3 ll! III 293 10063 2409 1 “it’s This is to certify that the thesis entitled A PREDICTION MODEL FOR DETERMINING STUDENT ENROLLMENT AT GENEVA COLLEGE presented by Robert E. Armstrong, III has been accepted towards fulfillment of the requirements for Ph.D. _dggreein Administration 8: Higher Education /Ma10r /professor Date //2//77 0-7 639 llllllllllllllllllfll||||||llllll|lllllllllllllfl Luau a y ‘* Michigan Stan: OVERDUE FINES ARE 25¢ PER DAY PER ITEM .Return to book drop to remove this checkout from your record. A PREDICTION MODEL FOR DETERMINING STUDENT ENROLLMENT AT GENEVA COLLEGE By Robert E. Armstrong, III A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY College of Education 1979 ABSTRACT A PREDICTION MODEL FOR DETERMINING STUDENT ENROLLMENT AT GENEVA COLLEGE By Robert E. Armstrong, III As private church-related institutions of higher education look to the future in the 1980‘s and 1990's, the financial crises precipitated by increasing costs and declin- ing enrollments becomes more of a reality. The purpose of the study was to develop a body of knowledge, including a prediction model that would help Geneva College predict the probability of a student enrolling at the institution. The study was based on data derived from a sample con- sisting of the 673 students who applied for admission to Geneva College during the 1976-77 school year. A comparison with those students who confirmed their intentions of enroll- ing at Geneva College was made with those students who with- drew from the admissions process and did not enroll at the College. The study consists of two parts. The first part was the development of a prediction ‘model which used 13 independent variables to predict the two dependent variables. The Enrollment Prediction Model devel- oped in the study was defined as follows: Y a A + ZijX. + e J where: Robert E. Armstrong, III the criterian measure; A = a constant term; B = the net regression weight that determines the relative effect on Y on a specific X value; X the specific predictor variable; e 3 an error term. This model is used to describe the effect that various specific predictor variables have on the dependent variable being examined. The prediction model in the study indicates that status, specifically, confirmed status, can be statis- tically predicted at a significance level of .05. There were three statistically significant variables developed in the study. The second part consisted of a series of tests for significant difference between the dependent variables de- fined in terms of each independent variable. It was deter- mined that six variables were statistically significant while the remaining seven variables were not significant at the .05 level. This determination was based on the examination of each independent variable using the appropriate statistic. It was further determined that the statistically significant predictors were also operationally significant in the Geneva College Admissions program. The data were analyzed on the Michigan State computer system using the Statistical Package for Social Science Robert E. Armstrong, III (SPSS). The evaluation of the data uses Pearson Product Moment correlation, multiple regression analysis, t-Tests, and chi-squared techniques. The .05 level was used in all tests as the level of acceptable statistical significance. The Enrollment Prediction Model developed in the study, although specifically related to and based upon data from Geneva College may be applicable to other church-related institutions. ACKNOWLEDGEMENTS I wish to express my sincere appreciation to Dr. Louis G. Romano, my Chairman, and to Dr. Vandel C. Johnson, Dr. Lawrence Borosage, and Dr. John W. Allen for serving on my committee. Their comments and suggestions have been invalu- able and necessary in the completion of this task. A special thanks to Geneva College, and particularly to Paul Arnold of that institutions for his assistance in the statistical analysis of the data. A personal thanks to Herrell DeGraff, Jack Allen, and George Farah whose direct and indirect encouragement enabled me to finish a task first inspired by Marian W. Armstrong. And finally, a debt of gratitude to Sandy for her understanding support throughout the process. ii TABLE OF CONTENTS CHAPTER 1 - INTRODUCTION Financial Crisis in Private Higher Education Enrollments The Changing Role of the Admissions Department The Role of the Christian College and the Need for Prediction Statement of the Problem Statement of Purpose Objectives Design of the Study Related work Scope and Limitations Definition of Terms Assumptions Hypotheses and Variables Hypotheses Variables: A List of General Predictor Variables Method of Presentation CHAPTER 2 - RELATED LITERATURE Forward Methodology Enrollment Patterns iii 10 12 12 13 14 15 15 17 18 l9 19 20 20 22 22 26 Overall Enrollment Trends Private Church—Related Institutions Prediction and Prediction Models: A General Background CHAPTER 3 - PLANNING AND DEVELOPING THE STUDY Population Procedures and Instruments Methodology and Analysis of Data The Model Limitations Summary CHAPTER 4 - ANALYSIS OF DATA Forward Part I: Hypothesis 1 - The Model Part II: Hypotheses 2 through 14 - Tests of Independent Variables Part III: Comparative Analysis and Operational Application Comparative Analysis Operational Applications Summary CHAPTER 5 - SUMMARY AND RECOMMENDATIONS Summary The Study Review of Literature Related Work iv 41 53 54 59 61 62 62 64 64 72 101 101 105 107 108 110 111 112 Conclusions Recommendations for Further Study Reflections APPENDICES Appendix A - Admissions Apple File Appendix B - Applicant File Map Appendix C - A Review of the Methodology of Models and Methods of Forecasting 117 120 Table Table Table Table Table Table Table Table Table Table Table Table Table Table 1-2 2-1 2-2 4-1 4-3 4-4 4-6 4—7 4-8 4-9 4-10 LIST OF TABLES Survey of 372 Church-Related, Four-Year Liberal Arts Colleges and Universities in the United States with 1975 FTE Enroll- ments of 250 to 2900 Students Experience of Private Colleges and Uni- versities in the Recruitment and Admis- sion of Undergraduate Students, 1969-70 through 1975-76 Statistical Summary Variables in the Final Regression Equation Variables Not in the Final Regression Equation Multiple Regression Variables Affecting Predictability of Status Correlation Coefficients for Variables in the Multiple Regression Analysis Inde- pendent Variables Affecting Status The Effect of Anticipated Term of Matricu- Withdrawals lation on Status--Confirmed vs. The Effect of Date of Campus Visit on Status--Confirmed vs. Withdrawals The Effect of Date of Acceptance on Status--Confirmed vs. Withdrawals The Effect of Indicated First Major on Status--Confirmed vs. Withdrawals The Effect of Source of Contact with Col- lege on Status--Confirmed vs. Withdrawals The Effect of Type of Student on Status-- Confirmed vs. Withdrawals vi 31 35 66 67 68 75 77 80 83 86 Table Table Table Table Table Table Table 4-11 4-12 4-13 4—14 4-15 4-16 4-17 The Effect of Church Affiliation on Status-- Confirmed vs. Withdrawals The Effect of Location of Hometown on Status--Confirmed vs. Withdrawals The Effect of Scholarship Aid on Status-- Confirmed vs. Withdrawals The Effect of High School Class Rank on Status--Confirmed vs. Withdrawals The Effect of High School Class Size on Status--Confirmed vs. Withdrawals The Effect of SAT Verbal Scores on Status-- Confirmed vs. Withdrawals The Effect of SAT Math Scores on Status-- Confirmed vs. Withdrawals vii 88 90 94 97 98 99 100 Figure 1-1 Figure 1-2 LIST OF FIGURES viii CHAPTER 1 INTRODUCTION Financial Crisis in Private Higher Education Between the years 1957 and 1962, the birth rate in the United States dropped from twenty-three births per thousand of population to fifteen births per thousand of population, and finally leveled off during the 1960's to approximately thirteen births per thousand where it has held ever since. During this same period, American higher education was in the midst of the largest expansion of physical plant in its history. Twenty years later, it is now faced with "excess capacity" and a "decreasing demand" for the services ren- dered. This situation means different things to different institutions. At the large, publicly-supported universities or the heavily endowed private institutions, it means a re-evaluation and probable curtailment of programs. At the community college, it perhaps will spell the end of proli— feration of programs, particularly as local tax payers be- come more resistant to increasing demands for funds. At the small, private college, the implications are much more serious because student enrollment is either directly or indirectly the primary source of revenue. A l relatively small decrease in enrollment, coupled with a high level of fixed costs, can mean financial disaster and the ultimate closing of the institution. The potential financial crisis is now well documented; in his article, "Colleges in the Red," Jack Magarrell re- ports that "The American Council on Education estimates that thirty-four percent of our nation's colleges and universities 1 This issue operated in the red in the past fiscal year." becomes even more relevant for the private colleges, as re- flected in an article from The Chronical of Higher Education: "A report of the financial health of private colleges and universities has found more than a fourth of them in serious distress with their survival in doubt."2 Since most private institutions view the majority of their operating costs as fixed or at least semi—fixed, this shifts the emphasis to the revenue side of the financial equation. The burden of survival is therefore shouldered by the Admissions Department who is usually responsible for generating the necessary revenues from enrollment. Enrollments Just as certainly as the period between 1957 and 1962 saw a sharp drop in the birth rate, so the period of 1977- 1979 is most certain to see a marked decrease in college enrollments. The impending financial crisis brought about by these reduced enrollments has been a foregone conclusion for the last three to five years as reflected by the analy- sis of high school enrollments; nevertheless, the college community seems to be playing ostrich in an attempt to avoid the issue. Projections such as the one developed by Frankel and Beamer have been available since 1974 and indicate the problem higher education will face as a result of declining high school enrollments: The latest USOES estimates indicate that 1976-77 will be the peak year for total number of high school graduates, with 3,199,000 pro- jected. Thereafter, high school graduates are expected to decline each year, so that by 1982-83 the estimated number will be 2,835,000 or a decreasg of about 11.3% over the figure for 1976-77. This information is validated by a number of other govern- mental agencies including the Department of Health, Education and Welfare, which indicates that the potential number of college students will be down by 25-30 percent by the year 1990.4 Again, the information was specifically directed to college administrators by none other than Garland Parker of the University of Cincinnati, in his report of 1976 in which he observed the following: Indeed, this writer has predicted all along that overall enrollments would increase through the 1976-77 period of the 1970's, begin to level off in the late 1970's, peak at the end of the decade or the beginning of the 1980's, and then, probably, start a slow declige that would accel- erate throughout the 1980's. Although there is some discrepancy as to the specific timing of Parker's projected decline in enrollments, apparently they are coming sooner rather than later, the trend and the difficulties resulting from that trend are extremely apparent. A comparative analysis of Table 1-1 and 1-2 indicates that declining emrollments were becoming a problem not only for higher education in general, but for church-related institutions specifically. Figures 1-1 and 1-2 reflect the fact that Geneva College (controlled by the Reformed Presbyterian Church- North America) was one of the 122 schools reporting losses during the 1965-75 period, as well as a slight increase in 1976 over 1975. The tables and figures de- pict a rather erratic enrollment picture reflecting the difficult time that Geneva has been having attracting the students that will generate the necessary revenue for the survival of the institution. The changing enrollment picture has affected the role played by the Admissions Department as they face their new and in- creasingly important responsibility. The Changing Role of the Admissions Department The role of the Admissions Officer has changed most drastically over the last twenty years, so much so that the function at most institutions is nor of admissions, but now one of recruitment and promotion. Again, this reflects the change in the capacity of the higher educational system to Survey of 372 Church-Related, Four-Year Liberal Arts Colleges and Table 1-1 Universities in the United States with 1975 FTE Enrollments of 250 to 2900 Students: Aggregate full-time-equivalent (FTE) enrollment 1965 compared to 1975: 1975 # 324,124 1965 287,052 +37,144 1* +13 (n) number of institutions reporting 327 (88% of 372) FTE enrollment 1965 compared to 1975 according to l 965 l 975 # %* n #gaining #losing Protestant Episcopal 2404 4094 + 1690 +70 4 4 0 Assemblies of God 1916 2810 + 894 +47 4 4 0 Baptists 7050 9745 + 2695 +38 9 8 1 Southern Baptists 30554 39941 + 9387 +31 30 25 5 Seventh-Day Adventists 5913 7474 + 1561 +26 7 5 2 Mennonite Church 2325 2782 + 457 +20 3 3 0 Roman Catholic 73069 86120 +13051 +18 95 55 40 Free Methodist 1940 2227 + 287 +15 3 2 1 American Lutheran 8891 10095 + 1204 +14 6 5 1 Christian Church Disciples of Christ 2818 3112 + 294 +10 3 2 1 Lutheran Church in America 14683 15672 + 989 + 7 12 7 5 Presbyterian, U.S. 7677 8116 + 439 + 6 11 7 4 Friends 4678 4939 + 261 + 6 6 4 2 Church of the Brethren 3839 3853 + 14 0 4 2 2 United Methodist 57826 57561 - 265 0 57 26 31 United Church of Christ 3617 3524 - 93 - 3 5 2 3 American Baptist 3881 3643 - 238 - 6 5 2 3 United Presbyterian, U.S.A. 19279 18020 - 1259 - 7 l9 8 11 Church of the Mazarene 6622 5950 - 672 -10 6 3 3** Others 28070 34446 + 6376 +23 38 29 9 religious affiliation: McCrath, Earl J. and Richard C. Neese. "Are Church-Related Colleges Losing Students?" Arizona University, Tucson-College of Education, Topical Paper #6. Source: *Percentages rounded to the nearest whole number. **If two Nazarene colleges founded in the late '603 were included for the ten-year period, aggregate enrollment for the Nazarene colleges would have increased by 660 for a nine percent gain. Aggregate full-time-equivalent (FTE) enrollment 1975 compared to 1976: 1976 # 345,717 1975 338,365 +7,352 %* +2 Table 1-2 (n) number of institutions reporting 343 (92% of 372) FTE enrollment 1975 compared to 1976 according to religious affiliation: 1975 1976 # Z* n #gaining #losing Church of the Nazarene 5950 6812 + 862 +14 6 5 1 American Baptist 3643 3906 + 263 + 7 5 2 3 Seventh-Day Adventists 7474 7811 + 337 + 4 7 5 2 Roman Catholic 87471 90833 + 3362 + 4 97 57 39** Church of the Brethren 3853 3963 + 110 + 3 4 3 1 United Presbyterian, U.S.A. 19632 20106 + 474 + 2 22 8 14 United Methodist 58665 60039 + 1374 + 2 58 39 19 Lutheran Church in America 16361 16680 + 319 + 2 13 4 9 Friends 4939 4992 + 53 + 1 6 3 2** Free Methodists 2227 2249 + 22 + l 3 2 l Assemblies of God 2810 2831 + 21 + l 4 2 2 Southern Baptist 44403 44729 + 326 + 1 34 20 13** Christian Church Disciples of Christ 3497 3501 + 4 0 4 2 2 Baptists 9745 9684 - 61 - 1 9 6 3 Presbyterian, U.S. 9386 9326 - 60 - l 12 5 7 American Lutheran 11278 11205 - 73 - l 7 3 4 Protestant Episcopal 4094 4055 - 39 - 1 4 2 2 United Church of Christ 4518 4436 - 82 - 2 6 4 2 Mennonite Church 2782 2680 - 102 - 4 3 1 2 Others*** 35637 35879 + 242 + 1 39 24 14** Source: McGrath, Earl J. and Richard C. Neese. "Are Church-Related Colleges Losing Students?" Arizona University, Tucson-College of Education, Topical Paper #6. *Percentages rounded to the nearest whole number. **The enrollment in one college remained unchanged. ***Others (religious groups with two colleges or less reporting) include: African Methodist Episcopal, African Methodist Episcopal Zion Church, Brethren Church, Bretheren in Christ Church, Christian Methodist Episcopal, Christian Missionary Alliance, Christian Reformed Church, Church of Christ, Church of God (Anderson, IN; Cleveland, TN; Findlay, OH), Evangelical Covenant Church of America, Friends United Meeting, General Conference Mennonite Church, Mennonite Brethren, Moravian Church, Reformed Church in America, Reformed Presbyterian, Reorganized Latter-Day Saints, United Brethren Church, Wesleyan, Interdenomin- ational, Multiple Protestant Denominations. Figure 1-1 1600 n— 1500 — 1477 1472 1400 1200 1100 1000- Note: The difference between the two lines for any year represents the night school enrollment. 