An assessment of the relationship between social integration variables and community college student retention

MISSING IMAGE

Material Information

Title:
An assessment of the relationship between social integration variables and community college student retention
Physical Description:
xiii, 186 leaves : ; 28 cm.
Language:
English
Creator:
Rom, Michael G
Publication Date:

Subjects

Subjects / Keywords:
College dropouts   ( lcsh )
Dropout behavior, Prediction of   ( lcsh )
Community college students   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1985.
Bibliography:
Includes bibliographical references (leaves 142-150).
Statement of Responsibility:
by Michael G. Rom.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 14772017
ocm14772017
System ID:
AA00011210:00001

Full Text








AN ASSESSMENT OF THE RELATIONSHIP BETWEEN SOCIAL INTEGRATION
VARIABLES AND COMMUNITY COLLEGE STUDENT RETENTION















BY

MICHAEL G. ROM


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY




UNIVERSITY OF FLORIDA


1985

















DEDICATION


To my parents Jean and George Rom,
Who taught me to accept challenges and conquer fear;

To my son David who had to play ball by himself,
Because I was not there;

To my daughter Michele who because of my absence,
Fell asleep while shedding a tear;

To my ever loving wife Lucile,
Who was both mother and father when I was not near;

To my God who when things appeared the darkest,
Gave me strength to see things clear;

This dissertation is dedicated in your honor.
















ACKNOWLEDGEMENTS

I would like to acknowledge several individuals for their

continual support. Without these people I might not have

completed this task.

My committee consisted of Dr. Al Smith, Dr. Steve Olenjik,

Dr. Gordon Lawrence, and Dr. Paul Fitzgerald. I would like to

thank Steve for his patience which helped me through the final

stage of my data analysis. I would like to especially thank Al

who at times had to show me the way, at times hold me back, and

several times had to come find me because I was lost. To each of

these men I am deeply grateful.

I would like to thank all the professors, secretaries,

clerks, and students who were a part of my study and experiences

at the University of Florida. I would also like to express my

deepest appreciation to all of my friends who have given me

support and love on a daily basis. It's the contribution of the

many parts, however small, that make up the total affect of any

experience. For the part each of you played in mine I thank you.

The individuals who contributed to this study by editing,

typing, reading, or adding words of encouragement are too

numerous to list. I would like to acknowledge some special

individuals who made my task easier: David Rom, Jean Rom,

Charles and Catherine Hudson, Bill Reynolds, Larry Eason, Jane

iii










Schmit, Dora Mae O'brien, Rindy Penegor, Debbi Amburgey, Juan

Burbano, Bob Judson, Debbie Depoy, Don Lester, Hugh Turner, Dave

Helfirch, and Susan Anderson. In addition much appreciation goes

to the staff, faculty, and administration at Pasco Hernando

Community College for their support of this project.

To my family, Lucile my wife, David my son, and Michele my

daughter, I pledge my deepest love and appreciation. Seven

years ago if I had known what I was going to ask of you, I would

not have. My ignorance of the sacrifices you would have to make

the next seven years allowed me to forge on step by step. I not

only thank you for standing by me when I needed you but I truly

thank you for doing this tremendous task with me. We finally

made it and received our Ph.D.'s. Thank you Drs. Lucile, David

and Michele Rom, Ph.D.'s for all your love and congratulations

for a well deserved honor.
















TABLE OF CONTENTS



PAGE


ACKNOWLEDGEMENTS.............................................iii
LIST OF TABLES....................... ........................vii
LIST OF FIGURES................................................ x
ABSTRACT................... ...................................xi

CHAPTER

ONE INTRODUCTION ..................... .....................1

Tinto's Model.........................................5
Statement of the Problem.............................9
Hypotheses............................................9
Assumptions ................. ......................... 12
Delimitations ..................... ...................12
Limitations..........................................12
Significance of the Study...........................14
Definition of Terms.................................. 16
Organization of the Dissertation........................19

TWO REVIEW OF THE LITERATURE AND RELATED RESEARCH..........20

Review of the Literature..............................22
Validity of Tinto's Model in Residential
Settings...........................................24
Validity of Tinto's Model in Non-residential
Settings.............................. ............25
Reconceptualization of Tinto's Model ..............28
Variables...........................................31
Family Background.................................32
Individual Attributes.............................33
Pre-College Schooling............................ 34
Institutional Commitment..........................34
Goal Commitment..... .........................35
Grade Performance................................35
Intellectual Development.........................36
Peer-Group Interaction.................. ..........36
Faculty-Student Interaction......................37
Summary of the Chapter................................37











THREE METHODOLOGY .... .........................................40

Instrumentation.................................... ..40
Description of Student Survey.....................40
Development of Student Survey......................42
Variables................ ..........................42
Pilot Study.........................................50
Present Study.........................................51
Subjects..... ....................................51
Data Collection....................................52
Data Analysis............. ........................52
Summary of the Chapter..............................56

FOUR FINDINGS........ .....................................57

Descriptive Analysis...... ........................... 60
Demographic Variables...............................60
Correlations........................................66
Regression Analysis..................................78
Hypotheses 1-4................................... 78
Hypotheses 5-13..................................92
Hypotheses 14-19................................107
Anova Between Campuses............................112
Summary of the Chapter.............................113

FIVE SUMMARY, CONCLUSIONS, IMPLICATIONS,
AND RECOMMENDATIONS..................................117

Summary ................... ..........................117
Conclusion..................... .....* ....* ...........122
Implications...... .................. ............... 134
Recommendations.................. ...................138

REFERENCES ...................................................142

APPENDIX

A PROGRAM DECLARATION........................... ....152
B WRITTEN ENGLISH EXPRESSION PLACEMENT
TEST................................................. 154
C READING PLACEMENT TEST................................ 164
D STUDENT SURVEY ....... ...............................175
E MONITOR INSTRUCTIONS.................................. 185

BIOGRAHICAL SKETCH............ ..............................186

















LIST OF TABLES


TABLE PAGE

1 Frequency and Percentage Distribution for
Demographic Variables....................................61

2 Pearson Product Moment Correlation for
Endogenous and Exogenous Variables--
Total Campus...............................................67

3 Pearson Product Moment Correlation for
Endogenous and Exogenous Variables--
East Campus........................................... ..70

4 Pearson Product Moment Correlation for
Endogenous and Exogenous Variables--
North Campus.......................................... ..73

5 Pearson Product Moment Correlation for
Endogenous and Exogenous Variables--
West Campus.............................................76
2 2
6 R R" Increase and Computed F Ratios for the
Full Model and Partial Regression Coefficients
Relating Students' Perceptions of Dtopout Decisions
With Social System Integration by Campus................80

2 2
7 R R Increase and Computed F Ratios for the
Full Model and Partial Regression Coefficients
Relating Students' Actual Dropout Decisions
With Social System Integration by Campus................80

2 2
8 R R Increase and Computed F Ratios for the
Full Model and Partial Regression Coefficients
Relating Students' Perceptions of Dropout Decisions
With Peer-Group Interaction by Campus....................83

2 2
9 R R Increase and Computed F Ratios for the
Full Model and Partial Regression Coefficients
Relating Students' Actual Dropout Decisions
With Peer-Group Interaction by Campus....................83

2 2
10 R R Increase and Computed F Ratios for the
Full Model and Partial Regression Coefficients
Relating Students' Perceptions of Dropout Decisions
With Faculty-Student Interaction by Campus..............87











TABLES-continued


TABLES PAGE

2 2
11 R R Increase and Computed F Ratios for the
Full Model and Partial Regression Coefficients
Relating Students' Actual Dropout Decisions
With Faculty-Student Interaction by Campus..............87

2 2
12 R2, R Increase and Computed F Ratios for the
Full Model and Partial Regression Coefficients
Relating Students' Perceptions of Dropout Decisions
With the Interaction Effect of Peer-Group and
Faculty-Student Interaction by Campus...................90

2 2
13 R R Increase and Computed F Ratios for the
Full Model and Partial Regression Coefficients
Relating Students' Actual Dropout Decisions
With the Interaction Effect of Peer-Group and
Faculty-Student Interaction by Campus....................90

14 Means, S.D., and Computed t-Statistic
Comparing White and Non-White Students
on the Social System Variable............................ .. 93

15 Means, S.D., and Computed t-Statistic
Comparing White and Non-White Students
on the Peer-Group Interaction Variable..................95

16 Means, S. D., and Computed t-Statistic
Comparing White and Non-White Students
on the Faculty-Student Interaction Variable.............97

17 Means, S.D., and Computed t-Statistic
Comparing Male and Female Students
on the Social System Variable............................99

18 Means, S.D., and Computed t-Statistic
Comparing Male and Female Students
on the Peer-Group Interaction Variable.................100

19 Means, S.D., and Computed t-Statistic
Comparing Male and Female Students
on the Faculty-Student Interaction Variable ............102

20 Means, S.D., and Computed t-Statistic
Comparing Students Who Have Had a Human
Relations Type Course With Students Who
Have Not on the Social System Variable..................104


viii










TABLES-continued


TABLES PAGE

21 Means, S.D., and Computed t-Statistic
Comparing Students Who Have Had a Human
Relations Type Course With Students Who
Have Not on the Peer-Group Interaction Variable........ 106

22 Means, S.D., and Computed t-Statistic
Comparing Students Who Have Had a Human
Relations Type Course With Students Who Have
Not on the Faculty-Student Interaction Variable....... 108

23 Correlation Coefficients for Accumulated
Semester Hours and Social System, Peer-Group
Interaction, and Faculty-Student Interaction
Variables by Campus.................................... 109

24 Correlation Coefficients for Age and Social
System, Peer-Group Interaction, and Faculty-
Student Interaction Variables by Campus................. 11

25 ANOVA Between East, North, and West
Campuses on Variables of Interest...................... 114

26 BONFERRONI Statistics: Faculty-student
Interaction on Student Survey Questionnaire............115

27 Results of the Hypotheses...............................127















LIST OF FIGURES


FIGURE PAGE

1 A Conceptual Schema For Dropout
From College (Tinto, 1975).......................... 3

2 Suggested Reconceptualization of
Tinto's Model (Pascarella, Duby,
and Iverson, 1983)..................................31
















Abstract of Dissertation Presented to the Graduate
School of the University of Florida in Partial
Fulfillment of the Requirements for the Degree of
Doctor of Philosophy




AN ASSESSMENT OF THE RELATIONSHIP
BETWEEN SOCIAL INTEGRATION VARIABLES AND COMMUNITY
COLLEGE STUDENT RETENTION


By


MICHAEL G. ROM


August 1985



Chairman: Albert B. Smith III
Major Department: Instruction and Curriculum



Using Tinto's conceptual framework model, the purpose of

this study was to determine what relationships existed between

social system integration variables (peer-group and

faculty-student interaction) and student dropout decisions in a

two-year community college. Two hundred students, at Pasco

Hernando Community College, were surveyed during the Fall term of

1984, using a Student Survey Questionnaire. One hundred

fifty-two usable responses were received for a 76 percent return

rate. The instrument measured the student's background

characteristics, commitment (goal and institutional), academic

integration, social integration (peer-group and faculty-student

xi










interaction), and perceived dropout decision. In addition, other

demographic variables were requested via this questionnaire.

