 Non-Applied l(4.5) 9(5.5) 12 Total 10 12 22 Critical Value: Alpha = .05, d.f. = l, X2 = 3.84. 2 Calculated X = 9.06. 186 TABLE A-12.--Individual Parent's Educational Classification: Combined Group. Application State No Some Classification College College Total' Applied 60( 69.1) 60(50.9) 120 Non-Applied 123(113.9) 75(84.l) 198 Total 183 ‘ 135 318 Critical Value: Alpha = .05, d.f. = 1, x2 = 3.84. Calculated x2 = 4.54. TABLE A-13.--Individual Parent's Educational Classification: ACT Group. Application State No Some Classification College College Total Applied 35(38.4) 31(27.6) 66 Non-Applied 71(67.6) 45(48.4) 116 Total 106 76 182 Critical Value: Alpha = .05, d.f. = l, X2 = 3.84. Calculated x2 = 1.13. 187 TABLE A-l4.--Individual Parent's Educational Classification: SAT Group. Application State No Some Classification College College Total Applied 25(30.6) 29(23.4) 54 Non-Applied 52(46.4) 30(35.6) 82 Total 77 59 136 Critical Value: Alpha = .05, d.f. = l, X2 = 3.84. Calculated X2 = 3.92. TABLE A-15.--Brother's and Sister's Educational Classification: Combined Group. Attended or Now Application State Attending College, Graduated Classification but not Graduated College Total Applied 28(23.l) 14(18.9) 42 Non-Applied S6(60.9) 55(50.l) 111 Total 84 69 153 Critical Value: Alpha = .05, d.f. = l, x2 = 3.84. 2 Calculated X = 3.18. 188 TABLE A-16.--Brother's and Sister's Educational Classification: ACT Group. Attended or Now Application State Attending College, Graduated Classification but not Graduated College Total Applied lS(lO.2) 2( 6.8) 17 Non-Applied 32(36.8) 29(24.2) 61 Total 47 31 78 Critical Value: Alpha = .05, d.f. = l, X2 = 3.84. 2 Calculated X = 7.23. TABLE A-l7.--Estimated Value of Homes Within the Neighborhood of Residence (excluding rural and farm): Combined Group. Application State $30,000 Below Classification or Above $30,000 Total Applied l9(l9.l) 17(16.9) 36 Non-Applied 32(3l.9) 28(28.l) 60 Total 51 45 96 Critical Value: Alpha = .05, d.f. = l, x2 = 3.84. 2 Calculated X = .01. 189 TABLE A-18.--Respondent Classification by the Number of Moves (last seven years): Combined Group. Application State No Two or Classification Moves One More Total Applied 29(29.2) 21(23.3) 7(4.4) S7 Non-Applied 50(49.8) 42(39.7) 5(7.6) 97 Total 79 63 12 154 Critical Value: Alpha = .05, d.f. = 2, x2 = 5.99. Calculated x2 = 2.79. TABLE A-19.--Vacation Trip Companion Preference: Applied Group. Group Family or Classification Self Friends Total ACT 16(12.l) 16(19.9) 32 SAT 6( 9.9) 20(16.l) 26 Total 22 36 58 Critical Value: Alpha = .05, d.f. = 1, x2 = 3.84. Calculated X2 = 4.50. 190 TABLE A-20.--Vacation Trip Companion Preference: ACT Group. Application State Family or Classification Self Friends Total Applied 16(ll.9) 16(20.l) 32 Non-Applied 17:21.1) 40(35.9) 57 Total 33 56 89 Critical Value: Alpha = .05, d.f. = 1, x2 = 3.84. 2 Calculated X = 3.58. TABLE A-21.--Descriptive Accuracy Associated With Purchase Pattern Descriptor Statement"C": Combined Group- Rank Order of Accuracy Application State Classification 1 2 3 or 4 Total Applied 35(34.8) 9(12.7) 9(5.4) 53 Non-Applied 61(61.2) 26(22.3) 6(9.6) 93 Total 96 35 15 146 Critical Value: Alpha = .05, d.f. = 2, x2 = 5.99. 2 Calculated X = 5.44. 191 TABLE A-22.--College Informed Classification, Before Senior Year of High School: ACT Group. Number of Variables More Informed Purchase Pattern than Uninformed about Colleges lst Ranked Descriptors 5 or less 6 or more Total Informed (B or C) l9(24.4) 55(49.6) 74 Uninformed (A or D) 10( 4.6) 4( 9.4) 14 Total 29 59 88 Critical Value: Alpha = .05, d.f. = l, x2 = 3.84. 2 Calculated X = 11.23. TABLE A-23.--Information Source Degree of Importance, Frequency Distribution for Sales People: Combined Group. Dzneecfi’hmnrfinme ImplkxuionSHBte (nasstfikwmion l