00F 11 I 1 l 1 1 J, 1 1 l 1 I l J .gJ .22 '63 '64 '65 '66 '67 '68 '69 '70 '71 '72 '73 '74 '75 *76 Source: Office of Institutional Research and Planning, Geneva College. 400 350 300 250 .200 150 100 50 F' 1-2 405 igure 272 269 £9 262 234 230 “e 247 25“ 226 235 223 200 r 186 200 ‘179 162 156 h o 142 00%? Qb 125 5‘ _ 100 70 51 47 48 l l l l l l l l l l 'asisr'“ '67 '68 '69 '70 '71 '72 '73 '74 '75 Source: Office of Institutional Research and Planning, Geneva College. supply a service in comparison to the demand for that ser- vice reflected by the society. As Thomas LaBaugh comments in his recent Michigan State University dissertation: Nicoll identifies a shift during that decade (1960's) as the admissions structure moves away from the role of counseling students seeking a suitable college, to a role of "recruiter of students for the college they represent." In October of 1970, L. Richard Meeth des- cribes the function of the admissions office in business terms when he asks: "Does the college have a clear and realistic understanding of its market?" This is the first time in the review of the literature that the term "market" arises.6 A good case could be built for observing that the "edu- cational industry" has gone through the same stages of devel- opment as most other industries. In the early stages of the cycle the need for education stimulated a demand for the service, which was met by increased capacity. As capacity increased, the services supplied met the demand and we reached a short period of comparative balance during the sixties. As we moved into the later stages of the cycle the demand declined, leaving the many problems typical of an industry in the more advanced stages of the product growth cycle. As indicated by the quotation from LaBaugh's disserta- tion, higher education has reacted similarly to any other industry with excess production capacity. The result has been a whole series of articles, chapters, and even books dealing with the role of the "Admissions Officer."7’8’9 As the actual pressure for survival increased, with continuing decrease in demand, coupled with a fixed supply 10 and no viable "substitution function" or "ease of exit", the tendency was to "sell the product." Again, the results of this kind of activity brought on all sorts of editorial comments as to the differentiation between "sales" and ”marketing" and the ethical behavior associated with each.10'11’12 Again the behavior can be explained in terms of the marketing functions of a product life cycle and the shift to what McCarthy and others refer to as'market orientation" rather than'product orientation."13 Those institutions in the education industry who start to adopt a market orienta- tion in an attempt to identify and satisfy student needs will not only be supplying a valid and ethical service, but will also have a much better chance of survival. The key to a market-oriented organization is the identification of a market segment which can best be met by that particular in- stitution. The Role of the Christian College and the Need for Prediction The preceeding section suggests that there is a speci- fic place for the church-related college, provided it can satisfy the need of a particular market segment; the conclu- sion of a study by McGrath and Neese completed in 1977 validates this position: It appears that the present condition and future prospects of these institutions is not as ll precarious as some observers have asserted, and that the educational and fiscal health of the church-related colleges will in large part de- pend on their tenacity in holding to the basic religious, spiritual, and moral principles that animated their establishment. Church support and government finacial aid to students who choose these colleges and universities will also be im- portant in assuring their survival and educational quality. But the schools will have to represent an attractive option {0 young people in order to compete successfully. 4 The last sentence reflects what McCarthy refers to as "mar- ket segmentation."15 One of the starting points of market segmentation is an analysis of present "customers," i.e., enrolled students, and those that were interested at one point, but at the last minute changed their mind (with- drawals). It is with this purpose in mind that a prediction ‘model for Geneva College is being examined. The grim facts are these: a 25 to 30 percent decrease in the national pool of potential students during the next fifteen years; a 50 percent decrease in the Western Pennsyl- vania area; 80 percent of the decrease in most cases coming in the middle income class; a trend away from the four-year liberal arts colleges,and a trend to the community colleges and vocational training in general. If Geneva College is to survive and prosper in the face of these changing market conditions , adequate market data and the tools to analyze it becomes in- creasingly important. Considering the obvious seri- ousness of the problem and the importance of student enrollment, the objectives of this study as they relate to Geneva College become quite specific. 12 Statement of the Problem The preceeding observations lead to a specific problem statement: to identify those students who are most likely to enroll at Geneva College. Student enrollment represents 85 percent of total revenues for Geneva and is thus one-half of the financial equation. Given the high percentage of fixed costs, the overriding problem for Geneva College is quite simply "where are the students going to come from,’ or in terms of marketing "who is going to buy our product," in this case "a service" called college education. The study of the development of a prediction model is therefore being undertaken due to the declining enrollments of Geneva College as reflected in Figures 1-1 and l-2. The problem is compounded by the fact that enrollment at other "Christian or church-related" colleges has not de- clined as severely as has the enrollments at Geneva (Tables 1-1 and l-2). Statement of Purpose A thorough review of the literature reveals that there are no prodiction models for enrollment at private, church- related institutions of higher education. The purpose of this study is to develop a body of knowledge, including the development of a prediction model which would: l3 1) Identify at the thne of application those students who are most likely to finally enroll at Geneva. 2) Identify those who will probably not enroll at Geneva, even if accepted. 3) Concentrate the recruitment efforts on those students who are most likely to ultimately enroll. 4) Examine the difference to see if anything can be done to increase the number of students who enroll. Objectives As a result of the proposed investigation of the prob- lems suggested by the preceeding Statement of Purpose, it is hoped that the findings will aid Geneva College to identify the kinds of students whose educational needs are ‘met by the institution, and thereby maintain sufficient enrollment so as to continue to be a viable church-related liberal arts college. The major objectives are: 1) 2) 3) 4) 5) To determine the characteristics of the confirmed students. To determine the characteristics of the "withdrawals" as defined by Geneva College. To compare the characteristics of the confirmed students with the character- istics of the "withdrawals." To determine if there are significant differences between the two groups. To examine the possible differences be- tween the groups in an attempt to build a model which helps predict the students who are most likely to enroll. 14 Design of the Study The study uses a sample made up of the total number of students who applied to Geneva College during the 1976-1977 school year for admission in the Fall of 1977. The total population was broken into three segments: those who were not accepted by Geneva; those who were accepted and "con- firmed" their intentions of enrolling, and; those who were accepted and "withdrew" (i.e., those who did not enroll). The data were collected by members of the Admissions Department of Geneva College using an "Admissions Apple File" form to record the data on each applicant (Appendix A). These data were stored in an "applicant file" (Appendix B) on the I.B;M. computer at Geneva College, which is part of the standard data bank kept on each Geneva Col- lege student. An examination was made of the data that was available on each student for the school year 1977-78 and pertinent data was key-punched on Fortran Computer cards by the Geneva College Data Processing Center personnel. The data from these cards were then used in testing hypotheses of the study. The Multiple Regression Model, available from the Statistical Packagefor the Social Science (SPSS) on the Michigan State University's Control Data Corporation 3600 Computer, was used to test the first hypothesis, while the remaining hypotheses were tested by use of chi-square technique and t-Tests. 15 Related Wbrk Aid in the design of the study was obtained from Dr. Billie Rader of the Department of Vocational Education, College of Education, Michigan State University, Mr. William Brown, Research.Assistant for SPSS, Michigan State Univer- sity's Computer Center, and from Dr. Paul Arnold, Professor of Statistics, Geneva College. Scope and Limitations The findings of this study will be limited to and based upon a sample defined as those students applying to and accepted by Geneva College during the 1976-77 school year for admission in the Fall of 1977. The total popu- lation for that year therefore represents a sample of all students that will apply to Geneva College in the future. The students that applied to Geneva College and were re- jected by the college, although originally included as part of the total population, were eliminated from the data base due to the fact that this information was of no value in predicting the students that would eventually enroll at the college. A much larger group of students that were eliminated from the study, due to a lack of data as well as financial and time constraints, were those stu- dents who requested information about the college but who never actually applied to Geneva. By defining the sample in this manner the total population was used as a basis for the study. 16 As a result of this restricted data base, the predic- tions derived from the study will be directly generalized to future enrollments of Geneva College. On the other hand, the prediction model itself may well be generalized to other institutions which are interested in a shnilar church- related market segment. This group would include some five-hundred institutions of higher learning. Due to the fact that the external variables are chang- ing rather rapidly, the results of this study will be valid only for that period of time during which those external variables remain relatively constant. A case in point might result from such things as a severe recession or an acute energy shortage. A more difficult problem might conceivably result from a shift in the way that the potential market perceives the educational services offered by Geneva College. In such a case, the shifts or changes in the external vari- ables are very difficult to determine. Finally, the study has been lflnited to Geneva College data to gain maximum results from the analysis in an attempt to offset Geneva's specific enrollment problems. By re- stricting the study of a sample consisting of the most recent group of accepted applicants, an attempt will be made to determine accurately those characteristics which identify students who will most likely enroll at the insti- tution. This, in turn, will allow for the highest degree of accuracy from the prediction model. 17 Definition of Terms In order that the reader will be able to understand some particular and peculiar terms used in the study, they are listed and defined below: Acceptance Data--date student was notified by Geneva that he/she was accepted. Confirmation Date--date student paid his deposit indicating hefshe would enroll. "Confirmed"--refers to those students who have applied, been accepted, and have indicated their intention to enroll at Geneva by pay- ment of a deposit. Denographic Characteristics--includes all those characteristics usedito describe a "student." Denomination--indicates broad denominational grouping, such as Evangelical, Protestant, Catholic and all others. First Major--area of study given by student as his/her preferred choice. Geographic Location--identification of student's home town broken into Beaver County, three- county area, Pennsylvania, Ohio, and all others. Model--an equation or set of equations depicting the causal relationships that are believed to generate observed data. Also, the expres- sion of a theory by means of mathematical symbols or diagrams. P.C.S. in Date--date when Geneva received the "Parents Confidential Statement" which is the basis for financial assistance. Source of Contact--the way the student found out about Geneva. 18 Special Scholarship Eligibility--indicates student is eligible for a special scholarship. Student--refers to high school or transfer stu- dents involved in the admissions process. Type of Student--indicates whether the student is full- or part-tfine, day school or night school, commuter or resident. Visit--means the student has actually visited the college. "Withdrawals"--those students who have been ac- cepted but have decided not to enroll at Geneva. Assumptions Due to the fact that the extensive review of the liter- ature shows no similar studies have been conducted, the following assumptions are offered: 1) The information on the student's application form is unbiased and accurate. 2) The information transferred and stored on the computer memory is also accurate. 3) The same information will be available in the future; thereby enabling the prediction model to be annually updated. 4) The information and the prediction model will help in the decisionamaking process at Geneva College so that the necessarily limited re- sources of the Admissions Department can best be used to offset the problem of declining enrollment. l9 Hypotheses and Variables Based on the review of the literature and the assump- tions of the study, the following hypotheses and list of variables are offered. Hypotheses 1) 2) 3) 4) 5) 6) 7) 8) 9) 10) ll) 12) The confirmed students can be predicted sig- nificantly from a set of variables. There are no significant differences in the Term of Matriculation between confirmed and withdrawals. There are no significant differences in the Dates of the First Campus Visit between con- firmed and withdrawals. There is no significant difference in the Acceptance Date between confirmed and with- drawals. There is no significant difference in First Majors between confirmed and withdrawals. There is no significant difference in Source of Contact between confirmed and withdrawals. There is no significant difference in Type of Student between confirmed and withdrawals. There is no significant difference in Deno- ‘mination between confirmed and withdrawals. There is no significant difference in the Distance from Home between confirmed and withdrawals. There is no significant difference in Type of Scholarship Aid between confirmed and 'withdrawals. There is no significant difference in Class Renk between confirmed and withdrawals. There is no significant difference in Class Size between confirmed and withdrawals. 