Four different population groups were considered; the total

student population, students attending the East, the North, and

the West campuses. Each of the 19 hypotheses was tested using

the four different population groups. Hypotheses 2 and 3, which

related peer-group and faculty-student interaction to the actual

dropout decision, showed significance at the .05 level for the

East and North campuses.

For East Campus students a significant proportion of the

variation in dropout decisions (11.2%) was explained by the

peer-group interaction variable (negative influence) and the

faculty-student interaction variable (8.6%), which was a positive

influence. For North Campus students a significant proportion of

the variation in dropout decisions (10.9%) was explained by the

peer-group interaction variable (positive influence) and the

faculty-student interaction variable (16.7%), which was a

negative influence.

Of the remaining 15 hypotheses relating age, sex, race,

accumulated semester hours, and a human relations type course

with the social integration, peer-group interaction, and

faculty-student interaction variables, six were found to have

statistically significant relationships at the .05 level.

The study showed that similarities between campus

populations were greater than the differences. The two main

xii










exceptions were peer-group and faculty-student interaction. The

variable relationships in Tinto's model did not appear to apply

evenly to this commuter school. When commuter populations have

characteristics in common with residential schools, then Tinto's

model of the dropout process may be more applicable than recent

research indicated.


xiii
















CHAPTER ONE

INTRODUCTION

Substantial time and interest have been dedicated to college

student attrition as is represented in the work of by Spady

(1970); Cope and Hannah (1975); Tinto (1975); Pantages and

Creedon (1978); and Lenning, Sauer, and Beal (1980). Despite all

the research, community college attrition rates have remained

high and virtually unchanged.

A national survey conducted in the spring of 1979 by the

American College Testing (ACT) Program and the National Center

for Higher Education Management Systems verified this consistency

in attrition rates (Beal & Noel, 1979). Retention after one year

in two-year public institutions was 55 percent in 1975-76, 55

percent in 1976-77, and 53 percent in 1977-78 as reported by 74,

82, and 92 two-year institutions respectively. More recent

statistics also confirmed a 50 percent dropout rate in

postsecondary schools (Grant & Eiden, 1982).

Prior to the 1970's, much of the research on attrition was

theoretical, identifying a variety of associations among various

student and institutional characteristics and attrition, but

lacked a theoretical base by which attrition could be studied.

Tinto (1975) attempted to bring some coherence to this research

as well as provide a conceptual framework to guide future

research. Tinto expanded Spady's (1970) work on student

I











attrition by developing a predictive, explanatory model of the

dropout process which has at its core the concepts of student

academic and social integration into the institution.

Tinto (1975), in his article "Dropout from Higher Education:

A Theoretical Synthesis of Recent Research," drew upon Durkeim's

(1961) theory on suicide which essentially theorizes that suicide

is more likely to occur when individuals are insufficiently

integrated into the fabric of society. Spady (1970) first

applied Durkeim's theory to college student dropouts by

suggesting that a college is a social system with its own values

and belief system.

Tinto (1975) further suggested that social conditions

affecting dropouts from college would be similar to those social

conditions resulting in suicide in society as a whole.

Specifically, he stated:

Insufficient interaction with others in the college and
insufficient congruency with the prevailing value
patterns of the college collectively will lead
to low commitment to that social system and will
increase the probability that individuals will decide
to leave college and pursue alternative activities.
(p.92)

The model Tinto developed to depict his theory on college

dropout decisions emphasized two main areas of integration: the

academic and social systems (see Figure 1 on p.3). These areas

of integration have been verified as causes of college dropouts

and as explanations for college student persistence (Bayer, 1968;

Denzin, 1966; Medsker & Trent, 1968; Rootman, 1972; Scott, 1976;

Spady, 1971).














III


0
II 0
1t 0 0
a 4)
a QQ
II I I
AA
I I
I I
manaa Hnil


SII II H a aI I a


0
ULJ

*O 0
-4 4!



ua
SUC
H
4).


a N
A
I
I
n
Hf


It It u 0 NI
A
I
I
I
I
I
I
I
I
I
I
I
I
I
I


U :3u H
A




U 3
c. u II
H f H eI HM A



S 0 o2 II


0 04 I4

0 4 S 4 IA




A II A NII


0
E
0
U
N IIN II
A
I




X3
O

-4 .w

(0

anaI a


I
I
I
I
I
I
I



I
I
I
I


II


H i n ui
aI aII



40

rtA

o U
M4 <
4-4

I J
3 e




0
P a2

U
alinAna


II II I

0
-4
ca

000
O
II0 4)


II I I I I II
II 0.

m -4
Sao
3 1.
H 0 -,

I NL





A
I





-4 -4


0 O
I









u X
JICs








lh





It N N N 0
SriI II I

U
a aa n


0



0 4
t I


S I-
ana


,J

41
i-4

r
0
U







0






CU
03












Tinto states that social integration occurs primarily

through informal peer-group associations, semi-formal

extra-curricular activities, and interaction with faculty and

administration within the college. Tinto's theoretical model

implies that successful social integration, via the above means,

results in increased retention of students in an educational

institution. Tinto stated:

Successful encounters in these areas result in varying
degrees of social communication, friendship support,
faculty support, and collection affiliation, each of
which can be viewed as important social rewards that
become part of the person's generalized evaluation of
the cost and benefits of the college attendance and
that modify his educational and institutional
commitments. Other things being equal, social
intergation should increase the likelihood that the
person will remain in college. (p. 107)

Pascarella and Chapman (1983), using 2,326 freshmen from 11

postsecondary institutions, and Pascarella and Terenzini (1983),

using a longitudinal study with three data collections and a

sample of 763 freshmen, verified the direct influence of social

integration on college dropouts as theorized by Tinto. These

studies were conducted in four-year residential schools. However

when Tinto's model was applied to both four and two-year commuter

schools, the results were non-supportive (Pascarella & Chapman,

1983; Pascarella, Duby, & Iverson, 1983). Pascarella and Chapman

(1983) found no direct nor indirect effects of social integration

on persistence in four or two-year commuter institutions. In

addition, Pascarella, Duby, and Iverson (1983), using 269 in-

coming freshmen, reported a negative direct effect of social







5




integration on persistence in a commuter school. This negative

effect was also reported in a residential school by Terenzini,

Pascarella, Theophilides, and Lorang (1983).

Even though social integration variables have been shown to

affect dropout decisions in particular institutions, the

application of Tinto's model across all institutional types is

still problematic. Further research is needed to verify the

influence of social integration variables, specifically

peer-group and faculty-student interaction, on college student

persistence as it relates to different types of postsecondary

institutional settings.

Tinto's Model

In the theoretical model of dropout decisions in college

diagrammed in Figure 1, Tinto suggests

the process of dropout from college can be viewed as a
longitudinal process of interactions between the
individual and the academic and social systems of the
college during which a person's experiences in those
systems (as measured by his normative and structural
integration) continually modify his goal and
institutional commitments in ways which lead to
persistence and/or to varying forms of dropout. (p.
94)

Individuals enter institutions of higher education with

background characteristics, pre-college experiences, a variety of

individual attributes, and family backgrounds, each of which has

a possible direct and/or indirect impact upon their performance

in college. In addition, these background characteristics and

individual attributes also influence the development of the











educational expectations (goal commitments) and commitments the

individual brings into the college environment (institutional

commitments). These goals and institutional commitments serve

both as predictors and reflections of the person's successes and

failures in the collegiate setting.

Tinto's model further suggests that if background

characteristics and commitments are given, then the individual's

integration into the academic and social systems of the college

is the factor that most directly affects the student's

continuance in that college. Tinto (1975) states, "Given prior

levels of goal and institutional commitment, it is the person's

normative and structural integration into the academic and social

systems that lead to new levels of commitment" (p. 96).

According to Tinto's model, other things being equal, the

higher the degree of integration of the individual into the

college system, the greater will be the student's commitment to

the goal of college completion. Through the model Tinto implies

that if preceding variables can be held constant, such as prior

college characteristics and/or experiences and commitment, then

academic and social integration will contribute to the

persistence of a student in college. According to Tinto, if

academic system variables could be controlled for, then social

system variables, specifically peer-group interaction and

faculty-interaction, would contribute to and explain the

persistence of a student in college.












There is adequate research to justify Tinto's inclusion of

social systems into his model of retention. Peer-interaction has

been shown to be an important variable in a student's decision to

stay in college (Denzin, 1966; Pancos & Astin, 1968; Slocum,

1956; Spady, 1970). Spady (1971) used a sample of 683 students

who entered the University of Chicago as freshmen in September

1965 to study interpersonal relations. He concluded that

interpersonal relationships accounted for over 12 percent of the

explained variance in social integration for the men and nearly

20 percent for the women.

The quantity and/or quality of faculty interaction has been

demonstrated as an important variable in the retention of

students in colleges (Centra & Rock, 1971; Cesa, 1980; Noel,

1976; Pascarella & Terenzini, 1976; Spady, 1971). Pascarella and

Terenzini (1977) used a sample of 536 students at Syracuse

University, to test Tinto's theoretical model of attrition.

Under discriminant analysis, Pascarella and Terenzini concluded

that informal student-faculty contact is a significant predictor

of college persistence with significant F-ratios being found on

three of the six faculty interaction categories.

Other things being equal, the greater the college student's

level of social and academic integration, the greater his or her

subsequent commitment to the institution and commitment to the

goal of college graduation, respectively. These subsequent












commitments, in turn, are seen, along with levels of integration,

as having a positive influence on persistence.

As suggested by Tinto's model, the student's social system

is an important part of the process that leads toward an

individual's decision to persist or dropout of college.

According to research, peer-group interaction and faculty-student

interaction have a unique contribution to student social systems

and student retention, but this may not apply across all

institutional types. Additional research needs to concentrate on

the influence of social system variables, such as peer-group and

faculty-student interaction, on college student dropout in both

residential and commuter institutions.

In order to determine the influence of social system

variables on the dropout decision, all other preceding and

related variables such as family background, individual

attributes, pre-college schooling, commitments, and academic

system need to be held constant. If this is not done, then one

cannot assume that social system variables alone contribute to

the student's decision to either leave school or remain.

This dissertation controlled for those variables in Tinto's

model, background characteristics, commitments and academic

system, while investigating the relationship between social

system variables and college student retention in a two-year

commuter community college.












Statement of the Problem

A problem facing community colleges today is high student

attrition or drop out rates. There is a need to investigate what

variables relate to community college student attrition in order

to better understand and prevent the dropout process.

Using Tinto's (1975) conceptual framework, the purpose of

this study was to determine what relationships exist between

social system integration variables (peer-group interaction and

faculty-student interaction) and students dropout decisions in a

two-year community college. The researcher hoped that the

results of this study would then be used by community colleges to

develop new programs designed to reduce student attrition and

subsequently enhance support for these colleges.

Hypotheses

The following null hypotheses were tested (alpha level .05):

Hypothesis 1. A significant proportion of the variation in

student dropout rates is not explained by selected social

system variables after controlling for student background

characteristics, student commitments, and the student

academic system (student grade performance and student

intellectual development).

Hypothesis 2. A significant proportion of the variation in

student dropout rates is not explained by peer-group

interaction after controlling for student background

characteristics, student commitments, the academic system,

and faculty-student interaction.












Hypothesis 3. A significant proportion of the variation in

student dropout rates is not explained by faculty-student

interaction after controlling for student background

characteristics, student commitments, the academic system,

and peer-group interaction.

Hypothesis 4. There is no interaction effect between peer-group

interaction and faculty-student interaction.

Hypothesis 5. There is no significant difference between white

and non-white student population measurements on the social

system variable.

Hypothesis 6. There is no significant difference between white

and non-white student population measurements on the

peer-group interaction variable.

Hypothesis 7. There is no significant difference between white

and non-white student population measurements on the

faculty-student interaction variable.

Hypothesis 8. There is no significant difference between male

and female student measurements on the social system

variable.

Hypothesis 9. There is no significant difference between male

and female student measurements on the peer-group

interaction variable.

Hypothesis 10. There is no significant difference between male

and female student measurements on the faculty-student

interaction variable.












Hypothesis 11. There is no significant difference between

students who have taken a human relations type course and

students who have not on student measurements of the social

system variables.

Hypothesis 12. There is no significant difference between

students who have taken a human relations type course and

students who have not on student measurements of the

peer-group interaction variable.

Hypothesis 13. There is no significant difference between

students who have taken a human relations type course and

students who have not on student measurements of the

faculty-student interaction variable.

Hypothesis 14. There is no correlation between student semester

hours completed and the social system variable.

Hypothesis 15. There is no correlation between student semester

hours completed and the peer-group interaction variable.

Hypothesis 16. There is no correlation between student semester

hours completed and the faculty-student interaction

variable.

Hypothesis 17. There is no correlation between student age and

the social system variable.

Hypothesis 18. There is no correlation between student age and

the peer-group interaction variable.

Hypothesis 19. There is no correlation between student age and

the faculty-student interaction variable.












Assumptions

The assumptions for this study were as follows:

1. The testing of some of the hypotheses relied on

self-reported data. Systematic error caused by method bias,

therefore, may have affected any relationships that have been

confirmed or questioned. However, there was evidence that

suggested that one's perception of social integration most

directly relates to college persistence (Pervin, Reik, &

Dalrymple, 1966; Rootman, 1972; Spady, 1971). This research was

relevant since the responses on the instrument used in this study

were based on the student's perception.

2. It was assumed the dropout rate for Associate of Arts

and Associate of Science degree seeking students would be

approximately 30 percent.

3. Decisions to remain in school or drop out are tentative

decisions and therefore conclusions derived from such data should

be considered in the same reference.

Delimitations

This study was confined to the student population of Pasco

Hernando Community College, a tri-campus college located in Dade

City, Brooksville, and New Port Richey, Florida.

Limitations

1. The ex post facto design of the study and the fact that

questionnaires were administered only once precluded advantages

inherent in experimental designs. The researcher was not able to












control all extraneous variables and to manipulate independent

variables.

2. The results of the study were interpreted within the

limitations imposed by the validity and reliability of the survey

instrument used in the investigation and the decisions made

concerning which items to include in the constructed scale scores

for each variable.

3. The generalizability of the findings was limited because

the subjects were not selected randomly from the total population

of two-year college students. They were randomly selected from

within the population of degree seeking students enrolled at

Pasco-Hernando Community College for the Fall term of 1984. Even

though Pasco-Hernando Community College students were felt to be

representative of other community college students, further

empirical study must determine the extent to which the findings

will be applicable in other two-year colleges or other

educational institutions.

4. Subjects were limited to students who had declared

programs on the Pasco-Hernando Community College Program

Declaration form (see Appendix A), of either an Associate of Arts

or Associate of Science degree. This assured at least a minimum

commitment to a college education by the students. This

eliminated students from the sample who declared undecided,

certificate program, vocational certificate program, or personal












objectives as their anticipated goal. By eliminating these four

groups, potential dropouts would not be eligible for the sample,

and therefore would limit the generalizability of the findings.

5. The dropout rate for degree seeking students in this

study was approximately 15 percent. This low dropout rate plus

the small sample size from each individual campus may limit the

validity of some of the statistical analysis.

6. Because of the large number of tests of hypotheses (92)

it might be expected, due to the error rate, that at least one

hypothesis might be found statistically significant in reference

to any given campus.

Significance of the Study

Academic failures are generally accepted as a category of

dropouts from educational institutions. There is a second

category of students, although academically capable, who lack

forms of personal and/or social qualities that prevent them from

becoming properly integrated into the college social environment.

Noel (1976), reporting on the findings of a national seminar on

retention, stated that the seminar group realized the need to

study the interaction between students and their institutional

environment. The University of California, Los Angeles, Academic

Advancement Program has also identified difficulty in adjusting

to campus life as one of the central areas of students' problems

(Moore, 1976). Edwards and Waters (1983) also suggested that lack

of satisfaction with the non-academic part of college could

contribute to students dropping out.












More could be done in two-year colleges toward both

redirecting non-persisters and increasing important recruitment

markets for student enrollment. There exists a great need in

community colleges to service all students who enter. Student

retention should not be promoted to the extent that students see

leaving community college as a failure. Students should feel

comfortable to come and go as they please (Hahn, 1974). There is

a need, however, to provide structured programs to assure that

potentials are realized by both students and the institution.

Students should be given every opportunity to become socially and

academically integrated into an existing college environment.

Programs presently exist that attempt to improve on those

personal and social qualities that are advantageous for social

integration into the college environment. Meyer (1975) concluded

the following concerning the impact of a human development course

on community college students: (a) a human development course can

have a significantly positive impact on college students'

self-esteem, definiteness about self, and sense of purpose and

meaning in life; (b) participation in a human development course

can significantly strengthen students' interpersonal feelings,

particularly feelings toward others; and (c) the human

development course is strongly endorsed by participants as being

personally relevant and meaningful. Similar findings supporting a

human relations-type course by actual participants were reported












by Wall (1979). Recent evidence of programs designed to improve

retention through improving the degree of social integration has

been cited by Beck (1980).

In light of the expressed need to retain students, and

considering the importance of student retention to the sustenance

of institutional vitality, the investigation of Tinto's

theoretical model of retention, specifically, social system

integration (peer-group and faculty-student interaction) and how

it related to dropout decisions in a two-year commuter

institution seemed appropriate. This study contains a description

of the contributions made by peer-group interaction and/or

faculty-student interaction variables toward the process of

dropout decisions described by Tinto (1975). Information provided

by this research should encourage educational institutions to

investigate whether single retention models are applicable to any

one institutional type or if a combination of theoretical models

might be more advantageous.

Definition of Terms

The following definitions were cited from Tinto (1975). A

complete operational definition of each term is provided in

Chapter 3.

Academic integration. Meeting certain explicit standards of

the academic system (grade performance) and identification with

the norms of the academic system (intellectual development).












Academic system. The combined effects of grade performance

and intellectual development on the student.

Actual dropout. Students who did not return the following

term

Background characteristics. Extraneous variables

characteristic of individual students, i.e., family background,

individual attributes, and pre-college schooling.

Commitment. A degree of obligation to a goal or

institution.

Dropout decision. The perceived intent of individuals'

educational plans or the actual decisions to drop out of an

institution.

Faculty-Student interaction. The degree to which students

evaluate both the quantity and quality of their relationships

with their instructors.

Family background. The highest level of formal education

obtained by a student's parent or parents.

Goal commitment. The degree of commitment to complete a

declared college program.

Grade performance (college). The grade point average (GPA)

of a student's academic performance.

Individual attributes. Academic ability characteristics

possessed by the student prior to entering college.

Institutional commitment. The educational expectations

involving specific institutional components which predispose the

student toward attending one institution rather than another.












Intellectual development. The self-perceived growth of a

student in the areas of: general knowledge, reasoning skills,

critical thinking skills, and appreciation of new ideas.

Peer-Group interaction. The degree to which a student

perceives an institution as being receptive socially and the

degree the student feels accepted by the institution. Also the

degree to which a student feels other individuals demonstrate

accepting behaviors and to what degree a student feels accepted

by others.

Pre-College schooling. Characteristic of the school setting

that students were exposed to prior to entering college.

Social system. The combined effects of peer-group

interaction and faculty-student interaction on the students'

integration into the social setting of an institution.

Social integration. The degree of social communication,

friendship support, faculty support, and collective affiliation

students perceive they possess in a social environment.

Organization of Remainder of the Research Report

The following chapters are utilized in the remainder of the

research report. Chapter Two discusses dropouts in general and

additional research and literature that were pertinent to the

investigation. Chapter Three contains the procedures used to

formulate the survey instrument, the pilot study, along with the

complete methodology used in the study. Three campuses of a







19




multi-campus community college were surveyed. Multiple

regression was used to analyze the main hypotheses of interest.

The findings and analysis of data are presented in Chapter Four.

Chapter Five includes a summary of the findings and the

conclusions drawn as a result of the study, as well as

implications for practices and further research.
















CHAPTER TWO

REVIEW OF THE LITERATURE AND RELATED RESEARCH

Postsecondary institutions are facing a serious situation

that has a number of important implications for institutions as

well as students. Considering the economic times and the

evolving demographics of student populations, there has been an

increased interest in research studies concerning student

retention. Until recently little attention has been directed to

the underlying dynamics of the phenomenon of student withdrawal,

rather the main emphasis has been theoretical and descriptive.

A number of theoretical papers (e.g., Bean, 1981; Spady,

1970; Tinto, 1975) have developed conceptual models. These

studies have made an important contribution to our understanding

of dropout behavior in postsecondary institutions. These models

provide both a comprehensive and an explanatory view of attrition

which provides direction to researchers confronted with the

problem of student dropout.

Tinto's (1975) schema has generated perhaps the most

extensive body of research. Using the work of Spady (1970),

Tinto has developed a longitudinal model which attempts to

explain the persistence/withdrawal process in postsecondary

education. This process is largely based on the degree of

personal fit between the institutional environment and the

20












individual student. The concept of personal-institutional fit as

an explanatory concept for student dropout has created much

interest in postsecondary research (Aitken, 1982; Baumgart &

Johnstone, 1977; Bean, 1981). It would follow then that the

validity of Tinto's (1975) model has been the focus of recent

research (Pascarella & Chapman, 1983; Pascarella, Duby, &

Iverson, 1983; Pascarella & Terenzini, 1983; Terenzini,

Pascarella, Theophilides, & Lorang, 1983).

Pascarella and Terenzini (1980) generally supported the

predictive validity of the major dimensions of the Tinto model

after sampling 763 students at Syracuse University. By adding

five institutional integration scales to a discriminant analysis

based on fourteen pre-college characteristics, freshman year

academic performance, and extracurricular involvement, they were

able to increase the correct identification of persisters and

dropouts from 58.2 percent to 81.4 percent.