20 13) There is no significant difference in the S.A.T. Verbal Test Scores between confirmed and withdrawals. 14) There is no significant difference in the S.A.T. Math Scores between confirmed and withdrawals. Variables: A List of General Predictor Variables VAR002 Term of Matriculation VAR003 Date of First Visit VAR004 Acceptance Date VAR005 First Major VAR006 Source of Contact VAR007 Type of Student VAR008 Denomination VAR009 Zip Code VAROlO Type of Scholarship Aid VAROll Class Rank VAR012 Class Size VAR013 S.A.T. Verbal VAR014 S.A.T. Math Method of Presentation Chapter 1 has been comprised of a general introduction to the dissertation, a statement of the problem, the general purpose of study, its design, definition of terms and hypo- theses, as well as its limitations. 21 Chapter 2 will deal with the related literature which gives support for the need to develop such a study. This chapter will be somewhat more general than might be the case due to the fact that there have been no prediction models developed for private, church-related institutions of higher education. Chapter 3 will deal with the planning and development of the study. It will include some comments on the need for this study, the decision to use the specific sample chosen, the advantages of the techniques involved, and the type of statistical analysis employed. Chapter 4 will present an analysis of the data, and Chapter 5 will draw conclusions and make suggested recom- mendations. CAHPTER 2 RELATED LITERATURE Forward If, as the Carnegie Commission Report has indicated in its now famous quote "the golden years of higher education" were during the sixties; and if, as a number of people have suggested, we are now in a period of ever-increasing crisis, the study of a prediction model for a college such as Geneva is all the more necessary. The study would, of course, not be complete without an examination of the related literature. This is the topic of this chapter, and the discussion will be broken into four basic parts: 1. 2. The general description of the methodology. The literature related to overall decreasing enrollments. The general use of prediction models. The limited use of thu specific enrollment prediction‘model. Methodology One of the relatively recent and certainly mosteffici- ent research tool which has been developed is the Education- al Resources Information Center, better known as ERIC. By 22 23 definition ERIC is a national information network for acquiring, selecting, abstracting, indexing, stor- ing, retrieving, and disseminating significant and timely education related reports. It con- sists of a coordinating staff in washington, D.C. and 16 clearing houses located at universi- ties or professional organizations across the country. These clearing houses, each responsi- ble for a particular educational agea, are an integral part of the ERIC system. ERIC was started in 1964 under the guidance of the U.S. Office of Education which maintained this system through July of 1972. Beginning in August of 1972, ERIC has been under the governance of the National Institute of Education (NIE). An integral part of the ERIC system, Research in Education, or since 1975 renamed Resources in Education (RIE), is a reference publication which provides access to the various report literature in the field of education. Because of the difficulty in covering journal articles, an additional index has been developed by the ERIC system, called the Current Index to Journals in Education (CIJE). This new publication makes available those periodicals that are of particular interest to practitioners in the field of education. It presently covers more than 700 publications which represent the core periodical litera- ture in the field. All of these articles, along with the educational documents previously listed, are covered by one of the 16 ERIC clearing houses. The heart of the ERIC system is the group of ERIC descriptors that are iden- tified in the thesaurus of ERIC descriptors which is 24 really a vocabulary deve10ped by specialists at the ERIC clearing houses. It uses an indexing system for the various documents, projects, and journal articles which are entered into the ERIC information system. These descriptors are based upon documents or journal articles previously indexed and currently included in the system. If one is interested in a particular topic, he or she would develop a search based on likely descriptors that might have been used in previous articles to describe the topic. This group of descriptors would then be used in what is referred to as an on-line computer search of the ERIC files. At the Michigan State University library, there are two experts who special- ize in the development of on-line computer search techniques of the ERIC system. Linda DeWitt is one of these experts and was of great assistance in this study as she helped the author develop the basic search mode which is described as follows. After careful examination of the various descriptors that were being used to describe the general area, it was decided that some twenty different descriptors be used for the broad-based search. These were divided into three major categories. In the first grouping the following descriptors were included: prediction, predictive measurement, predictor variables, multiple regression analysis, cohort analysis, simulation trend analysis, mathematical model, and models. As a result of this sort, some 9,189 possible articles and documents were identified by the computer search. The 25 second group of descriptors which yielded 4,170 possible sources included: enrollment, enrollment influences, college admission, enrollment projections, enrollment trends, and enrollment rate. The third group which yielded an unbeliev- able 45,382 possible sources included the following descrip- tors: higher education, post-secondary education, colleges, universities, church-related colleges, and private colleges. From this potential of some 58,000 sources, the methodology was simply to identify a relatively wide scope and then a relatively narrow scope literature search. By combining the three groups previously listed, it was possible to identify and later investigate a combination of 212 articles, which in essence had to include at least one descriptor from each of the three categories. This search then gave a much broader background than would necessarily be needed. The next step was to restrict the search to those works which would be most directly related. It was decided to confine the search by restricting the third group of descriptors so that it included only studies dealing with private colleges and/or church-related colleges. Of those two descriptors, 970 specific references were identified and examined; and when combined with the descriptors in groups one and two above, the identification of 18 references was developed. It is of great interest to the author to realize that in addition to the other search techniques such as the exami- nation of dissertations and related articles, that it is possible to develop through the ERIC system a methodology 26 which would include some 58,000 sources as a potential base of references and then restrict that to a very workable 200 references. As a result of this system, then, the informa- tion in the literature search seems to be much more thoroughly covered than would ever be possible through the more tradi- tional system of sorting through various card indexes and journal topical indexes. Enrollment Patterns As suggested by the descriptors discussed in the devel- opment of the methodology and outlined in the preceeding section, a major portion of the related literature deals with the overall enrollment trends for higher education during the last 15-20 years. Four different perspectives can be‘ specifically related to this study: first, the shifts in general overall enrollment for higher education; second, changes in enrollments for a private, church-related insti- tution; third, the variations and similarities between general enrollment patterns and those for church-related in- stitutions; and fourth, the anticipated effects on specific church-related colleges. Overall Enrollment Trends One of the most thorough and most recent articles relat- ing to the first general heading was published in January, 1978, by the Library of Congress. It was entitled, "Future 27 of Higher Educational Enrollments: An Analysis of Enrollment 17 Projections,‘ and was done by James Stedman. In the study, Stedman examined three recent demographically based enroll- ‘ment projection studies. The first of these projections was prepared by the National Center of Educational Statistics (NCES). The second set of projections was the one developed by the Carnegie Council on Policy Studies in Higher Educa- tion, while the third set was based on the work of Allen Cartter. Stedman's comparative analysis of these three studies is particularly important as it gives some indica- tions of the necessity of being aware of the enrollment pro- jections, and also points out the degree of difficulty of interpretation of these enrollment projections. The synopsis of these various studies, as Stedman sees them, are as fol- lows: the NCES study projects that total enrollments will grow by some 22 percent from 1975 to 1983 when they peak, and that by 1985 those same enrollments will have dropped by 2.1 percent from that 1983 peak. On the other hand, the Carnegie projections show a slightly different pattern with head count enrollments projected to grow 2.8 percent between 1930 and 1985 and then fall 4.1 percent before 1990. Enroll- ‘ments would then, according to Carnegie projections, grow during the last decade of the century by some 6 percent. The projections developed by Cartter, while he was at the University of California, indicate an increase in undergradu- ate enrollments up to 1982 when a decline period will begin which will shrink 7.2 percent by 1990 and continue to decline 28 through 1993 when a recovery period begins. James Stedman begins with a word of warning for the academic administrator as he attempts to analyze these pro- jections. At the outset, it should be noted that strict comparisons of these projections is particularly difficult. . . we would argue that only the trends described by the numbers should be directly com- pared, and then, only if care is taken to assess the assumptions used in the type of enrollment being‘measured. Stedman points out that the NCES projections are based on the assumption that the 1974 percentage of full-time under- graduate enrolhment of 18 to 21 year-olds will follow the same pattern that was established between 1965 and 1975; this is in contrast with the assumptions made as a basis for the Carnegie estimate of head count. The author goes on to sug- gest that the Carnegie projection shows that the decline will occur in the second half of the next decade rather than during the first half. Although this is perhaps significant, it is probably a result of a different treatment of that 18 to 21 year-old cohort which is examined in each instance. As a result then of a comparison of the first two sets of projections, the author stresses that both of these projec- tions are based on assumptions which do not include a rela- tionship to the labor market for college-educated individuals so that in the direct sense, they are both demographically based studies. In contrast to the Carnegie projection and the NCES projection, the work done by Cartter is influenced by changing demands for undergraduate enrollment in the labor 29 market. As a result of injecting the influence of the de- mand, Cartter's enrolhment projections tend to be slightly different from those of the preceeding analyses.19 The message pointed out by Stedman's concluding para- graph is painfully clear. He suggests that no single pro- jection of enrollment will supply all the answers for every institution, but it is apparent that those institutions who have depended on the traditional source of 18 to 21 year-old students will have to compete with other institutions for their share of that shrinking market. He observes that, . . in final analysis, 'many institutions will probably find themselves fighting the battle be- tween maintaining traditional methods and tradition- al missions and instituting change which may consztatite substantial redirection of those basic purposes. In his final word of warning, Stedman points out two things: The projections of change in enrollments do not give individual institutions much more than a warning that planning and analysis are in order. Demographic change, such as the decline in the 18 to 21 year-olds in the next two decades constitute only one of the many fgfces influencing higher education enrollments. The kinds of forces that will influence higher educa- tion enrollments in the coming years is suggested by a recent Bureau of the Census publication enrirled, "School Enrollments, Social and Economic Characteristics of Students."22 As a result of this analysis, the Bureau of the Census indi- cates that there are a number of forces that are influencing changes in college enrollments including shifts in family income, the changing size of the armed forces, the variations in the means of obtaining high school diplomas, the types of 30 degrees being sought by college students, and the availabil— ity of income generated through the "G.I. Bill." The con- cluding results of this study and of similar studies points out that it is not sufficient to analyze the demographic changes in the 18 to 21 year-old cohort without taking into consideration numerous other social and economic influences that are presently affecting enrollments in higher education. Private Church-Related Institutions The next relevant group of related literature suggests that the changing enrolhnent picture is one which is affect- ing the private church—related institution just as certainly as it is affecting the overall enrollment pattern. An exam- ination of the enrollment and admissions picture as reflected in the private sector, particularly the church-related sector is perhaps best introduced by the following quotation from an article by Bowen and Minter. Perhaps the most conspicuous mark of a healthy college or university is its capacity to attract and hold students and among the most important in- dicators signaling impending or actual distress are declining numbers and qualifications of applicants, incregging student attrition, and declining enroll- ment. Although the enrollment and admissions patterns seem to be somewhat mixed as to interpretation and trends, Table 2-1 reflects rather specifically the plight of the private col- leges and universities from the period 1969-70 through 1975- 1976. The indication is that the numbers of completed 31 woeuuooo canon. u< .n nun «an nan non non man «an consume ocean-«loo no sooueoouoa on ease-cum seasons» .u unn «on an~ nah was was «cu . eo>uooou aeo«u¢u««nes concumsoo no ensueoouua an ocean ease-sun osu cu cans-«eve no ououuo .0 con ca on don don can oou «vouuuaea .uouoeuuu use seen-cum..ous can. can 11 Isa-nausea: aeouq>usco usuuuaasu no noses: .n _ . 04 -~ and and «mg and and com «vouudnue canons». ususeuuu can 9» Isscuuuovcs aeouo>usoo osuuuansu no names: .c 11 «m cm om «o ea om cod accede case-cum one on censuses Aw nusoosu- uso~o>wsvn eauuuuasu uo nonasz .n T a a 3 2.. «S «S 2: «23.... 53.33 .5 3 sounowado venouuo «unsounmno «a Means: .N «a ea no am no «on can condo unannouu on noun-«av. you . . eo>uooou “souuaowaemq sou-45300 no amass: .« .uNmu «~34 «Nag mama «Nme dead o~a~ inhma umsau annma numaa Idnoa ranma amomn use» scan «a casua< «astound zuaodzh chlmoau .mhzaashm aa<=auza 52¢ muouAJou ua<>~ua ho muzmuaumuu Howard B. Bowen and W. John Minter, Private Higher Education. 