Additional support for Tinto's model was reported by

Terenzini, Pascarella, Theophilides, and Lorang (1983) in which

an earlier path analytic study (Pascarella & Terenzini, 1983) of

the predictive validity of Tinto's theory of college student

attrition was replicated. The study by Pascarella and Terenzini

(1983) produced 24 significant paths in Tinto's model and

Terenzini, Pascarella, Theophilides, and Lorang (1983) identified

22. Sixteen or 72.7 percent of these paths were common to the

two studies. Both institutions were large, comprehensive,












research-oriented universities, with undergraduate enrollments of

approximately 11,000. The first institution (Pascarella &

Terenzini, 1983) was an independent residential, private

institution and the other was a public residential institution.

Pascarella, Duby, and Iverson (1983) partially verified

Tinto's model using a sample of 269 students from an urban,

commuter university setting, rather than a residential setting.

They suggested that when applied to a commuter institution sample

not all dimensions of Tinto's model functioned according to

expectations.

The validity of Tinto's model appears to be generally

accepted, particularly in reference to residential settings. In

addition, individual variables in Tinto's model appear to have

predictive and explanatory power concerning dropout decisions.

Despite this growing body of research on persistence/withdrawal

behavior in colleges and universities, there lacks sufficient

interest in two-year, community college commuter institutions.

Thus a major purpose of this study was to assess the relationship

between social integration variables and community college

student retention using Tinto's (1975) conceptual framework.

Review of the Literature

Research in the past has generally come to the same

conclusions concerning the particular characteristics of students

who dropout from college. Dropouts generally (1) lack direction

(Summerskill, 1962) and plans for the future (Pancos & Astin,












1968; Wessell, Engle, & Smidchens, 1978); (2) lack institutional

commitment (Gottfredson, 1982; Hackman & Dysinger, 1970); (3)

lack interpersonal orientation and friendship support to some

degree (Astin, 1964a; Fiedler & Vance, 1981; Medsker & Trent,

1968; Spady, 1971; Yourglich, 1966); (4) have less success in

academic areas (Aitken, 1982; Baumgart & Johnstone, 1977; Slocum,

1956); (5) lack either insight and/or capacities for

self-analytic, critical thinking or reject these processes as

important parts of their personality (Daniel, 1963; Faunce,

1966); (6) lack acceptance of themselves to some degree (Lavin,

1965; Stevens, 1956); (7) are less conforming, flexible, or

adaptable (Gurin, Newcomb, & Cope, 1968; Stern, Stern, & Bloom,

1956; Summerskill, 1962); and (8) possess fewer social skills

that provide for positive social integration (Bourn, 1976; Spady,

1971; Tinto, 1975).

Of the many variables associated with attrition, two

particularly stand out: social integration and academic

integration (Spady, 1971; Tinto, 1975; Wider, 1981). The degree

and the direction to which these variables affect college student

retention varies. This variation is partially dependent on the

type of institution under investigation (residential or commuter)

and the uniqueness of the individual study (Pascarella & Chapman,

1983; Pascarella, Duby, & Iverson, 1983; Terenzini, Pascarella,

Theophilides, & Lorang, 1983).












Validity of Tinto's Model in Residential Settings

Research findings related to Tinto's (1975) model by

Pascarella and Terenzini (1983) concluded that

persistence/withdrawal behavior is essentially the result of a

longitudinal process of person-environment fit as theorized by

Tinto. Specifically, background characteristics and

institutional commitments explained little variance in

persistence with reported R2 increases of only .9 percent and 1.3

percent respectively. With alpha levels set at .01, significant

R2 increases occurred with the addition of the academic and

social integration scales. Therefore, both academic integration

and social integration had a direct influence on persistence.

Terenzini, Pascarella, Theophilides, and Lorang (1983)

support the major constructs and their causal linkages in Tinto's

model of college student attrition, with some noteworthy

exceptions. In comparing their study of a public, residential

school (study 2) with Pascarella and Terenzini's (1983) study of

a private, residential school (study 1), the following

differences were reported. In the second study the investigators

found that a significant and direct path was lacking between

students' level of academic integration, as was reported in the

first study and college persistence. This direct path was

present until subsequent institutional commitment was added to

the model. It is suggested that this non-significant direct path

may be artificial due to the fact that academic integration still

has a strong indirect effect on persistence.












Another discrepancy reported by study 2 was that the direct

path between social integration and persistence was marginally

reliable (p<.15) and the influence was negative. This finding

was in direct conflict with Tinto's theory and previous

research. Possible speculative explanations were provided by the

authors of study 2. They suggested excessive social involvement

may reduce time spent studying, and withdrawal due to the

student's recognition of poor academic performance was preferred

over academic dismissal. The authors emphasized the fact that

these were only speculations.

Study 2 confirmed findings in study 1 that indicated

students' background characteristics had no direct effect on

retention of students. Instead, background characteristics

influence was dependent on the student's interaction with the

institution and the student's experience in college.

Validity of Tinto's Model in Non-residential Setting

There is a growing body of research investigating the

appropriateness of Tinto's model in relationship to

non-residential schools. The research of Pascarella and Chapman

(1983) compared the validity of Tinto's (1975) model of college

withdrawal in three different types of institutions: four-year

residential institutions, four-year commuter institutions, and

two-year commuter institutions. The pooled analysis generally

supported a number of Tinto's theoretical expectations, but there

were variations between institutional setting types which were

relative to this study.












The major difference between residential institution and

commuter institution was the role played by academic and social

integration variables. The residential sample reported academic

integration having neither direct nor indirect effect on

voluntary persistence. Social integration was found to have a

significant direct effect on persistence. Conversely, in both

the four-year and two-year commuter institutions, social

integration had neither a direct nor indirect influence on

student persistence. Academic integration indirectly influenced

persistence through its direct effects on institutional

commitment.

The pooled analysis of this study suggest that Tinto's

model is potentially useful in predicting and explaining

persistence/withdrawal behavior. Under separate analysis,

results for different institutional settings may vary

substantially.

Another study concerning the efficacy of the

person-environment fit theory promoted by Tinto was conducted in

a non-residential school by Pascarella, Duby, and Iverson (1983).

Background characteristics were reported as having greater

influence on persistence than social integration. This apparent

influence was partially explained by the differences between

residential and commuter institutions. Commuter students'

environments were found to be generally less rich in terms of

social integration opportunities than residential students'












settings and in addition the students enrolled in these

institutions usually spent less time on campus (Chickering,

1974). From this research one could assume that the background

characteristics which the commuter student brings to college

might have a stronger direct impact on subsequent persistence

than the background characteristics of residential students.

Additional findings showed that academic integration had a

direct effect on persistence, a finding which was consistent with

several previous studies conducted in residential schools (Bean,

1980; Munro, 1981; Pascarella & Terenzini, 1983; Terenzini &

Pascarella, 1978). Social integration was found to have a

negative influence on persistence. This was inconsistent with

previous research in residential institutions (Pascarella &

Terenzini, 1983; Pascarella & Chapman, 1983) but supported the

findings of Terenzini, Pascarella, Theophilides, and Lorang

(1983).

A possible explanation for this negative influence was

given by the authors based on findings by Pascarella and Chapman

(1983). They concluded that students with high levels of social

integration tend to have high affiliation needs. Because of

these needs, these students may be more sensitive to the limited

opportunities for social integration satisfaction than their less

socially integrated counterparts. This may increase the chance

of the socially integrated student transferring to a residential

school in order to fulfill these affiliation needs.












Another possible explanation for this negative influence of

social integration is evidence suggested by Astin (1973) and

Chickering (1974). They reported that commuter college students

are a different population to begin with than residential college

students. These initial differences may be a significant factor

affecting the patterns of variables involved in the retention

process across commuter and residential institutions.

If these explanations are feasible, then what is the

possible flaw in Tinto's model that accounts for this negative

influence. As Pascarella, Duby, and Iverson (1983) suggested,

the flaw may not be in the model, but rather in the population to

which it is applied. When Tinto's model is applied to

residential schools, then Tinto's assumption that the institution

provides ample opportunities for social integration applies. But

when Tinto's model is applied to commuter school samples, the

social integration component of the model may have an influence

quite different from the initially hypothesized model.

Reconceptualization of Tinto's Model

The reported negative influence of social integration on

persistence of college students ( Pascarella, Duby, & Iverson,

1983; Terenzini, Pascarella, Theophilides, & Lorang, 1983) may

have important implications in reference to the association

between person-environment fit and college persistence.

Possibly, this person-environment fit only influences persistence

when institutions provide means for students to achieve social


integration.


P












Pascarella, Duby, and Iverson (1983) have revised Tinto's

model to better reflect their findings. Based on their research,

a reconceptualization of Tinto's model was offered (Figure 2

p.30). This revised model was intended to provide more

explanatory power in a non-residential institution.

The model assumes that the characteristics which students

bring to college will not only influence their interactions with

the college environment, but will also have important direct

effects on persistence. In Tinto's original model, these

characteristics were seen basically as determinants of students'

integration with the academic and social systems rather than

having any direct influence.

Even though social and academic integration were retained

as major elements of the model, some revisions were made.

Academic integration was hypothesized as having a direct

influence on persistence and an indirect effect through its

influence on goal commitment. This was consistent with Tinto's

model. The direct effect of social integration was hypothesized

to be either non-significant (suggested by Pascarella & Chapman,

1983) or negative. This departure of social intergrations'

influence from Tinto's model was based on two assumptions.

First, commuter schools generally provide fewer social

integration opportunities than residential schools and second

this fact may lead to a more complex relationship between social

integration and persistence than originally hypothesized by

Tinto.















U
an
N
0 e-.
to

ui
W
Ha

A


ia a












I Z
I
I I I














0
NZ
H U
H f-4
N -
Il
IA
I 5
I 2
I i-












A
I I


II


II
H
II

II


II
I


I I I I I I I I I I


I Z


SZ N
1 O



II a


I


I I I


a H II H NI I I II U


H z. :


o 0 *)O




.* .
0 Z 0
U U U












0 -







0 0 U) 0
) UOl










*U. *ZU











I II
H






to* 4

(d i d




O O u nn O


II a a H A I H 'a a #a H II a
C I
O H
0
Un -f >, N

Z H1 E 3 -'< r-i u
0 th 0 O e-4 W C O O Io
Sa: X 0 (0 0 a U t II

u (u 0 0 U0 l u 0 l
i F-1 u = m o0 ucOa
U a)
< < r r: K I
O U 1
U I W a
IH aII N a IIN II H I N HI H 'a a:


I I I


0 II



II Z

* U -I
O
I
I
I
I
* a1 -


I I I

A
i a a a


I I I I I I


I,



I 4







I
I a




10



0






a I

a? I
at
aa

at


'a


II
a








N *


n

if


0c0





0
3 4 C
r) a t)
= K


rc




0 C




o o
00






0 i
Ie.










t4
0
1.4









rO
4 -


o "
4




3 1
0u





u,4




0
00




. "

t)
v












This reconceptualization of Tinto's model for use in

non-residential institutions needs to be tested across different

samples. Only a partial description of the model was included so

as not to imply a complete testing of its validity by the present

research investigation. Both models were included to provide the

reader with background information concerning existing research.

This present study mainly investigated the influence of social

integration on persistence in reference to Tinto's model.

Results obtained may, in addition, provide further information

that will enhance the promotion of the revised model of

Pascarella, Duby, and Iverson (1983).

Variables

In order to relate Tinto's model of retention and

corresponding literature to this investigation, it was necessary

to isolate and document characteristics and/or measurements that

most substantially represent the variables under consideration.

The following variables in Tinto's model of retention were

investigated for research support: background characteristics

(consisting of family background, individual attributes, and pre-

college schooling), commitments (consisting of goal commitment

and institutional commitment), academic system (consisting of

grade performance and intellectual development), social system

(consisting of peer-group interaction and faculty-student

interaction), and dropout decisions. The following sections

contain a review of the research in these areas.












Family background. The three most accepted characteristics

of family background were (1) socioeconomic status, (2) parental

education, and (3) quality of relationship.

Research has shown that the socioeconomic status of the

family is inversely related to student dropout rates from college

(Astin, 1964b; Brown, 1980; Pancos & Astin, 1968; Pascarella &

Terenzini, 1983). Even when intelligence was held constant,

children from lower status families exhibited higher rates of

dropout than did children of higher status families (Sewell &

Shah, 1967).

The educational status of the student's parents is another

variable related to dropout from college. The higher the level

of formal education by the parents the more likely the student

will persist in college (Chase, 1970; Jaffe & Adams, 1970;

Kowalski, 1982; Pascarella & Chapman, 1983; Pascarella, Duby &

Iverson, 1983; Ramist, 1981; Spady, 1971; Terenzini, Pascarella,

Theophilidles, & Lorang, 1983). Bennet and Bean (1983) found this

particularly true for black students.

Additional research has indicated that the quality of

relationships between parents and students is an important factor

related to dropout rates from college. The quality of

relationships not only includes the quality of communication

within the family but the expectations that parents and/or family

members have concerning the student's education (Bean, 1981).

College persisters' home environments tend to be characterized by











more open democratic, supportive, and less conflicting

relationships (Congdom, 1964; Trent & Ruyle, 1965; Willner,

1980). In addition, these parents expressed more interest and

offered more advice concerning college experience than

non-persisters' parents (Trent & Ruyle, 1965). Hackman and

Dysinger (1970) also confirmed that the greater the parental

expectations the more likely the student would remain in school.

Slocum, in his 1956 study, received a positive response from 81

percent of college persisters and only 35 percent from the

dropouts to the following question, "Do your parents want you to

finish college?"

Individual attributes. There are many individual

characteristics that could be correlated with dropout behavior,

such as personality traits and attitudinal differences. These

variables certainly should not be discredited concerning their

possible influence. Two particular characteristics have received

the most research support concerning their relationship to

student persistence in college: grade point average (GPA) in high

school and scores on a standardized test.

The grade point average in high school indicates the

students' ability and serves as a measurement of their past

success (Bean, 1982; Blanchfield, 1971; Chase, 1970; Coker, 1968;

Hutchenson, 1980; Jaffe & Adams, 1970; Lavin, 1965; Pancos &

Astin, 1968; Prather, 1982; Taylor & Hanson, 1979; Willner,

1980). Tinto (1975), Edwards and Waters (1982), and Pascarella












(1968) confirmed that GPA in high school is one of the best

predictors of college persistence. Tinto (1975) also supported

Astin's (1973) conclusion that measures of ability, as obtained

on a standardized test, are a significant predictor of college

persistence.

Pre-College schooling. Nelson (1972) suggested that the

characteristics of the high school attended by students were

important variables in determining the probability of a student

either persisting in college or dropping out. Two variables

confirmed as predictors of college persistence by Davis (1966),

Nelson (1972), and St. John (1971) were the students' ability

level and their social status composition in the school. These

two variables not only appear to affect the individuals'

perception of their own ability but also their expectations for

future college education. Additional research by Pascarella and

Chapman (1983) found students' percentile rank in high school and

high school GPA to be consistent predictors of college

persistence.

Institutional commitment. In researching the variables that

may influence an individual's decision to remain in school or

dropout, one would have to investigate to what extent an

individual was committed to a particular institution (Pascarella

& Chapman, 1983). A valid procedure would be to determine

whether specific institutional components exist that would

predispose a student toward attending one institution rather than

another (Pascarella & Chapman; Spady, 1970; Tinto, 1975).












Goal commitment. There seems to be little doubt that lack

of goals in life decreases our motivational drive; thus college

persistence depends on degree of career goals (Day, 1982;

Churchill & Iwai, 1981; Jacobs, Bringman & Friedman, 1982).

Previous research supports this statement: Slocum (1956)

emphasized the need for occupational plans; Summerskill (1962)

reported students need direction; Astin (1964b) implied" What to

study?" was the important question for college students; Wessell,

Engle, and Smidchens (1978) cited the need for a clear purpose

concerning educational persistence; Beck (1980) reported that an

important factor related to college dropout is inadequate

clarification; and Simpson, Baker, and Mellinger (1980)

demonstrated that voluntary withdrawals had less commitment than

persisters. Research reports that if students have some degree

of direction they are more likely to persist in college.

Grade performance. Researchers reported that grade

performance is an important variable in students' college success

(Avakian, MacKinney, & Allen, 1982; Bean, 1982; Edwards & Waters,

1983). Spady (1970) reported that grades are the single most

important factor related to persistence in college. Tinto's

(1975) synthesized research also confirmed the importance of

grade performance. There is little doubt that grades indicate to

what degree students are academically integrated into the college

environment (Creamer, 1980).












Intellectual development. Tinto (1975) stated that

intellectual development deals with intrinsic forms of reward;

this development is the individual's evaluation of the academic

system. Medsker and Trent (1968) referred to intellectual

development as the degree to which students value their college

education as a process of gaining knowledge and appreciating

ideas. Spady (1971) suggested that intellectual development is

exposure to stimulating ideas and experiences.

Though not as important as grade performance, intellectual

development was found to be an influencing variable in students'

decision to drop out (Rootman, 1972; Spady, 1970; Summerskill,

1962; Tinto, 1975). Tinto states,

Though grade performance and intellectual development
appear as separate components of a person's integration
into the academic system, it's clear that persons with
high grades are more likely to be high in measures of
intellectual development. (p. 106)

The distinction may be that grade performance is generally

measured objectively, whereas intellectual development is more

likely a subjective measurement.

Peer-Group interaction. Terms used to describe peer-group

interaction included friendship support (Flacks, 1963), social

fit (Rootman, 1972), supportive groups (Hanson & Taylor, 1970),

and normative congruence (Spady, 1970). In general, researchers

concluded if students perceive themselves as being accepted by

some form of peer-group, college persisters will be enhanced












(Baker, 1980; Beck, 1980; Simpson, Baker, & Mellinger, 1980;

Spady, 1970; Spady, 1971; Tinto, 1975). In addition to research

indicating that informal peer-group associations are related to

persistence (Gardiner & Nazari, 1983), semi-formal

extra-curricular activities were also related to students

persisting in college (Chase, 1970; Ramist, 1981; Spady, 1971).

Faculty-Student interaction. Social interaction with

faculty by students, in various forms and degrees, has been shown

to be related to persistence in college (Bean, 1982; Gardiner &

Nazari, 1983; Noel, 1976; Pascarella & Terenzini 1976; Penick &

Morning, 1982; Tinto, 1982). Additional research has concluded

that some types of interaction are more effective than others.

The strongest form of interaction appeared to be informal contact

concerning subject and/or career related material (Pascarella &

Terenzini, 1977). A general conclusion that has been drawn from

the existing research is that increased quantity and/or quality

of faculty-student interaction is advantageous to persistence in

college (Centra, 1971; Hutchenson, 1980; Keim, Van Allen, &

Anderson, 1982; Pancos & Astin, 1968; Reed, 1981; Slocum, 1956;

Spady, 1971; Tinto, 1975; Tinto, 1982).

Summary of the Chapter

Research has shown that social system integration,

peer-group interactions, and faculty-student interaction

variables can potentially affect whether students decide to

remain in school or drop out. According to Tinto's theoretical











model, the influences of these variables are dependent on

preceding and related variables, background characteristic,

commitments, and academic integration. The strength of the

research concerning social integration may lie in variables not

adequately under the researchers control such as type of

institutional setting. Additional variables not under the

researchers control are family background, individual attributes,

pre-college schooling, commitments, and academic system.

If the research on social integration is to be valid and

useful in the educational system, investigators must control

adequately for extraneous variables. Programs developed to

enhance the educational climate of students should be based on

the best available data on college student persistence.

Retention programs emphasizing human relations type courses

have been promoted in educational settings to improve individual

interpersonal skills and a sense of purpose in life (Meyer,

1975). Little (1971) and Wall (1979) found that a human

relations course significantly strengthened students'

interpersonal feelings, particularly feelings towards others. If

these courses and programs are successful, then according to the

research improved retention should be a by-product.

Recent evidence of programs designed to improve

interpersonal skills and therefore improve retention have been

cited by Beck (1980). He reported the following findings from

Lurleen B. Wallace Junior College: (a) the Human Potential












Seminar (HPS), which is part of the freshman orientation, has

proven to be a successful retention strategy; (b) the dropout

rate for HPS students was less than for students without HPS; (c)

follow-up studies, approximately one year later, showed the HPS

dropout rate was 15 percent, compared to 29 percent among those

without HPS; (d) on an evaluation questionnaire, 99 percent of

the students responded that the HPS was helpful. This research

supported the concept promoted by Tinto (1975) that peer-group

interaction and to some extent, faculty-student interaction are

important ingredients for social integration, and this

integration is advantageous to a successful retention program.

The important issue is whether the increased retention of

college students in these institutions is due to a cumulative

affect of many variables as Tinto suggested or whether programs

of this type can stand on their own merit as vital retention

programs. In addition, can the same results be expected

regardless of the type of institutional setting in which the

programs are being conducted, either residential or commuter.
















CHAPTER THREE

METHODOLOGY

Using Tinto's (1975) conceptual framework, the purpose of

this study was to determine what relationships exist between

social system integration variables (peer-group interaction and

faculty-student interaction) and student dropout decisions in a

two-year community college.

Instrumentation

The survey questionnaire developed for this research

investigation was formulated by this researcher using the most

recent and popular measurements of the variables under

consideration. The questionnaire provided measures of each of the

concepts in Tinto's (1975) theoretical model under investigation,

except for those provided by the institution: the students' grade

point average, number of college semester hours completed, the

actual dropout rate and the Guidance Placement Test scores

(Appendix B and C). In addition, the instrument obtained

information regarding demographic characteristics, i.e. social

security number, age, sex, and race. The instrument contains

forty-one items and appears in Appendix D.