1 (Washington, D.C. Association of American Colleges, May, 1976), p. 13. 32 applications are down along with the numbers of applicants offered admission to freshman class, while the number of full-time undergraduate transfer students has markedly in- creased. The observations can be summed up by suggesting that the trend-line is a declining one and that the mix of students in the student body is changing. Again Howard Bowen reflected the same kinds of conclusions in his posi- tion paper a year previous when he observed that the posi- tion in private colleges has changed little but that the projections are mixed with "the student body being somewhat less prepared, faculty salaries being somewhat hurt by in- flation, student-faculty ratios remaining constant and en- rollments up slightly in overall numbers."24 Continuing with the same line of reasoning, the article published by the American Council on Education suggests the following: Public and private education share many of the same problems . . . the private sector, however, is especially vulnerable to certain pressures because tuition is a primary source of its revenue. 5 The report examines the profiles of eight types of institu- tions as an attempt to offer a summary of the important dif- ferences between private and public institutions. In the report, important functional aspects are considered as they affect these differences and they include special roles such as church affiliation, differences in student clientele, and differences in academic character such as particular educa- tional viewpoints. The examinations of these functional 33 differences tend to highlight the variation in the roles played by public and private higher educational institutions. A further examination of the importance of tuition as a source of revenue for the private sector leads Richard Anderson to the following conclusion. Examination of the financial problems of pri- vate institutions must go beyond the readily avail- able statistics on enrollments and surpluses or deficits into how private colleges compete and what is happening to that gompetitive position.vis-a-vis public institutions.2 At this juncture, it would therefore seem valid to observe that although the base problem of admissions and enrollments are the same for the public, the private, and the private church-related segments of higher education, there are some specific differences that should be examined. 7.27 "Are church-related colleges losing students The question is, perhaps, directly answered in the opening com- ments of this study. Frequent statements in the public press suggest that institutions which continue to abide by their founding religious purpose offer programs and life- styles less and less attractive to today's American youth. These opinions have often led the general public to conclude that the church-related college has become a disappearing academic species. Facts related to enrollment permit no such generalization. These prognostications fail to take into considera- tion the significant fact that though the rate of growth in attendance in private institutions has slowed in relation to public sector since 1950, the number of enrollees has increased considerably.28 This rather extensive study conducted by McGrath and Neese analyzes the basic question of the relative position of the church-related colleges. They concur with Bowen and Miller 34 that the privately supported colleges are not in nearly as precarious a position as some professional analysts would suggest.29 The authors point out that private colleges and universities as a group do receive great quantities of state and federal funds and that there is a broad base of overall support for these institutions by the affiliated churches that control or influence their existence. Another major observation is the fact that some institutions have been able to survive and flourish even without increasing enroll- ments and have instead intentionally remained relatively small, concentrating on superior academic performance. Even though this is the general finding of the report, it is nec- essary to examine the enrollment trends in church-related colleges. An.examination of Table 2-2 would suggest that 201 presidents reported an increase in the number of full- time equivalent students from the year 1965-66 through the year 1975-76, while 122 schools reported a drop in full-tune equivalent enrollments for that same ten-year period of time. Even though this was a long-term positive upward trend, the 2.2 percent decline reflected for the 1975-76 year suggests enrollments are moving into that leveling off period sug- gested by the preceeding sources of data. Even with this apparent declining trend starting in the Fall of 1976 or 1977, the reasons for the enrollment growth of some church- related schools are complex. It was the opinion of McGrath and Neese that the chief administrators of these various colleges could provide valuable information on the topic of 35 Table 2—2 Statistical Summary* Number of church-related senior colleges and universities with enroll- ments of 250-2900 surveyed: 372 FTE Enrollment: Fall Fall # 7* (n) number of institu- 1965 1975 ° tions reporting 287,052 324,124-+37,144-+13 327 (882 of 372) Fall Fall # 7* (n) number of institu- 1975 1976 ° tions reporting 338,365 345,717 +-7,352 -+ 2 343 (922 of 372) Of the 327 colleges reporting for the 1965-1975 period, 201 (612 of 327) gained students, 122 (372 of 327) lost students, and 4 (1% of 327) re- mained unchanged. Taken by themselves, the 201 colleges reported an aggregate enrollment increase of 35 percent (+57,937 students) over the ten-year period from 165,924 students for Fall 1965 to 223,861 students for Fall 1975. On the other hand, the 122 schools reported an aggregate enrollment loss of 18 percent (-20,893 students) over the ten-year period from 118,383 students for Fall 1965 to 97,490 students for Fall 1975. Of the 122 colleges reporting a loss of students, 81 (66% of 122 or 25% of 327) sustained enrollment losses of 10 percent or more from 74,557 students for Fall 1965 to 58,202 students for Fall 1975; an aggregate loss of 16,355 students (22%). 0f the 343 colleges reporting for the 1975-1976 period, 197 (572 of 343) gained students. 142 (41% of 343) lost students, and 4 (11 of 343) re- mained unchanged. Taken by themselves, the 197 colleges reported an aggregate enrollment increase of 7 percent (+13,251 students) during the one-year period from 195,672 students for Fall 1975 to 208,923 students for Fall 1976. On the other hand, the 142 schools reported an aggregate enrollment loss of 4 percent (-5,799 students) during the one-year period from 139,378 students for Fall 1975 to 133,579 students for Fall 1976. *Percentages rounded to the nearest whole number. Source: McGrath, Earl J. and Richard C. Neese. "Are Clurch-Related Colleges Losing Students?" Arizona University, Tucson-College of Education, Topical Paper #6. 36 continued growth. A representative series of quotations from college presidents reflects the fact that such items as vigorous recruiting programs and new courses of study have increased their attractiveness to students; but most of these college presidents attribute the growth of their institutions to their religious affiliation and its corresponding moral and spiritual values. Perhaps the position of most of the college presidents is best summed up by Clark Kurr, as he writes the introduc- tion to "A Profile of Christian Colleges," done for the Carnegie Commission by Robert Pace. Pace finds the evangelical fundamentalist col- lege in some ways outside the mainstream of social change, at least as it is understood by students in the nation's colleges and universities. Yet, because of the strong assistance they receive from those who financially and spiritually support their educational philosophies, their future looks secure. So, ironi- cally, does the future of the colleges with very loose ties to their supporting church, and which are in some ways, more liberal than the avowedly non- sectarian liberal arts colleges. Perhaps in greatest difficulty are those main-line denominational col- leges that do not now seem committed to either a strong religious philosophy or a strong academic program. It was the opinion of McGrath and Neese as they drew to a close their observations about church-related colleges that "a large number of the institutions included in this study reflect the soundness of Pace's conclusion that evangelical colleges' future looks secure."31 They do offer a word of warning and some observations concerning those colleges and universities in the study which dropped considerably in enrollment over the past decade, and suggest that these 37 institutions should focus on their particular difficulties as they see them. They suggest that there are a number of possible problems which these institutions might be facing including . ineffective administrative leadership, poor recruiting efforts, ineffective marketing practices, the failure of administration and staff to envision new programs appealing to today's high school graduates, and their own possible strained 32 relationships with their respective church bodies. If this reflects the general position of the church-related institutions, it would be useful to examine the specific positions of two major denominations that have been actively involved in post-secondary education. One of the most recent and certainly most encompassing studies conducted by a major denomination has been done under the auspices of the Lutheran Educational Conference of North America, entitled, "Trends in Higher Education (A Review of the Literature)."33 It includes six monographs encompassing some 335 pages of text with extensive notes for reference. The overall position taken in the concluding monograph, "Im- plications for Undergraduate Colleges," suggests that the role and position of private higher education will be a vital part of this nation's educational system for many years to come. The issue of the future of private and church-affiliated institutions of higher education is examined from a number of different perspectives in the various monographs and summar- ized finally by the observation that: 38 . . if one were to base conclusions regard- ing the present importance and future possibilities of private higher education solely on the relative proportion of enroleents these colleges maintain within higher educational institutiopz as a whole one could become rather pessimistic. But as Carol Shulman observes, the role of private education in the United States makes two major contributions. First, it provides "variations in size, philosophy, curricula and communal feelings that are not generally available in public colleges"; and it provides a "preventative pressure against excessive governmental interference in the academic life of public colleges."35 The Lutheran study pointed out that practically every survey of higher education conducted in recent years has stated the vitally important role of the private system. This point was reinforced most convincingly by the Carnegie Commission in "The Capital and the Campus."36 If the basic conclusion drawn from this rather extensive study is that the independent, privately funded, church- related institutions will be a viable part of higher educa- tion, then they have a particular role that must be played. This role is perhaps best summarized by John Sibler, Presidem: of Boston University, when he said, . . the challenge to church-related colleges is to. maintain a certain distinctiveness while still acknowledging the broader role which is that of pre- paringimen and women to work for the greater public good. It is this rather optimistic view that is the concluding po- sition taken by the Lutheran denomination. 39 The year 1976 saw the publication of another similar document, edited by Michael Elliott and others, entitled "Toward 2000, Perspectives on the Environment for the United ‘Methodists and Independent Higher Education, 1976."38 This report reached a muchumore sobering conclusion as it examinei the position that the United Methodist church found its in- stitutions of higher education facing. The last part of this paper is, in the Opinion of this author, worth quoting as follows: With reasonable safety, it may be asserted that if higher education continues to deal only ‘with the primary clienteles it has served for the last two decades, its enrollments will no more than hold their own in thg 1980's and may very well drop precipitously. ' This is the starting point for this study which continues: While a gray picture for higher education is generally painted in this paper, the picture is darker still for independent and especially United ‘Methodists institutions, particularly in terms of enrollment outlooks. If past trends continue, the decline in the size of 18 - 24 populations in the 1980's will be a bad.omen for higher education and worseéstill for the United Methodist higher educa- tion. Even in light of this somewhat pessimistic outlook, the re- port makes some suggestions as to the role that church- related, higher educational institutions can play if they are to remain a viable part of the overall higher educational picture. Successful institutions will likely be those that carefully and continuously review their own operations in order to select the most effective programs and staff for continuance and expansion . this concern for the internal process of the 40 institution must focus on the question of quality and efficiency and is especially ezfential for any process for institutional renewal. It concludes with the observation that, . . . those institutions which can make a unique claim for their institutional purpose will likely find a market. Those institutions which can abandon the traditional views of institution- al autonomy and join together with others to share resources and creative ways may cope with their economic and dgmographic environments. New ‘models are possible. 2 Starting from.what was then a somewhat pessimistic overall view of the future of church-related and, particularly, United Methodists' higher education institutions, the report builds to a final rather opthnistic observation that, . independent and church-related institu- tions must not just sit back and let the future hap- pen to then; they must seize it and shape it. We must command the future if independent, church- related higher education is to continue to serve the vital social functions and czgrch purposes which they have traditionally pursued. The anlysis of the general enrollment and admission literature would suggest private church-related institutions need to pursue an analysis of potential markets using all the possible tools that are available to then. It is from this position that the next section of this study is to be examined. It consists of a study of the work that has been done with prediction and prediction models in an attempt to anticipate shifts and changes in general enrollment patterns, as well as an examination of specific enrollment goals for particular institutions such as Geneva College. 41 Prediction and Prediction Models: A General Background Controversies over the extent and duration of projected enroleent declines have made the tech- niques involved in such precitions a matter of interzzt to everyone involved in educational plan- ning. This editorial comment introduces a very complete, yet re- latively brief, article which surveys the current methodolo- gies involved in modeling and enrollment forecasting. The authors identify and briefly describe forecasting methods including ratios, cohort survival, Markov models, regression, optflnization, combination methods and guesstimations; these definitions are spelled out specifically in Appendix 2-1. The authors also discuss the use of different types of models being employed at national, state and institutional levels. The review of the development of models indicated by the authors identifies the fact that Markov models are presently being used at the national levels in Australia, Britain, and Norway. Lyell and Toole go on to point out the development of modeling at various state levels and suggest that the technique has progressed from simple ratio methods used in 45 Michigan in the early years to the more sOphisticated Markov models presently being used in the State of Washington?6 In.more recent years, there have been a.number of re- gression models developed to look at institutional problems such as the ones presently being used for student flowvmodel- 48 ing by NCHEMS47 and CAMPUS. Another interesting variation. 42 that uses a constant work concept as a definition of a co- hort model, has been developed by Marshall and Oliver at the University of California. This model will be examined in somewhat more detail later in this section. Perhaps the con- clusion reached by Lyell and Toole that best summarizes the prediction model techniques is summarized by the following: When it comes time to make a forecast that has to be lived with at the institutional level, model results must be tempered with insight and experience. Some of the methods may look well on paper but they are a waste of time and money unless they can pro- vide an administrator with more aficurate estimates than he could obtain intuitively. 9 With the work of Lyell and Toole as a general background, it is of interest to examine a few'of the models presently being used at various institutions for various purposes and then suggest the reasoning for the development of a specific 'model for Geneva College. A report entitled, "Projecting 0,"50 is typical of the Institutional Enrollments, 1974-199 type of projections being done by many state coordinating organizations. This one was conducted by the Minnesota Higher Education Coordinating Commission in St.Paul. It is part of the second stage of projections presently being car- ried out by that state and is described in the introduction section for the model as follows: In the projection model, resident, post- secondary students are conceptualized as flowing out of each regional high school, graduate cohort, as new entering full-time and part-time freshmen, i.e51 into each post-secondary institution in Minnesota. The reasoning and the approach to this type of projection is an attempt by the state of Minnesota to project a head count 43 of the number of students that are likely to be enrolled in ‘Minnesota higher educational institutions for the period 1972-1990. It is an example of a very traditional type of cohort analysis and was used, and is still being used, by ‘many state coordinating agencies. Typical of the types of student flow models and pro- jections that are being used across the country are the ones presently being employed in the state of Pennsylvania. An examination of two different studies indicates that the ‘models are becoming more saphisticated as the problems fac- ing administrators increase in intensity. The first study conducted in the state of Pennsylvania is reflected in a report published in July, 1973 and followed by an up-dated study in May, 1976. The 1973 report indicates the type of activity that was being conducted at the time and is reflec- ted in the following quotation from the overall summary. The patterns of movements of students in a multi-campus institution of higher education had a significant impact upon the magnitude and major ififiiiifincT?°?iii°ls°§fi$§jEffigfi‘éfinii 222152; institution be able to measure and predict these patterns of flow and to reflect tggm in their projections over the enrollments. An examination of this study indicated the necessity for the development of a‘Marovian-type of model that would project the needs of the Pennsylvania State University System in anticipation of student enrollments. The model developed was an offshot of two previous models; but, as Tillman and Newton suggest, this newer model differed from its prede- cessors in that it had a higher level of what they referred 44 to as "disaggregation." ‘This'made it suitable not only for the state-wide system, but for a particular university cam- pus. It did this by providing a relatively efficient way to project future enrollments as well as aiding in the data that would be necessary in anticipating various instructional loads at the institution. This model was interesting con- ceptually in that it was one of the earlier applications of the Marovian approach for enrollment projections, and led to the development of the model used in the report published in 1976. As patterns of college enrollment continue to be- come more complex, the changing composition and magnitude of student needs required new projection techniques. By itself, the changing characteristics of our population indicate the need for a new and some- what more complex model for projecting enrollments, one which recognizes a variety of different popula- tion sectors and their unique characterized patterns of participation in various educational experiences. The preceeding quote indicates the reasoning for the estab- lishment of a new; more sophisticated prediction model at Penn State University in an attempt to cope with the ever— more frustrating "what if" types of questions that were being asked by administrators as they attempted to accurately allo— cate resources and develop programs for this ever-shifting market. This was an attempt to develop a model which would differentiate among this wide variety of p0pulation sectors and, more accurately, project enrollment patterns that are a result of changes in population compositions. 45 The state of Maryland has been one of the more progres- sive states in attempting to project future enrollments based on a number of different models used over a considerable period of time. The development of these various projects is reflected in a document entitled, "Enrollment Projec- "54 This is a rather extensive study and includes tions. the history of the development of projection.models in the state of Maryland starting with the earliest model developed for the state by Peat,‘Marwick, Mitchel, and Company in 1968. Following that model, various other more sophisticated models were developed and utilized for prediction purposes. The present model is basically a cohort survival model but a rather sophisticated one that consists of three different modules which reflect various cohorts that are deemed import- ant by the Maryland system. The document by Gustafson and Hample, in addition to an examination of the history of the model, is involved with more specific technical data includ- ing the logistic curve of the model and the description of the calculation procedures. It also includes a manual for the running of the model on facilities such as the Univac 1180 system. An examination of related literature in the field of the development of prediction models would not be complete with- out reference to the work that has been done by the University of California which was started at a rather early date and reported in two publications in 1968 and 1969. The 1968 re- port, authored by Rs M. Oliver, entitled, "Models for 46 Predicting Growth Enrollments at the University of Califor- published by the University of California, Berkeley, under a Ford Foundation Grant, is the first of the two re- ports which should be mentioned. It was an extensive program drawing on the best of the enrollment forecasting techniques that were available at the time and was one of the early adaptations of the Markov models developed by Gani in his work in Australia. This Australian model was completed in 1963, and Gani later adopted a revised model for use at Michigan State University in 1965. The specific model that was finally developed and used by the state of California has been referred to as the Grade Progression Ratio method (GPR). This model was considered to be a more sophisticated approach that accounted for a higher degree of accuracy when predicting growth enrollments for the system. The second model was developed by K. T. Marshall and Robert M. Oliver and is described in "A Constant Wbrk.Mode1 for Student Attendance and Enrollment."56 This model was developed by the authors in an attempt to predict undergraduate student attendance. It relies on five different parameters, one of which is referred to as a total work parameter which is really an analysis of the relationship between the amount of work previously completed and the likelihood of enrollment in the University of California. Marshall and Oliver found that they could rather accurately predict the probability of a student graduating based upon the total work completed toward the degree. In other words, the more work a student had 47 previously completed, the more likely he was to re-enter once he had dropped out of the institution. This is another indi- cation of the degree of saphistication in the development of various modeling techniques. In addition to using models to come up with enrollment estflnates for state systems and universities, models have also been used in an attempt to predict certain shifts in seasonal demands. Two such attempts are identified in the literature and should be examined at this point. Federal City College in Washington, D.C., conducted a study in an attempt to develop a model which would help them predict the demand for required freshmen.courses. This study was carried out by T. S. Chidasbaras and entitled, "Enrollment Forecasting in an Open'Admissions.Environment."57 The paper discusses the utilization of a cohort survival model in an attenpt to forecast the required number ofsections that would be needed in various courses to meet the demands of freshman year enrollment. The last study to be examined under the general heading of enrollment models is one that was published in 1977. This study reflects the increasing complexities of enrollment predictions and the fact that they are becoming evermore important in establishing programs at various types of institutions. One of the more recent prob- lems that administrators are facing is an attempt to differ- entiate between long-term trends and short-term variations that tend to break established trend lines. The misinterpre- tation of this data can cause the misallocation of resources, 48 space, and staff. The attempt by~Georgia State University to deal with this was the reason for the development of a model discussed in the paper, "Helping Administrators Identify Shifts in Enrollment Patterns."58 The model and the report that it generated dealt with the following types of problems: the identification of past seasonal, cyclical, and trend variations in enrollments; the correlation of cyclical enrollment variations with the business cycle; and the deve10pment of a simple regression model to predict short-run variations. This model is one that should be useful in helping administrators focus on long-range as well as short-term program demands. Enrollment Versus Admissions The literature search revealed limited amounts of infor- mation on decisionumodels which attempted to deal with actual enrollment predictions for specific institutions. The review of enrollment models dealt primarily with large- scale aggregate figures rather than specific enrollment de- cisions of individual students considering a specific institution. The reason for the lack of information or the few number of prediction models dealing with actual enroll- ‘ment decisions is suggested by Richard Spies. Analytically, a student's choice of a college should be treated as two separate (but not inde- pendent) decisions. The first decision is where to apply. The second is which college to attend among all those to which he was admitted. A great deal of research has been done on the college 49 selection process but most of it has failed to dis- tinguish between the application and the enrollment decision . . . The enrollment decision presents a number of complex analytical problems, the most serious of which is that a student's decision to enroll at a particular school depends heavily on his alternatives, the other schools to which he has been admitted. It would be very difficult to account for the presence of these alternatives in the emperical model. The application decision is much simpler.5 Spies goes on to make some observations about the application decision in comparison with the enrollment decision. He suggests that the application decision is one that is quite different from the enrollment decision and suggests the following: we have found that students in the higher ranges of academic ability generally select the type of school they wish to attend and then appéa almost exclusively to schools in that category. Given the position taken by Richard Spies, what the studies have in reality been examining are the results of the appli- cation process as a basis for the models. A cohort survival technique superimposing a given percentage of the students that have applied to an institution adds the "enrollment" dimension to the model. This, then, is a statistical tool defining enrollments. It must be pointed out that from Spies‘ point of view, this is a.much different process than attempting to predict the actual "enrollment decision" as it is madeby a particular student as he considers one or more academic institutions. It is this "enrollment decision" pro— cess which is the basis of the prediction model being devel— oped in this study. 50 "The Carleton Application Pool: An Emperical Study"61 was conducted in an attempt to identify the variables which address the "enrollment decision" process and therefore lends credibility to the development of an enrollment prediction model for Geneva College. The introductory comments in the Carleton Study are concluded with the following quote: 1. Given the fact that the Carleton Admissions Office receives approximately 7,000 inquiries each year from prospective students, why is it that less than 20 percent actually submit an application? 2. What aggects an accepted applicant's deci- sion to attend? It is an examination of point number two in the Carleton study that is of particular interest and bearing on the Geneva prediction model. The reason for examining what the Carleton study re- ferred to as yield, i.e., the number of students that actually enroll compared to the number of students accepted, is the feeling that if the yield were increased, Carleton could be more secure in its recruiting position. It was also felt that the study of those factors which affect yield would aid in the evaluation of the admissions process. A similar set of reasons have been suggested by the administration of Geneva College. The methodology of the two studies, Carletads and this study, is somewhat different. Carleton College used a questionnaire to identify the reasons given for the deci- sion to either enroll or not to enroll while the Geneva pre- diction model is based on the actual historic results of comparing the confirmed students with the withdrawals. 