Description of Student Survey

The Student Survey Questionnaire (Appendix D) provides

measures of five main areas under consideration: background

characteristics (family background; individual attributes;

40












pre-college initial commitments (goal and institutional);

academic system (academic grade performance and intellectual

development); social system (social peer-group interaction and

faculty-student interaction); and dropout decisions. In

addition, item 8 measures whether an individual has completed a

human relations type course. The specific breakdown by category

and item number follows.




Category Item Number

Social Security Number................................. 1
Age .............. .................................... 2
Human Relations Course............................... 8
Background Characteristics
Family Background
parental education............................ 7
Individual Attributes
sex........... ............................*.. 3
race.......................................... 4
Pre-college Schooling
high school percentile rank.................. 6 6
high school grade point average................ 5
Initial Commitment
Goal Commitment
highest expected degree....................... 9
importance of graduation....................... 11
Institution Commitment
rank of subject institution as choice......... 10
probability of transfer...................... 13
confidence of choice of subject institution... 12
Integration
Academic System
students' perceived intellectual development.. 15-21
perceived faculty concern.................... 22-26
faculty out-of-class contact (academic)........ 14(1,2,3)
Social System
student extracurricular activity............... 40
student-peer interaction.......... .......... 27-33
faculty-student interaction..................... 34-38
faculty out-of-class contact (social)......... 39(1,2,3)
Dropout Decisions
intent of future educational plans............ 41












Development of the Student Survey

The student survey utilized in this study was formulated

from the combined information received from the following

research:

A Multiinstitutional, Path, Analytic Validation of Tinto's
Model of College Withdrawal by E.T. Pascarella and D.W. Chapman
(1983);

A Test and Reconceptualization of a Theoretical Model of
College Withdrawal in a Commuter Institution Setting by E.T.
Pascarella, P.B. Duby, and B.K. Iverson (1983);

Predicting Voluntary Freshman Year Persistenece/Withdrawal
Behavior in a Residential University: A Path Analytic Validation
of Tinto's Model by E.T. Pascarella and P.T. Terenzini (1983);

and Path Analytic Validation of Tinto's Theory of College
Student Atrition by P.T Terenzini, E.T. Pascarella, C.
Theophilides, and W. G. Lorang (1983).

Variables

As presented in Figure 1, Tinto's model consists of five

major constructs or variable sets in a causal sequence: (a)

background characteristics (family background, individual

attributes, and pre-college schooling); (b) initial commitments

(goal commitment and institutional commitment); (c) integration

(academic system and social system); (d) subsequent goal and

institutional commitments; and (e) withdrawal decisions.

Five constructs, background characteristics, initial

commitments, academic system, social system, and withdrawal

decisions, were operationalized as follows. In addition the

measurements of the variables under consideration were selected

using the following criteria.












Background characteristics

The background characteristics' value for each student was

the sum of the Z-scores from family background, individual

attributes, and pre-college schooling divided by 3. These

Z-scores resulted from changing raw scores to standard scores,

adding a constant of 10, and expressing the scores in standard

deviation units. This approach of standardizing the scores was

used by Terenzini, Pascarella, Theophilides, and Lorang (1983)

and was adopted for this study to be consistent for comparison

purposes.

Family background. Parental education was the most

frequently used measurement of family background in the studies

reviewed. Two studies, Pascarella and Terenzini (1983) and

Pascarella, Duby, and Iverson (1983), used, in addition to

parental education, parental income and parental financial

support respectively. Neither of these measuremnets were used in

this study since the average age of PHCC students was 26 years

and they were considered adult learners.

Family background was a single variable, item 7, consisting

of the average of each parents' level of formal education. A

seven point scale was used with a value of 1 being assigned to

"some grammar school" to a value of 7 assigned to "graduate

degree." After dividing the total of both parents' score by 2,

this raw score was converted to a Z score or if only one parent's

education was reported, then the single score was converted to a

Z score, then a constant of 10 was added.












Individual attributes. Scholastic Aptitude Test (SAT) scores

were used in the majority of studies for. academic aptitude with

American College Test (ACT) scores being used in one additional

study. Since the majority of PHCC students do not take the SAT or

ACT tests, another standardized instrument was used to measure

academic aptitude. The Comprehensive Guidance Placement test

(English and Reading) is administered to virtually all entering

students at PHCC and was the instrument utilized. The

Comprehensive Guidance Placement test scores (English and

Reading) were furnished by the PHCC registrar. These two scores

were divided by 2 after being converted to Z-scores and adding a

constant of 10, resulting in a single Z score measurement for

individual attributes. Due to the variety of math competencies

of students entering PHCC, three different levels of exams must

be administered. These exams vary to such a degree, that for

standardization purposes the math scores were not used.

A few studies requested information concerning "major area

of study." The wording in these studies was inconsistent, and

these studies were conducted in four-year institutions, rather

than two-year; therefore this item was omitted as one of the

measurements of individual attributes.

Sex, item 3, was coded O=male and 1-female. Race, item 4,

was coded 0Owhite and I-non-white. To be consistent with

previous research these items were not included as measurements

of individual attributes. Both items are generally considered as

descriptive variables in reference to dropouts.












Pre-college schooling. The two most popular measurements of

pre-college schooling in the research were (a) students'

percentile rank in high school, item 6, and (b) indication of

high school grade point average or grade achievement, item 5.

Other variables, such as high school preparation (Terenzini &

Pascarella, 1983) and extracurricular activities in high school

(Pascarella & Terenzini, 1983), were used in previous research,

but their isolated usage eliminated them from consideration in

this research.

Percentile rank in high school used a seven ordinal category

assigning a value of 1 to "70% or below" through a value of 7 for

"top 10%," and the grade point average in high school used a

seven point scale assigning a value of 1 to "D or below" through

a value of 7 for "A/A+." Each score was converted into Z-scores,

adding a constant of 10, summed, then divided by 2.

Initial commitments

The commitment value was the sum of the Z-score of goal

commitment, plus ten, and institutional commitment, plus ten,

divided by 2.

Goal Commitment. Highest degree expected, item 9, and

importance of graduating from college, item 11, were consistently

used as measurements of goal commitment. A value of I was

assigned to "Associate of Arts/Science" through a value of 6 for

"LL.B. or J.D. (law)." In reference to importance of graduating,

a value of 4 was assigned to "extremely important"












through 1 for "not at all important." Each score was converted

into a Z-score, adding a constant of 10, -summed, then divided by

2.

Institutional commitment. The following were used by all or

a majority of the studies to measure institutional commitment:

(1) institutional rank as a college choice, item 10; (2)

probability of transferring before graduation, item 13; and (3)

confidence or satisfaction that choosing the subject institution

was the right choice, item 12.

A value of 4 was assigned to "1st choice" through a value of

1 for "4th choice" on item 10, a value of 5 to "SD" through I for

"SA" on item 13, and a value of 4 to "extremely confident"

through 1 for "not at all confident" on item 12. These three

values were converted to Z-scores, adding a constant of 10 to

each, summed, and then divided by 3.

Academic system

Academic system was operationalized as the sum of the

following scales or variables: (1) grade point average, provided

by the registrar; (2) a seven-item factorially derived scale

measuring a student's perceived level of intellectual

development, items 15-21; (3) a five-item, factorially derived

scale measuring a student's perception of faculty members concern

for student development and teaching, items 22-26; and (4) the

frequency of a student's out-of-class contact with faculty of 10

minutes or more for each of the following purposes: (a) "to get

basic information and advice about my academic program";












(b) "to discuss intellectual or course-related matters"; (c) to

discuss matters related to my future career", item 14(1,2 & 3).

The factorially derived scales were originally developed by

Pascarella and Terenzini (1980). Initially 55 items were

constructed and subsequently reduced to 34 items. The specific

factorially derived scales used in this study to meausre the

academic system were taken directly from Terenzini, Pascarella,

Theophilides, and Lorang (1983) and Pascarella and Terenzini

(1983). In these two studies an internal consistency (alpha)

reliability of .60 and .64, respectively, were reported for the

academic scale.

The above mentioned four variables were used in virtually

all research investigated. Value placement and/or Z score

conversions were as follows: (1) GPA was converted to a Z-score,

plus ten; (2) a value of 5 was assigned to "SD" through 1 for

"SA" for item 15 and the reverse order being used for 16-21. The

sum of these items was divided by seven and converted to a

Z-score, plus 10; (3) a value of 5 was assigned to "SA" through

1 for "SD" for items 32, 23, and 26 and the reverse values for

items 24 and 25. The sum of these items was divided by 5 and

converted to a Z-score, plus 10; (4) The sum of each of these

items was divided by 3 and the Z-score calculated.

The sum of these four Z scores, calculated from the raw

scores of the academic system variables, was divided by 4. This

quotient represented the academic system measurement.














Social system

Social system integration was operationally defined as the

sum of the Z-scores of peer-group interaction and faculty-student

interaction variables divided by 2. The factorially derived

scales used to measure peer-group interaction and faculty-student

interaction were originally developed by Pascarella and Terenzini

(1980). Terenzini, Pascarella, Theophilides, and Lorang (1983)

reported an internal consistency (alpha) reliability of .47 for

these items and Terenzini and Pascarella (1983) reported .46.

Peer-Group interaction. The vast majority of research

studies operationalized peer-group interaction as the sum of the

scores for (1) the number of hours spent per week in

extracurricular activities, item 40; and (2) a seven-item

factorially derived scale measuring the extent and quality of a

student's interaction with peers, items 27-33. Items 27-30 of

the peer-interaction question were assigned a value of 1 for "SD"

through 5 for "SA" and the values were reversed for items 31-33.

The total score was divided by 5 and converted into a Z-score,

adding a constant of 10. This Z-score was summed with the

Z-score conversion of the raw score for extracurricular

activities, then divided by 2.

Faculty-Student interaction. The majority of the research

studies operationalized faculty-student interaction as the sum of

the scores for (1) a five-item factorially derived scale

measuring the quality and impact of a student's out-of-class

contact with faculty, items 34-38, and (2) the frequency of











student non-classroom contact of 10 minutes or more with faculty

concerning personal and/or social matters, item 39(1, 2, & 3).

For the items concerning quality of faculty contact, a value of 5

was assigned to "SA" through 1 for "SD" for items 34, 36, 37, and

38. The reverse value assignment was used for item 35. The

frequency scores for items 39(1), 39(2), and 39(3) were summed,

divided by 3, then converted into a Z-score, adding a constant of

10. The frequency score was summed with the Z-score which was

the result of totaling the values from the items 34-38, dividing

by 5, then transforming into a Z-score, adding a constant of 10.

The sum of these two Z-scores was divided by 2, resulting in the

faculty-student interaction measurement.