51 Operationally, the two studies are really quite similar. In the analysis section of the Carleton study, the authors iden- tify the following crucial question: "The question that needs answering, however, is: Do acceptances who differ on any of the above background factors enroll at Carleton in ?"63 The approach.used to identify different proportions these variations is a series of hypotheses that attempt to find statistical differences between the enrollees and those that do not enroll. In answer the authors conclude, "Analy- sis shows that they do in a couple of ways; differences in SAT scores and in regions of residences produce different "64 An examination of the data shows that yield rates. Carleton did a better job of enrolling students from the ‘Midwest and the South than they did from the geographically ‘more removed western regions and Eastern regions. One of the hypothesis examined in both the Carleton study and in the Geneva prediction model concerned the influ- ence of Admissions representatives. In the Carleton study there was a positive correlation between enrolhments and con- tact by Admissions people. The result was that those studenUs who were contacted were 29 percent more likely to enroll at Carleton. Another hypothesis that has been identified in both studies dealt with the influence of campus visits on student enrollments. It is significant to note that 65 per- cent of the acceptances who visited the campus enrolled at Carleton compared with only 41 percent of those who did not visit the campus. Another area that is of interest to 52 Admissions personnel is the area of financial aid. To quote from the Carleton study, "no study of yield would be com- plete without considering the impact of financial aid."65 An examination of the information from the Carleton study suggested that Carleton's chance of enrolling a student was 26 percent higher when the college offered aid to that student. Another area that was examined in the Carleton study which has implications for the Geneva prediction model concerns the influence that parents have on students‘ enroll- *ments at particular institutions. It was found in the Carle- ton study that the parents were often the first source of information about Carleton and had a positive effect on the student's decision to eventually enroll in the college. The final conclusions of the Carleton study is particu- larly relevant to the Geneva study. Further, it is obvious from the yield study that certain types of actions could increase Carle- ton's yield. ‘Most would involve increasing the percentage of positive contacts students have with aspects of the Admissions process and decreasing the percentage of negative contacts. While this ‘may sound like a trivial conclusion, the greatest value of the yield study lies in showing which of the many things that could affect yield actually do, hOW'tO change the things which affect yield for the better is not something which can be deduced from this particular research. The Carleton study provides a very sound and valid conclusion to the literature search as it relates to the Geneva study. CHAPTER 3 PLANNING AND DEVELOPING THE STUDY Chapter 2 provided the background which will enable the reader to understand the need for the development of an "enrollment decision" prediction model rather than the more general projection.models that are reflected in the review of the literature. Chapter 3 will examine the planning and development of this study. Papulation A sample, made up of and defined as those students who applied to Geneva College during the 1976-77 school year for admission in 1977, is used as a basis for this study. This population was chosen as a representative sample of those students that will be applying to Geneva College in future years. The sample can be subdivided into various groups. The first group is those students who applied to Geneva College. The second group is those students who.having applied to the College,are accepted by the College. The third group is those students who, after being accepted by the College, de- cided to enroll and confirm their intentions to do so, hereinafter referred to as "confirmed." The fourth group is those students who, having been accepted by the College, 53 54 decide not to enroll and withdraw from the admission pro- cess, hereinafter referred to as "withdrawals." Another group that is not included in the analysis are those stu— dents who applied to the College and were denied admission. Of the total 673 students who applied to Geneva College during the 1976-77 year, 124 of them were declined admission. This left a total of 549 students who were accepted by the College. Of that group, 330 students confirmed their inten- tion of enrolling at the College while 219 withdrew from the admissions process. It is hoped that by examining and comparing the con- firmed and withdrawals that a sufficient body of knowledge ‘will be developed so as to enable the Admissions personnel to better determine those students who are most likely to fall into those particular categories in the coming years. Procedures and Instruments As indicated by the hypothesis and variables found on page 19 and 20 of Chapter 1, it was decided to develop a procedure made up of two different parts. The first part was the development of a prediction model which would use the 13 independent variables to predict the two dependent variables. This first part is implicit in Hypothesis 1. The second part consisted of a series of tests for signifi- cant difference between the dependent variables defined in terms of each independent variable, as implied by Hypotheses 2 through 14. 55 The examination of the specific hypotheses would sug- gest the following: Hypothesis 1: That confirmed students can be predicted significantly from a set of variables. This hypothesis simply attempts to determine whether or not the status of the student, confirmed or withdrawal, can be significantly predicted from the 13 independent variables. This hypothesis includes the development of the prediction model discussed in greater detail under the Methodology and Analysis of Data section of this chapter. Hypothesis 2: There are no significant differences in the term of matriculation between con- firmed and withdrawals. In considering the term of matriculation as it relates to those students who are likely to enroll at the College, it is the general feeling of Admissions people that the students who apply for admissions in the fall term are the group that are most important simply because of sheer num- ber of applicants. Hypothesis 3: There is no significant difference in the dates of the first campus visit between confirmed and withdrawals. The whole topic of campus visitations always seems to be of great interest to Admissions people. It is felt that if you can get the student to visit your campus, he or she 56 is much more likely to ultimately enroll at your school. It is also generally felt that the student who visits the cam- pus when the campus looks its best is more likely to enroll. Hypothesis 4: There is no significant difference in the acceptance date between confirmed and withdrawals. There is always a great urgency in answering students' inquiries for information. There is also the feeling that the earlier a student is informed of his acceptance, the more likely it is he or she will enroll at a particular institution. Hypothesis 5: There is no significant difference in first majors between confirmed and withdrawals. Over a period of time, certain majors are more popular than others. These tend to follow cyclical patterns and the emphasis is always placed upon those majors that are most popular at the time. Again, the general consensus is that students applying in the popular majors are the most likely to enroll. Hypothesis 6: There is no significant difference in the source of contact between confirmed and withdrawals. With the decreasing demand for the services offered by a college, there has been an increasing emphasis placed on 57 recruitment procedures and practices. There is a general belief that personal contact is the best way to assure that a student will enroll at an institution. If this is the case, then there is justification for the increase in the importance being placed on recruitment, alumni departments and public relations. Hypothesis 7: There is no significant difference in type of student between confirmed and withdrawals. An examination of the enrollment pattern at Geneva College would suggest that there is a marked change in the type of students that have enrolled at Geneva College over the past 10 years, with a decline in the commuting student and an increase in the resident student. Hypothesis 8: There is no significant difference in denomination between confirmed and withdrawals. As Geneva is a church-related liberal arts college particularly interested in students of reformed evangelical church affiliation, it is felt that those students who are ‘most likely to attend Geneva College will come from that type of common background. Hypothesis 9: There is no significant difference in distance from home between confirmed and withdrawals. 58 A college such as Geneva with a limited recruitment budget is naturally interested in the location of its ”mar- ket". It is a great deal less expensive to recruit students that are relatively close to the campus than it is to re- cruit students at some distance from the campus. Hypothesis 2 through 9 deal in general with non—acadenic demographic variables in that those independent variables examined in Hypotheses 2 through 9 are not directly related to academic achievement or institutional academic character- istics. On the other hand, Hypothesis 10 through 14 deal with, or are based upon,past academic characteristics; and therefore are somewhat different in this regard. Hypothesis 10: There is no significant difference in type of scholarship aid between con- firmed and withdrawals. Hypothesis lO examines the relationship between types of financial aid offered different groups of students and the probability that these students will enroll at Geneva College. Often one hears the comment that a college could recruit better students or more students if a specific scholarship were available for a particular purpose. The use of scholarship funds is of ever-increasing importance in recruitment. This hypothesis examines the most effective use of scholarship funds. Hypothesis 11: There is no significant difference in class rank between confirmed and with- drawals. 59 Hypothesis 12: There is no significant difference in class size between confirmed and ‘withdrawals. Hypothesis 11 and 12 deal with factors that most insti- tutions think are relatively important. There is some evi— dence that class size is an indicator of interest in a particular sized institution. Class rank is thought to in- dicate that a student would tend to enroll in an institution of particular academic quality. If there was a correlation in either case, it would be very helpful in the recruitment process. Hypothesis 13: There is no significant difference in the SAT verbal scores between confirmed and withdrawals. Hypothesis 14: There is no significant difference in the SAT math scores between confirmed and withdrawals. Each institution would like to think it is capable of attracting students with good college board scores. It would be rewarding to any institution to find that their academic program satisfies the needs of those students with relatively high college board scores. Methodology and Analysis of Data The data used for the study are collected as a normal part of each applicant‘s record. These data are already 60 available at the institution and the only additional cost would be the cost of analysis. The information being readily available would suggest that it should be analyzed in an attempt to determine if there are any statistically signi- ficant differences between confirmed and withdrawals, and if those differences are predictable. The available data on each student that applied to Geneva College during the 1976- 77 year for admission during the 1977 school year was key- punched on IBM cards by the Geneva College Computer personnel. The data were sorted into the necessary categories and grouped together so that it could be easily handled as it related to the various specific hypotheses. The data were analyzed on the Michigan State computer system using the Statistical Package for the Social Sciences (SPSS). The evaluation of the data uses Pierson Product Moment correla- tions, multiple regression analysis, t-Tests, chi-squared techniques and summary and percentage computations. The .05 level was used in all tests as the level of acceptance statistical significance. In Hypothesis 1 advantage is taken of the multiple regression analysis that is available on SPSS as a tool for development of the Enrollment Predic- tion'Model. For Hypotheses 2 through 10, chi-squared techniques will be employed to determine significant differences for validating the hypotheses. For Hypotheses 11 through 14, ‘t-Tests will be used in a similar manner, and where appro- priate, means and standard deviations will be included. 61 The,Mode1 A model is: "An equation or set of equations depicting the causal relationships that are believed to generate ob- served data. Also, the expression of a theory by means of 'mathematical symbols or diagrams."66 The Enrolhment Prediction model developed in Hypothesis 1 and used in this study is defined as follows: = + . X + Y A iJBj j e Y = the criterion measure; A = a constant term; B = net regression weight that determines the relative effect on Y of a specific X value; X = specific predictor variables; e 3 an error term. This model is used to describe the effect that various specific predictor variables have on the dependent variable being examined. The criterion measure, Y, would represent the dependent variable and has been given an arbitrary numerical value. The expression ZijXj represents the effect that the independent variables have on the constant term A. As A is affected by the expression ZijXj moves closer in value to Y the degree of predictability increases. When the value A equals the value Y,one has perfect prediction. 62 Limitations Each private, church-related college is a unique insti- tuion and as such must develop Admissions programs which are suited to its own particular needs. The body of knowledge, including the enrollment prediction model, developed in this study was based upon data available from and specifically applicable to Geneva College. The Enrollment Predictor Model developed for Geneva College may not meet the specific enrollment needs of another institution due to the variations in Admission goals. As a result of this study it is the opinion of this author that the Enrollment Prediction Model developed for Geneva College is applicable to the general needs of other private, church-related colleges attempting to meet the needs of the same or similar student market. Summary In this chapter the design of the research methodology was presented. The sources of data and the methods of ana- lyzing that data were examined. The various hypotheses utilized in the study were identified. Included in this chapter was the definition of the Enrollment Prediction.Model developed in Hypothesis 1. The research instruments used in the study were discussed. The ‘methods of statistical analysis used to develop the predic- tion model and to test the other hypotheses were outlined. The methods were multiple linear regression, Pierson's 63 Product Moment Correlation, chi-square tehniques, and t-Test. The lunitations of the study were also identified. CHAPTER 4 ANALYSIS OF DATA ,Forward After the presentation of the importance of the problem in Chapter 1, the review of the past and current status of Admissions projection in Chapter 2, and the presentation of methodology of the study in Chapter 3, this chapter now turns to the analysis of the data. The chapter is divided into three parts: Part I analyzes the results of the prediction model developed in the first hypothesis. Part II analyzes the data as it relates to the remain- ing hypotheses, testing the significance of the independent variables in each case. Part III analyzes the results of the prediction model and the information generated by the other hypotheses which give direction to Admissions Programs at Geneva College. Part I Hypothesis 1 - The Model Hypothesis 1: The confirmed student can be predicted significantly from a set of variables 64 65 An examination of Tables 4-1, 4-2, 4-3 show that the con- firmed student can be predicted with a level of statistical significance above the .05 level. Both forward and backward multiple linear regression techniques were used to test the hypothesis. Using the general model: Y=A+ .B.X.+, tJJJ e data were analyzed and only statistically significant pre- dictors were included in the final model, the results of which were: - 0.531 X Y = 10.008 - 0.490 X - 0.507 X l 2 3 Where, Y = Predicted Y - the Criterion measure. X1 = Commuters: (Commuters as l, 0 All others). X2 = Asummer: (Asummer as l, 0 All others). N II 3 Referrals: (Referrals as 1, 0 All others). In descriptive terms this model suggests the following: that when "confirmed" status is given an arbitrary value of 9, and "withdrawal" status is given an arbitrary value of 11, the probability that a particular student would enroll at Geneva College merely by chance is reflected by the con- stant value 10.008. Certain variables tend to move the constant value closer to 9 and further from 11 and would increase the prob- ability that a student would confirm his intentions of en- rolling at Geneva. It is those predictors that this study 66 Table 4-1 Variables in the Final Regression Equation Variable Commuter Referls Asummer (Constant) Net Regression Weight Significance -.h90 .000 —.531 .009 -.507 .001 Overall F - 11.90039 Significance - 0.000 67 Table 4-2 Variables not in the Final Regression Equation Partial Regression Variable Weight Significanqg T.Fall .02816 .600 T.Expm. 1.00000 1.000 T. Spring -.04358 .416 T.Summer .00031 .995 Visit -.05497 .305 A.Fall -.05475 .307 A.Winter .06796 .205 A.Spring -.02297 .668 Bible .0418? .435 Business -.05693 .288 Engineer .06564 .221 Physci .06482 .226 Soc.Sci. —.04452 .406 Arts -.01293 .810 Educ. -.00253 .962 Other -.03936 .463 Resident .02675 .618 PartDay —.02882 .591 Eve.Sch. -.01492 .781 Protest -.00920 .864 Evangel -.00810 .880 Catholic .02714 .613 Non.Demo. -.00767 .886 Test -.10038 .061 Noid .06672 .213 Direct .03337 .534 CampLife -.o3973 .459 Magazine ~0755h .159 Conferen .08102 .130 Persons .07374 .169 Rank 207390 .168 Size .03552 .508 Verbal -.08117 .130 Math .03933 .463 Zip -.02121 .693 68 moo. n oeseoacasmsm omomo.a [.m1wssso>o mmaoo. HmeH. mwfloa. awn. zoom omcoo. maowa. waooa. mww. muflhmm.e omaoo. mwmmd. memmm. mom. QHN mmmoo. momma. mmhmm. Hom. nooowmum oomao. manna. oomom. moo. Henso> mgooo. mamzd. mowvm. mmw. ofiaospoo mowao. Abazd. mzmpm. :Ho. uoassm.< :maoo. «mama. Nommm. :mm. Hamm.< mHHoo. momma. moomm. omm. some ommoo. monH. eooem. wow. sespo wmooo. mmmHH. ommam. mmm. mvhd ahaao. Nmmaa. mmmzm. mzo. mauomom ooooo. mseod. mmemm. moo. .osmo.soz Nmmmo. QJNOH. :mhmm. HOO. hopfifiaoo ooooo. omamo. Haoom. oeo. .Hem.eom mamao. mmaoo. eommm. New. meomsmm m0boo. mommo. movmw. :40. new: mmaoo. mommo. mawam. Ham. poohfim ommoo. sosmo. oemmm. mom. venom oaooo. mommo. mommm. mes. .eem.u>m moooo. hmamo. mmmwa. hem. pmop0hm mmhoo. :mamo. armba. ooa. oohomooo mmooo. Nozmo. magma. cam. pamw> wmmoo. mammo. mmmma. had. oufluowdz armoo. marflo. wmoma. :ba. .Hom.h£m Hmooo. :mzflo. mvaH. th. muwm oomoo. msmao. mamas. mmo. escapee haooo. mzaao. ommoa. mmw. oqumado omaao. omaao. memos. sew. Haoa.e uwoono meadow m ousdmw m m mamwpasz ooosowwwumHm mapwflhw> mopspm to assassoeoaoosm weapooees moaossss> seammosmem oaoepasz MI: oHQmB 69 is attempting to identify. In this study the three predictor variables, "Commuter", "Asummer" and "Referrals", all showed negative net regres- sion weight factors which would thereby reduce the constant and move the value closer to 9, the numerical value assigned to the confirmed status. (Tables 4—1 and 4-4) In summary it could be concluded from this study that a student who was (1) applying as a commuter, (2) accepted in the summer prior to fall term, and (3) referred to the College by an alumni, faculty, or other friend of the insti- tution, would more likely fit into the "confirmed" status than a student who did not meet these criteria. An extensive list of independent variables was examined, those identified in Table 4-1 and 4-2 showed significance in the prediction process. This list is an extension of the list of general predictor variables found on page 1%) in Chapter 1. Tfall, Texpm, Tspring, Tsummer - indicates term of ‘matriculation, Tfall = Fall Term and etc. Visit - indicates the student visited campus. Afall, Awinter, Aspring - indicates the time of the year the student was accepted, Aspring = the spring of the year. Bible, Business, Engineer, Phy. Sci. Soc. Sci., ArtsL Educ., Other - indicates an area of academic interest or major, Arts = an English, History, or Music major. 70 Table 4-4 Correlation Coefficients for Variables in the Multiple Regression Analysis Independent Variables Affecting Status Variable , Status Variable,__- Status T.Fall .10612 Commuter -.20779 T.Spring -.08834 Resident .21700 T.Summer -.o6387 PartDay -.oh398 Eve.Sch. -.03382 A.Fall —.04138 Protest. .00023 A.Winter .11591 Evangel. .00822 A.Spring .0469? Catholic .00110 A.Summer -.l8229 Non.Demo. -.01572 Bible .05128 Test —.10243 Business -.02537 Noid .05503 Engineer .07014 Direct .05696 Phy.Sci. .05675 CampLife .01252 Soc.Sci. -.04268 Magazine .08618 Arts -.04207 Conferen .08521 Educ. —.03948 Persons .11241 Other -.02940 Referls -.1h771 Rank .06759 Size .03402 Verbal -.07655 Math .05122 Zip .06398 Visit -.OlO97 Resident, Protest.L 71 Commuter,_Part-time, Eve. Sch. - indi- cates the student will live on campus, commute from home, be part-time, or in evening school. Evangel., CatholicL Nondeno. - indicates the student's religious affiliation, i.e., Protestant, Nondenominational, and etc. Verbal and Math - indicates the College Board test scores of the students. Rank - indicates the student's rank in his or her high school graduating class. Size - indicates the size of the student's high school graduating class. Zip - indicates the location of the student's home by U.S. Postal zip code zones. The following variables describe the source of firSt contact with the College: Referrals - indicates a person not directly associ- Persons - ated with the recruiting process such as an alumni, friend, pastor, or faculty member. indicates a member of the institutions whose job deals directly with recruit- ment, i.e., admission counselors, athle- tic coaches, or religious service personnel. Test - indicates the American College Testing Service. 72 Direct - indicates any one of a number of college directories that are available as refer- ence sources. Magazine - indicates an advertisement in a "student" magazine. Letters_- indicates letters sent out by the Admis- sion Department. Noid - indicates non-identified prospective students. Conference - indicates conferences held by admis- sions personnel at various high schools and churches. Decision: The hypothesis is accepted. Part II Hypotheses 2 through 14 - Tests of Independent Variables Hypothesis 2: There is no significant difference in the term of matriculation between confirmed and withdrawals. An examination of Table 4-5 shows that there is a sig- nificant difference between confirmed and withdrawals as they relate to their anticipated terms of matriculation. By far the most interesting observations to be drawn from this table are reflected in the data for those students that have applied for admission in the Spring term and Summer term. The table would indicate that those students who have been accepted for admission in the Spring term all 73 Table 4- S The Effect of Anticipated Term of Matriculation on Status--Confirmed vs. Withdrawals % 1 Term of Matriculation Confirmed Withdrawals (N=33o) (N=219) Fall Term 59.0 41.0 Spring Term 100.0 0.0 Summer Term 87.5 12.5 Chi-square = 8.737 Significance = .0127 74 confirmed and entered the College. This is educationally interesting and implies that this group of students is look- ing for a college that will meet their immediate needs, and ‘may well be transfer students. These students may have considered Geneva as a second choice when they were shop- ping for colleges earlier in their career and having not found what they wanted at another institution, are much more certain as to their needs. The figures comparing con- firmed and withdrawals for admission in the summer program again reflects the fact that this group of students is much more likely to confirm and enroll at the College, perhaps as a result of having made up their minds well in advance and wanting to get started on their academic programs. As a result of this first table, it would seem that the Admis— sions Department would want to pay special attention to those students who apply for admission to either the spring or the summer term because of the markedly higher percentage of confirmed status students. Decision: The hypothesis is rejected. Hypothesis 3: There is no significant difference in the date of campus visit between the confirmed and with- drawals. Table 4-6 shows the result of visiting the campus as it affects a student's decision to enroll at Geneva. The data show a statistical significance between confirmed and 75 Table 4-6 The Effect of Date of Campus Visit on Status—-Confirmed vs. Withdrawals Date of campus Visit Confirmed Withdfawals (N=326) (N=217) Fall 66.? 33.3 Winter 50.0 50.0 Spring 79.2 20.8 Summer 61.5 38.5 Chi-square = 10.1723 Significance = .0321 76 withdrawals as related to a campus visit. In addition to the statistical significance of the data in general, there are some very important educationally significant differ- ences depending on the time of the year the campus visit is made. The first thing that becomes apparent is the fact that there is no difference between confirmed and with- drawals if the campus visit is made during the winter months. This reflects the fact that the student might be seeing the campus at a period of time when it is not particularly attractive or any one of a number of other reasons. 0n the other hand, it is apparent that those students who visit the campus during the spring of the year are much more like- ly to be included in the confirmed category. Spring is therefore the most desirable tine to visit the campus. It would be fair to conclude that everything possible should be done to encourage the student to visit the campus in the spring or early summer, and that particular care should be taken to see that there be a very thoroughly conducted and informative program. Decision: The hypothesis is rejected. Hypothesis 4: There is no significant difference in the acceptance date between confirmed and withdrawals. The data in Table 4-7 indicate that there is a statis- tically significant difference in status between confirmed and withdrawals as affected by the date of acceptance. An The Effect of Date of Acceptance 77 Table 4-7 on Status--Confirmed vs. Withdrawals Date of Acceptance Confirmed Withdfawals (N=326) (N=217) Fall 67.0 33.0 Winter 57.6 42.4 Spring 50.3 1‘9-7 Summer 78.3 21.7 Chi—square = 20.528 Significance = .0001 78 analysis of the data show' the following significant dif- ference. The normal practice of high school students would be to apply to and expect an answer from a number of col- leges during the fall and winter months of their senior year in high school, thereby hOping to have a broad choice for the fall of the next year. This group, reflected in the categories of winter and spring acceptance dates, shows very little difference between confirmed and withdrawals, indicating that these students are shopping and have had a enumber of choices of institutions that are willing to accept them. On the other hand, the other two categories indicate a quite different set of circumstances. The group of stu- dents in the category of fall acceptance have apparently shown an early interest in the institution, have applied earlier than normal, and if accepted are more likely to confirm and enroll. In like manner, the relatively high percentage of confirmed students in the summer acceptance period reflects a group of students who have, for any one of a number of reasons, applied to Geneva at a relatively late date in hopes of acceptance for the fall term which is only a month or so away. Again, this group appears to be much more likely to confirm and enroll if they are accepted. The whole question of the acceptance of late applicants is another major Admissions problem, but the statistical indi- cation is that those students who are accepted late in the session are the most likely to confirm and enroll at the College. This is verified by the data developed by the 79 prediction model in Hypothesis 1. Once again, this indicates special attention should be paid by the Admissions personnel to those students who are outside the normal pattern. In other words, they should watch those students who either apply early and are accepted early or those students who apply late and are accepted at the last moment. Both of these should be handled with par- ticular care as the instance of them enrolling at the Col- lege appears to be much greater. Decision: The hypothesis is rejected. Hypothesis 5: There is no significant difference in choice of first major between confirmed and withdrawals. Table 4-8 shows the result of the examination of the effect of various majors on enrollment at Geneva College. There is no statistical difference in confirmed and with- drawals for the indicated first major category in total, but a closer examination indicates some meaningful educa- tional differences. As the popularity of various academic programs shifts (over a period of time) from one program to another, the recruitment emphasis as well as college curricula and new programs seem to follow. Today, Business Administration and Accounting, as well as Engineering, are very popular and are in great demand. On the other hand, Education and the Liberal Arts seem to be somewhat out of vogue. A 80 mmmm. u oouoowmwuwwm m:m.m u macadmlfino m.mN :.:> mph< Hmhopwq o.:: o.mm onaooufiwom m.mm P.mm oowpoodom w.mm N.mm oouowom Hofioom e.os m.mm mosesem steamers o.mm o.m: oHpHm H.mm m.mm our soapmhpmfiowsowuwwMMMMMM N.H: w.wm dvdwovdcb Aonqu Aommuzo massesospaz eessausoo sane: pesos oepeeaosH u u madawnonpwz .m> ooahwmooollmdpopm no sons: poses oopaesosH to scream use mu: eases 81 comparison of the confirmed and withdrawals in those areas would indicate that while Business Administration and Accounting as well as Engineering have ratios of close to 2—to—l favoring the confirmed category as compared to with- drawals, the areas of Liberal Arts and Education are doing even better and that Liberal Arts is running a 3-to—1 ratio favoring the confirmed over the withdrawal category. Another worthwhile observation, particularly for a college such as Geneva, which prides itself on being a Christian college with a strong Bible Department, is the fact that the ratio of confirmed to withdrawals in the Bible Major shows a slightly higher percentage of withdrawals, indicat- ing that the choice among institutions offering a major in Bible must be sufficiently large enough to offer students a broad choice. The point is relatively clear (because of the large numbers of applicants), it is operationally important to ‘meet the demands of those academic disciplines that are in vogue. 0n the other hand, it should be equally clear that a college such as Geneva should stress the programs such as Liberal Arts. The Admissions people should be particularly conscious of the fact that those students who do apply for admission in one of the less popular programs are more likehr to enroll than those students who are interested in the more popular programs offered by a large number of schools. Decision: The hypothesis is accepted. 