Dropout decision

The majority of the research studies used an intent item

(intention of remaining in school) and official records from the

registrar to measure withdrawals. Dropout decisions, in this

study, was operationally defined as the student's perceived

educational plans for the future, item 41. Each student was

placed in a group according to his or her degree of intent to

persist as indicated by his or her selection of one of these five

options: (1) I plan on returning to PHCC next term; (2) I plan on

returning to PHCC but not necessarily next term; (3) I plan on

attending another institution next term; (4) I plan on attending

another institution, rather than PHCC, but not necessarily next

term; or (5) I am not planning to attend this or any other

institution anytime in the foreseeable future. These responses











were then converted to a dichotomously coded dependent measure

with response (1) indicating persistence (coded 1) and responses

(2), (3), (4), and (5) indicating withdrawal (coded 0). In

addition, the dropout decision variable was also operationalized

as actual persistence or non-persistence the following term based

on school records.

Four university professors reviewed the preliminary

instrument, Dr. Al Smith, Dr. Steve Olejnik, Dr. Gordon Lawrence,

and Dr. Paul Fitzgerald. These individuals were instructed to

take note of possible areas of revision in terms of clarity or

wording in instructions, possible ambiguity of items, and

relevance of items to the variables under consideration in this

research study.

Pilot study. The investigator conducted a pilot study to

determine the suitability of the instrument format and to provide

data for analysis of the items.

Subjects. The pilot sample consisted of a total of 40

students. Ten subjects were selected randomly from the East

Campus of PHCC, 10 subjects from the North Campus of PHCC, and 20

subjects from the West Campus of PHCC. The West Campus has a

student population approximately equal to the East and North

combined; therefore twice the number of subjects was required to

increase the generalizability of the pilot study findings. A

total of 38 usable surveys was returned. The survey












questionnaire proved to be an adequate and serviceable instrument

in investigating variables of interest.

The same two criteria for selection that were used in the

present study applied to the pilot study with the exception of

number of hours completed. In the pilot study students were

eliminated, as possible subjects, if they were within nine hours

of graduation, indicating a persister. These students from the

pilot study could graduate at the end the 1984 Summer term by

taking an acceptable load. Therefore, students were eliminated

from participating in the pilot study if they had accumulated more

than 50 semester hours.

Present Study

Subjects

The study sample consisted of a total of 200 students at

Pasco-Hernando Community College who were enrolled for the Fall

term of 1984. A total of 152 useable questionnaires were

received for a 76 percent response rate. Since the West Campus of

PHCC was approximately twice the size of either the East or North

campuses of PHCC, which were approximately equal in size, 100

students were sampled from this campus and 50 from each of the

remaining two campuses. A random selection process was used to

select subjects from all students who had declared an Associate

in Arts degree and/or Associate of Science degree on their

Program Declaration Form (Appendix A). This limitation (a)

assured relatively equal degree of commitment to the college on

the part of the students and (b) attempted to eliminate











confounding variables involved with individuals who might be

taking a limited number of courses to complete a certificate

program, vocational certification program, personal objectives

(i.e.,teacher recertification) or have declared themselves in the

category of undecided. The only other students who were excluded

from participation in the study were individuals who were within

15 semester hours of completing their degree program. These

students, by nature of their accumulated hours, were considered

persisters.

Data Collection

The survey questionnaire was distributed to the randomly

selected students, in their classes, on each of the three

campuses by their classroom instructors. This took place during

the eleventh week of the fifteen week term, three weeks after the

distribution of mid-term grades and one week after the last day

of official withdrawal from classes without penalty. Any

student, from the original sample, withdrawing prior to the

distribution date was contacted, by mail or personally, in an

attempt to complete the survey questionnaire. Each participating

student received in class a survey questionnaire (see Appendix D)

and was asked to complete the form in class as accurately as

possible. The instructions provided to the monitors are found in

Appendix E. All forms were received and then analyzed by the

researcher.











Data Analysis

The following null hypotheses were investigated:

Hypothesis 1. A significant proportion of the variation in

student dropout rates is not explained by selected social

system variables after controlling for student background

characteristics, student commitments, and the student

academic system (student grade performance and student

intellectual development).

Hypothesis 2. A significant proportion of the variation in

-student dropout rates is not explained by peer-group

interaction after controlling for student background

characteristics, student commitments, the academic system,

and faculty-student interaction.

Hypothesis 3. A significant proportion of the variation in

student dropout rates is not explained by faculty-student

interaction after controlling for student background

characteristics, student commitments, the academic system,

and peer-group interaction.

Hypothesis 4. There is no interaction effect between peer-group

interaction and faculty-student interaction.

Hypothesis 5. There is no significant difference between white

and non-white student population measurements on the social

system variable.

Hypothesis 6. There is no significant difference between white

and non-white student population measurements on the

peer-group interaction variable.











Hypothesis 7. There is no significant difference between white

and non-white student population measurements on the

faculty-student interaction variable.

Hypothesis 8. There is no significant difference between male and

female student measurements on the social system variable.

Hypothesis 9. There is no significant difference between male and

female student measurements on the peer-group interaction

variable.

Hypothesis 10. There is no significant difference between male

and female student measurements on the faculty-student

interaction variable.

Hypothesis 11. There is no significant difference between

students who have taken a human relations type course and

students who have not on student measurements of the social

system variable.

Hypothesis 12. There is no significant difference between

students who have taken a human relations type course and

students who have not on student measurements of the

peer-group interaction variable.

Hypothesis 13. There is no significant difference between

students who have taken a human relations type course and

students who have not on student measurements of the

faculty-student interaction variable.

Hypothesis 14. There is no correlation between student semester

hours completed and the social system variable.












Hypothesis 15. There is no correlation between student semester

hours completed and the peer-group interaction variable.

Hypothesis 16. There is no correlation between semester hours

completed and the faculty-student interaction

variable.

Hypothesis 17. There is no correlation between student age and

the social system variable.

Hypothesis 18. There is no correlation between student age and

the peer-group interaction variable.

Hypothesis 19. There is no correlation between student age and

the faculty-student interaction variable.

Multiple linear regression analysis was used to test

hypotheses 1-4. Hypotheses 5-13 were analyzed using a

t-statistic. Hypotheses 14-19 used a Pearson product moment

correlation. An alpha level of .05 was used in each case to

reject the hypothesis.

In all instances where scales operationalizing components of

the model were constructed from variables with different metrics

(e.g. academic and social integration), the same two-step

procedure was employed. First, each individual item or scale was

standardized to provide the same metric (Z score), and second, a

constant of 10 was added to eliminate negative numbers. The

scale was then formed by summing across the standardized items

(Armor, 1973-1974).











Multiple linear regression analysis was employed to

determine the incremental increase in the explained variance in

the persistence/withdrawal behavior (R2 increase) associated with

different variable sets in Tinto's model. The sets of variables

were entered in an a priori, hierarchical manner consistent with

the causal sequence of the model: (a) background characteristics

(family background, individual attributes, and pre-college

schooling), (b) commitments (goal and institution), (c) academic

system, (d) peer-group interaction, (e) faculty-student

interaction, and social system.

Summary of the Chapter

This chapter contains the methodological procedures used in

this investigation. The development of the survey questionnaire

was described and justified. Finally, this chapter contains a

description of subjects, data collection, and data analysis

techniques.

















CHAPTER FOUR

FINDINGS

The purpose of this study was to determine what relationship

exists between social system integration variables (peer-group

interaction and faculty-student interaction) and student dropout

decisions in a two-year community college. The investigation

used Tinto's (1975) conceptual framework model to guide the

research. The student's home campus, age, sex, race, total

semester hours completed, and whether a student had completed a

human relations type course were the demographic variables

considered. Background characteristics, commitment (goal and

institutional), academic integration, and social integration

(peer-group interaction and faculty-student interaction) were the

independent variables under investigation.

Two different measurements were used for the dependent

variable. The first dependent variable was students' perceptions

of whether or not they would return the following term. The

second dependent variable was whether the student actually

returned the next term. This was determined from the official

registration of Term II 1985. Even though the total dropout rate

was similar for both, 13.8 percent for dropout perception and

14.5 percent for actual dropout, the individuals differed

greatly. Of the 21 students who replied on the Student Survey

they would not be returning the following term, only 10

57












ultimately withdrew. This resulted in 52.4 percent change of

response. Of the 131 students who said they would return to PHCC

the following term, 119 actually did return; this resulted in a

9.2 percent change in students' decisions concerning dropout.

Therefore, two different dependent variables were independently

measured and analyzed: (1) students' perceptions of whether they

would return to Pasco Hernando Community College the following

term and (2) actual dropout rate the following term.

The researcher pilot tested the survey instrument to be used

in the study on students attending all three campuses of Pasco

Hernando Community College in the Summer term of 1984.

Permission was obtained from the administration of PHCC to run

the full study on the student body of PHCC the Fall term of

1984. The Student Survey (appendix D) was designed to obtain

information concerning the independent and dependent variables

under investigation.

The original intent of this study was to test 19 hypotheses

on the total population of 152 subjects. An additional

consideration of interest developed; specifically, was there

differences between the three individual campuses that comprise

the total population in reference to the 19 hypotheses. Because

of this concern a total of four different population groups was

tested.

This chapter describes the results of the study. First,

demographic data of the participants are presented and discussed.

Then results pertaining to each of the hypotheses under












investigation are described in terms of four groups: total campus

of PHCC, East Campus of PHCC, North Campus of PHCC, and West

Campus of PHCC. The four groups are discussed in reference to

the influence social integration, peer-group interaction,

faculty-student interaction, and the interaction of peer-group

and faculty-student variables had on dropouts with reference to

both dependent variables, perception of dropout decision, and

actual dropout decision.

Multiple linear regression was used to test each of the

first four hypotheses. Each of the four hypotheses was tested for

each of the population groups, total campus, East Campus, North

Campus, and West Campus, using first the students' perceptions of

their dropout decisions as the dependent variable then the

students' actual dropout rate. The total number of analyses was

32. The next nine hypotheses investigated whether there was a

difference between white and non-white students, male and female

students, and students who had taken a human relations course and

those who had not on the social system, peer-group interaction,

and faculty-student interaction variables. A t-test was used to

test these hypotheses. Each of the nine hypotheses was tested

using each of the four groups as separate populations. This

resulted in 36 different analyses. The remaining six hypotheses

investigated whether there was a relationship between accumulated

hours and age with social system, peer-group, and faculty-student

interaction variables. Each of the six hypotheses was tested

using each of the four groups as separate populations. This












resulted in a total of 24 analyses. The total number of tests

for the four separate groups was 92. A significance criterion of

.05 was used to reject each of the 19 hypotheses.

Finally, an ANOVA was used to analyze differences between

individual campuses on variables of interest. Variables that

resulted in a significant F-statistic were analyzed with a

computed Bonferroni-statistic. A .05 level of statistical

significance was used to indicate a difference.

Descriptive Analysis

A subprogram of the Northwest Analytical Statpak computer

program was used to calculate the frequencies and percentages for

the demographic variables. Other subprograms of the NWA were

used to calculate the ANOVA, t-statistic, correlation

coefficients, and multiple linear regression results. A total of

200 questionnaires were distributed with 152 useable surveys

being returned. This resulted in a 76 percent response rate

available for analysis.