82 Hypothesis 6: There is no significant difference in the source of contact between confirmed and withdrawals. The data from Table 4-9 show that there is a signifi- cant statistical difference between confirmed and with- drawals as affected by the source of contact with the college. A closer examination shows that there are marked contrasts between the various sources of contact as they affect the confirmed students and those students who with- draw and do not enroll. The data in the table can be divided into two major areas, one might be referred to in.marketing terms as the areas of "personal selling" while the other might be re- ferred to as "general promotion." In the area of personal selling or contact by persons employed by or interested in the college, there is a wide variation in the degree of success as one attempts to get students to confirm their intentions to enroll at Geneva. The personal referrals from alumni, faculty and friends of the college and those students who come from local families show the greatest positive effect favoring confirms over withdrawals. In marked contrast to the personal referrals and local resi- dents, the category entitled Special Personal Contacts made up predominantly of non-Admissions recruiters had a very low success rate in generating confirmed students. In de- fense of these people, these withdrawals may have been highly sought after by a number of colleges, and might well 83 Table 4-9 The Effect of Source of Contact with College on Status--Confirmed vs. Withdrawals Source of Contact Confgrmed Withdfiawals (N=329) (N=218) izzizznsiiiizs: 6w Recruitment Letters 54.1 45.9 Non—identified 48.3 51.7 Telephone Calls (recruitment) 73.7 26.3 Visit to Campus 71.0 29.0 Local Resident 81.8 18.2 College Directories 46.2 53.8 Other Magazine Promotion 50.0 50.0 "Campus Life" 58.8 41.2 "Christian College Profile" 56.3 43.8 High School & Other Conferences 48.4 51.6 Special Personal Contacts 41.9 58.1 Personal Referrals 82.9 17.1 Chi-square = 31.7609 Significance .0015 84 be top athletes or top-ranking scholastic scholars who have a wide range of choice from among a number of schools. It is equally clear that recruitment telephone calls and campus visits have been highly successful as a tool in generating confirmed students compared with withdrawals. On the other hand, the general category of conferences such as those con- ducted at high schools or churches have apparently been ineffectual at best. An examination of what we might call the promotional sources of contact would also indicate some wide variations in the comparison of confirmed with withdrawals. The Campus Life magazine and the Christian Collgge Profile are running well ahead of the other magazine publications as their confirmed to withdrawals categories would indicate. Listing in college directories have related more specifi- cally to withdrawals than to confirmed while the great emphasis on recruitment letters seems to have a marginal success. It is of interest to note that those high school students who found out about Geneva College through the American College Testing Service publications have somewhat better than a 2-to-1 ratio favoring confirmed over with- drawals. A possible explanation would be to suggest that these students feel that this is a less biased source of information. It would seem that the examination of this table and data reflected in it should help Admissions people sort out the sources of first contact and concentrate on those that 85 tend to be most effective in generating confirmed students. This would reduce the expense of maintaining the less pro- ductive methods and would provide additional funds that could be directed to those tools that seem more beneficial in the generation of confirmed students. Decision: The hypothesis is rejected. Hypothesis 7: There is no significant difference in type of student between confirmed and withdrawals. During the last 10 to 15 years, Geneva College has changed from an institution with a student body make up of two-thirds commuting students and one-third resident stu- dents, to an institution with a resident student body that now makes up two-thirds of the total enrollment. The ef- fects of this change have been of marked importance in the life of the institution. Table 4—10 reflects these data as it relates to confirmed versus withdrawals, and they show that there is a statistically significant difference between the two categories. The three categories, commuter, part- time day, and evening students, are all groups that commute to the college. All of these groups show a statistically significant difference between confirmed and withdrawals, heavily favoring the confirmed category. In contrast, the data for the resident student body ShOW’ no significant dif- ference between the confirmed and withdrawal categories, indicating that this more traditional group of students has 86 Table 4-10 The Effect of Type of Student on Status--Confirmed vs. Withdrawals % % Type of Student Confirmed Withdrawals (N=329) (N=219) Commuter 75.5 24.5 Resident 56.3 43.7 Part-time Day 100.0 0.0 Evening 85.7 14.3 Chi-square = 14.6223 Significance = .0022 87 wider range of choice between Geneva and other institu- tions. The implications of these data are rather clear; while the resident students make up the biggest segment of the student body, the students that indicate that they are going to be commuting either as full-time, part-time, or evening students are much more likely to confirm their in— tentions of enrolling in the institution. The important thing for the people in Admissions is to realize that although the primary emphasis is being given to the resi- dent student body, that particular attention should be paid in the recruitment process to those students from the local area that indicate an interest in the College, for there is a.much greater probability that they will enroll at Geneva. Decision: The hypothesis is rejected. Hypothesis 8: There is no significant difference in denominations between confirmed and withdrawals. Frequently, colleges tend to affiliate themselves with a particular denomination or group of denominations, or at least a particular theological position and often emphasize that position in their recruiting efforts in an attempt to draw from a specific market segment. The data in Table 44J. show the effect of such affiliation as it relates to the status of confirmed versus withdrawals for the four major denominational groupings; Nondenominational, Evangelical, 88 Table 4-11 The Effect of Church Affiliation on Status--Confirmed vs. Withdrawals Church Affiliation Confgrmed Withdiawals (N=330) (N=218) Non—denominational 51.9 48.1 Evangelical 63.0 37.0 Protestant 62.2 37.8 Catholic 54.4 45.6 Chi—square = 3.7022 Significance = .2955 89 Protestant, and Catholic. The data show that there is no significant difference between confirmed and withdrawals as affected by church affiliation. 0f operational signifi- cance, there is, as might be expected, a slight difference favoring confirmed over withdrawals for the Evangelical and Protestant groupings which shows about a 60-40 split in favor of the confirmed. On the other hand, the nondenomina- tional group and the Catholic group show no significant dif- ference between the two status categories. All that can be concluded is that denominational dif- ferences are not all that significant, as they seem to have a minimal effect on the status of confirmed versus with- drawals. Decision: The hypothesis is accepted. Hypothesis 9: There is no significant difference in the distance from home between confirmed and withdrawals. The geographic dimension of any market is always of major interest to any institution supplying a service, and colleges are no exception. Table 4-12 shows the results of analyzing location data based on the status of confirmed versus withdrawals. This analysis indicates that there is a statistically significant difference between the two groupings. A detailed analysis of the data on commuting versus residential students in connection with Hypothesis 7 would suggest the fact that students living close enough to 90 Hmzo. u ooqmofimwumwm mmmm.:H u onoodmuwno o.me o.mm our noos< ”WWW ”WM es 0.s a-mwfiwfimofim as as sing-” ”mammomhm m.wm m.Hm Ofino unopmom m.Hm m.w: pond ofinmaoooafinm m.m: N.Hm owoo>ahmooom onovmos H.hm a.mw mowpgsoo heapsm m huoswoaad .oouohamq a.mm m.o~ hpqzoo ho>oom Azamuzv Aommuzv maoawnoapfiz ooSMwmooo GSOpoEom no coapoooq u u madauuonpw3 .m> ooahwwuoollmupopm no osopoaom mo cOapdooq mo poowwm one mums canoe 91 commute would ShOW'a significantly greater number in the confirmed category. Beaver County, where Geneva College is located, showed that 70 percent of the accepted students actually confirmed their intentions of enrolling at the College. Lawrence, Allegheny, and Butler counties and the eastern Ohio group also showed a significantly higher number in the confirmed category. In contrast, the rest of western Pennsylvania and the Philadelphia area, Which includes part of the states of Delaware and New Jersey, showed no difference between con- firmed and withdrawals as is reflected by the 50-50 split shown in the table. This might well reflect and indicate the fact that Geneva has some rather strong competition 'within this general geographical area. The difference between confirmed and withdrawal status for the next two most distant locational areas--those indicated as the zip code areas 0, l, 2, and 4, and zipcode areas 3, 5, 6, and 7 representing the North Atlantic and New England and Southeast and Central areas, respectively--show a slight increase favoring the confirmed status category. Again, one can speculate as to the reasons and might con- clude that students who are interested enough to apply to a college that is some 500 to 1,000 miles away from home, might be more likely to attend that institution because they see Geneva as significantly different enough to make it worth the additional travel time. Finally, the confirmed versus withdrawals figures for the areas farther west, 92 represented by the major geographic area in zipcodes 8 and 9, would simply indicate that Geneva College is too far away to attract very many students. The data shows that three-fourths of the students who do apply eventually with- draw their applications and elect to go someplace else. The message for the recruitment efforts again seem.re- latively clear; those students close by are the most likely to confirm their intentions of enrolling at the college, and those students at great distances are the most likely not to enroll in the institution. The areas in-between are, of course, the most difficult to analyze; but it would seem that the emphasis should be placed by the recruitment per- sonnel on attracting students from the tri-county, that is, the Lawrence, Allegheny, and Butler counties, and the bor- dering counties in eastern Ohio. Beyond that, all that could be suggested is that the recruitment effort should follow a pattern of concentration in areas where the college has successfully attracted students in the past. This would relate to some of the previous data dealt with under Hypothesis 6 which suggests that the source of first con- tact can be of great importance. Decision: The hypothesis is rejected. Hypothesis 10: There is no significant difference in type of scholarship and between confirmed and withdrawals. 93 The data in Table 4-13 indicate that there is a signi— ficant statistical difference between confirmed and with- drawals as related to the type of scholarship aid offered to various students. Of major interest to the Admissions Office and Financial Aid Office is the fact that certain scholarships seem to increase the probability of the con- firmed status while other scholarships seem to have little or no effect. Two comparisons seem especially relevant. The first is a comparison between the scholarship group entitled "clergy only" and the scholarship group identified as "RPNA". Both of these groups reflect relatively small scholarships awarded on the basis of religious association. The clergy scholarship is awarded to children of clergy and would seem to have no positive effect on status. In con- trast, the RPNA scholarship is awarded to children of the ‘members of the Reformed Presbyterian Church of North America, which is the church that controls the College. Those stu- dents offered this scholarship show a very high probability of attending Geneva, an event which would probably occur regardless of the scholarship aid being offered. It could, therefore, be argued that neither of these scholarships greatly increase the number of students that eventually en- roll at Geneva College. The second comparison of educationally significant interest is between first and second honor scholarships. The first honor scholarship group is made up of those stu- dents who graduated in the t0p 10 percent of their high 94 Table 4-13 The Effect of Scholarship Aid on Status--Confirmed vs. Withdrawals % % Type of Scholarship Aid Confirmed Withdrawals (N=107) (N=70) Clergy Only 50.0 50.0 lst Honor-Top 10% Only 69.8 30.2 2nd Honor-Top 20% Only 50.8 49.2 NMSQT National Merit and All Combinations 57'1 "2'9 RPNA-Church and All Combinations 91'7 8'3 Science and A11 Combinations 50.0 50.0 Chi-square = 11.1656 Significance = .048 95 school class. The second honor scholarship group repre- sents those students who graduated in the tOp 20 percent of their high school class. The two groups are mutually ex- clusive. There appears to be some gain in the confirmed group represented by the first honor category. The second honor group shows, at best, a neutral relationship to the scholarship being offered, and it could be argued that the effect might even be negative. Likewise, a comparison of first honor scholarships and science scholarships, i.e., special scholarships offered to students interested in the hard sciences who meet certain qualifications, shows the same relationship as the one between first honor scholar- ships and second honor scholarships. It would therefore appear that the scholarship program ‘might very well be re-evaluated to examine the benefits of eliminating some of the less productive scholarships and adding those funds to scholarships which produce the great- est return in confirmed students. Decision: The hypothesis is rejected. Hypothesis 11: There is no significant difference in class rank between confirmed and withdrawals. Hypothesis 12: There is no significant difference in class size between confirmed and withdrawals. 96 Tables 4-14 and 4-15 indicate the data dealing with class size and rank in high school class. The data indi- cate that there was no statistical significant difference in either of these two majors as they affected the status of confirmed versus withdrawals. In terms of educationally significant differences, the average student in the con- firmed category came from a slightly smaller size high school class and stood somewhat higher in that class. The compara- tively high standard deviation values in both tables would indicate a wide range of students in terms of both high school class size and standings within that class. Decision: Hypothesis 11 is accepted. Decision: Hypothesis 12 is accepted. Hypothesis 13: There is no significant difference in the SAT verbal test scores between confirms and withdrawals. Hypothesis 14: There is no significant difference in the SAT math test scores between confirms and withdrawals. The data in Table 4-16 and 4-17 indicate that there is no statistically significant difference between confirmed and withdrawals as affected by either SAT verbal or math scores. The confirmed group of students had slightly higher average verbal test scores and slightly lower average math test scores than did their corresponding withdrawal 97 mfimxaou< HoOfipmeopm opofinmoammmoH I xxx xxx Hwa. mm.maa om.moa :ma mamamhdnpr Manx mmmdo om.aoa Po.mm mmm ooahfimooo Hoonom swam osae>ue oooaoaoaomam .n.m one: 2 neuron oapaaso> madawhonpwz .m> ooaufimooonlmsvopm no room anode Hoooom swam uo rectum one as -s oases 98 mxmxaoq< Hoowpmxpopm opoxxmoxamooH I xxx xxx mom. os.msm mo.smm xmx adornsoopxa oaxm annxo ms.mos ox.xmm mom oossxxooo Hoooom swam osan>ue ooeeoxoxomxm .o.m our: 2 museum expoxso> madaoxospxs .m> ooahxmuoollmopopm no oaam anexo Hoooom swam so season one m~.!: panda 99 mxmxamua Hooxpmxpopm opaxxmoxmmooH u xxx xxx max. ox.mo s:.mss max nxosasoooxz osoem Hanso> oo.oo mm.mma saw oossxeooo saw osxo>ua oooaoxoxomxm .o.m coo: z narrow oxoaxso> madamxonpxz .m> doaxxmooollmopoum no nonoom Hopso> saw so septum one 0a-: oases 100 mxmxamu< Hmoxpmxpovm opmxxmoxmmmuH I xxx xxx xmm. oo.om se.xom max extensoooxz osoom son: ss.mm xm.xmx :xm eoenxoooo mam osxo>Ie oooaoxuxomxm .o.m are: 2 noooom oxnoxso> mxosmxonpxz .m> ooahxmooollmopopm no nosoom our: e