Demographic Variables

The data regarding the demographic variables are presented

in Table 1. Inspection of Table I revealed the proportion of

usable responses from each of the three campuses was proportional

to the total student population attending those individual

campuses. The total population under investigation was 1146

students, East Campus (23 percent), North Campus (25.7 percent),

and West Campus (51.3 percent). All statistics

























'0 C4

SCO
4-4





0%o


0% Co .r -~ clh

CO0 -9 4 C'4



-4nrc?- c


ecOn *4-*000







COCO(M'<4OO000


*0
>0 -


%O-4uOuM00

Nu%~C'40'40


%0 '0M (l c0 -40 0
Cq












T M M Mc tn p















0 (0 N 0 0 o s









4M N rl M 4 4 T ir mtr


00 C4
A* -




t o
















**






In in
-4 N'


MCN
* *
NI C






NN





(cV(








(r
* *



M


r C4o


C-4 c
('N


*
Cl o






o'
4 -4
99


I0
Le) 0


IC 0

Src


























C %o IT N 4 mC

4 IA f s4 U' N 14


* 4 N -4




SNW .


OOONO
N(N


cM \O
(C.


00owo00 ^0

c-4 C4 n ino
-4 -4
*****cn
Ncam


M 0 iM IA N N 0
-4 -4 nru


4 4 N --











N Cl 4 tn
111111+
-00% CM N '0
-N cM o cr n


In kn
* 4







NC


C M


C 0C
4(1 4(


pi


NFC

40O




Ifs 0


W~ U

u
U


n-4 -


m N


C


3 r(
0


0
0.


C
u 0

0 0

o u
0

Qa











concerning the total population were obtained from the Office of

the Registrar. The sample consisted of 152 students, East Campus

(25.6 percent), North Campus (25.6 percent), and West Campus

(48.6 percent). There were slight variations in percentages

between the sample and the actual percentage enrollment figures

for the total population of each of the three campuses.

Slightly more than half of the respondents (54.6 percent)

were between the ages of 17 and 21. The remaining age groups and

percentages were, 22-26 (13.2 percent), 27-31 (13.8 percent),

32-36 (5.9 percent), 37-41 (3.9 percent), 42-46 (5.3 percent),

47-51 (1.3 percent), 52-56 (1.3 percent), and 57+ (.7 percent).

The mean age for the sample subjects was 25.1, which approximated

the mean age of the total student population of 26.

The females out-numbered the males slightly better than two

to one, with 67.1 percent of the respondents being female and

32.9 percent being males. These percentages are comparable to

the total population, which contained 61.5 percent females and

38.5 percent males. Each of the three campuses was represented

by percentages that were within 5 percent of the percentages of

their respective total populations.

The white and non-white participants showed the same

proportional trend that was reported for the different sexes.

The sample population of 93.4 percent white and 6.6 percent

non-white was within 2 percent of the total college student

population of 5.4 percent non-white and 94.6 percent white.

Different percentages were reported for each campus, but these












differences adequately reflected the percentages of white and

non-white students attending those individual campuses.

The percentage of students, in the sample, who had had a

human relations type course was 38.8 percent. Similar

percentages were found on the East and West campuses, 38.5

percent and 36.5 percent respectively. A higher number of

subjects had a human relations type course on the North Campus

(43.6 percent). Even though accurate records were not available

for the total population, in reference to this variable, this

researcher's general impression was that similar participation

had occurred for the total population.

The next item listed in Table 1 revealed the frequency range

of accumulated credit hours. Students with 1 to 9 hours account-

ed for (28.2 percent) of the total population, 10 to 18 (28.9

percent), 19 to 27 (18.4 percent), 28 to 36 (11.1 percent), 37 to

45 (7.8 percent), 46 to 54 (4.6 percent), and 55+ (.6 percent).

Differences appeared in accumulated semester hour frequency

figures between campuses. The total campus and East Campus had

similar percentages of students with less than 28 semester credit

hours accumulated, 75.5 percent and 76.8 percent respectively .

These percentages indicated a larger number of freshmen in

attendance than sophomores. West Campus had an even larger

proportion of freshmen with 85 percent. North Campus appeared to

have the most balanced freshmen and sophomore classes with 56.3

percent of the students having less than 28 semester hours

accumulated.











The mean for accumulated semester credit hours for the

sample was 19.0 which approximated the total population mean of

20.2. This slight difference may have been attributed to the

fact the sample had a limit on the maximum number of accumulated

hours a student could have in order to be eligible for the

study.

The last variable listed was dropout decision, both the

students' perceptions and actual dropout decision. The number of

individuals who perceived themselves as not returning was 21.

This was very close to the actual dropout rate of 22. The range

between campuses was 3.2 percent for the perceived decision and

5.9 percent for the actual decision. East Campus was very

consistent with 15.4 percent recorded for both perceived decision

to drop out and actual dropout decision. North Campus lost two

fewer students than perceived which decreased their perceived

percentage from 15.4 percent to a 10.3 actual dropout

percentage. West Campus lost three more students than

anticipated which increased their percentages of dropouts from a

perceived dropout rate of 12.2 percent to an actual dropout rate

of 16.2 percent.

Additional statistics of comparative value, which are not

recorded in Table 1, were the English and reading scores from the

Guidance Placement Test (Appendices B and C). These entrance

tests reflected students' ability in reading comprehension and

English usage. The total college population mean for reading was

24.7 and 25.1 for the sample under consideration. The











total college population mean for English was 26.5 and 26.8 for

the sample. These data, combined with the demographic

information in Table 1, reflected very similar data between the

sample and the total population under consideration.

Correlations

Pearson product moment correlations for all variables under

investigation for the sample are shown in Table 2. Correlations

for East Campus are listed later in Table 3, North Campus Table

4, and West Campus Table 5. The dichotomous variables were coded

as follows: persisters 1, withdraws 0; males 0, females 1; white

students 0, non-white 1; and having a human relations type course

1, not having the course 0.

Most of the intercorrelations for total campus were

non-significant. Only fourteen, out of 91 correlations, were

found to be significant at the .05 level. Because retention was

the dependent variable the two significant correlations relating

to retention are presented first. The higher one's background

support scores, the more likely one would actually remain in

school (r=.197). The only other variable relating to actual

dropout rate was perception of dropout decision and it showed a

positive correlation (r=.377) with the actual dropout rate. The

following correlations are presented in the order in which they

appear in Table 2. Older students seemed to be academically

integrated to a greater degree than younger students (r-.272).

More females had had a human relations type course (r--.691), and

were more committed than males (r=.206). The more

























4 O0
IT



.K

C 1 -4
0 ',







0 0
I I


c IT r-

- *4 -

K*


in T



c0
*~


I-4




1
0



























0


0
14



C


0
-M

















4.1
0
CA
0



0

o4
0





-4


-I
8)

0
















4)




0.





C.0 o


V 41
3
U) U


0 C
0 0
c-:-
*O *J

4) Os
cr




< a;


c
0







0




-4







0 0
, oL
I 1-1


U U
0 o
b> LI


c:
0
.jI


u
I O
u U




0 0
0 0
S.

o o
p a


0 .4 N '-
'0 r ~c- 0 --


I 4
I 0
I r -i

I I i



.3


I-a


'-4


41
0

V

















0




















CM

























N Cl 0 B0
0 0 0


Cl


L O OD 0 -4 0 % Q
0 0 0 0 0 4 0 0
O .
I I I I I




0 0 0 0 0 0 0 C0
* *
I I I I I I






I I

0 0 0 0 i 0 0




0 0 0 0 0 0 0

I I I I



4 0 0 O O '.0

* *
I I
('4 0o l o N "7 .0








04 0 0 IT D
0 ( 0 0- 0 0











I0



0 O O V
o W
CM 0 )0










4) tv u 2 0
oOC 4 c o 3 o -
























o0

eq 0 0 c


0 0 2 0 0 0
* *








00 (LCCU)

1.. *

Cl 0
0 -U *i














* *
*- m fi -^ /i a r c


0 en co N






S0 0
0 0
* *






* .
* *



M
4C
0
Cl


0

-C



u
c 0

u c
(a ( 0
*i *

C

3 eU









*
1 0
1.. p ",


c I
0
a .

0 0 I


a. <



I

a 0 10
* Iv
CM m IM r


'n


'C
<












credit hours a student accumulated, the higher were the scores on

the three measurements that comprised background characteristics

(r=.194) and the more likely the students were to have had a

human relations type course (r-.384). The following

relationships were found to have a positive correlation with

academic integration: commitment (r=.296), peer-group interaction

(r=.208), faculty-student interaction (r=.415), and social

integration (r=.200). Peer-group interaction also had a positive

correlation with faculty-student interaction (r-.318) and social

integration (r-.575). Finally, faculty-student interaction was

positively correlated with social integration (r=.612).

East Campus

Among the 78 correlations related to the East Campus, only

eight were significant at the .05 level (see Table 3). The two

statistically significant correlations relating to the dependent

variable are presented first. The higher the academic

integration the more likely the students were to actually stay in

school (r=.446). Second, females perceived themselves as

returning to school more often than males (r=.486). The

remaining six statistically significant correlations are

presented in the order in which they appear in Table 3. Female

students scored higher on the peer-group interaction scale

(r=.357) than male students. The more credit hours accumulated

by a student indicated a greater likelihood of a student taking a

human relations type course (ri.476). The more a student

indicated a commitment to goals and institution,
























N


I.







sw

0
fu


C
0
$4
1-4





















0

0




-4







,I
0
us







C










0
us
0
















4)
0
















E- -
0
-4

'Ai
u,




















E*4 04


- c- C M

'. N -


*r(
u



u
-,4







-M
c c
C U
UI ()


" C

u v
4 r4
o o
0 0
rL O u


0
-A


u


i1 4J
e U


0 0

0 0
a a
o o


-* 0 N 0 -
* .


i I






-I N N



- 0 4








*-





0
h N




loo


4) (0


0 .
0 41 1

0 4 0)


Ai r.
*r- U L


1 1 41 E



0 3 EO 4
) X u U I u e 0
00 4) 0 m0 0
45: u0 d


0




-4
O



































0

04

























CS N a r -i
n % .0 Cl)
N N -





N ,0 0' N 1n 0'
O 0 4 N 4 -
CN 0 C 0 0
* *.





c N N N
in 4 inI 0 0
0 0 0






0 0O 0 4 N N
N 0\ 0' an N c
o N M 0 0 0
* C






4 in o c '







N 4 N 0 '.0 0
N 04A 4 0
N C N 4 o Cl
h~gr~nI


C
S0

U U
4 L) C 1I
O I
L 4) 0 Li U
0 4 U C



0o r 00 u
M 0 0 w < 4)
41 U ( 1 U
S0 o $ C 3.

Si C C 3 m


E 0 DD e 4) -4
:I C IV s

00 4) 0 U o 0 u 4) (
C4 1; o e 1- 0 >
3. 0


C C
O 0
4 4

.. 4) 4

C 0 3
u U


-4 3 3

0 0 0
0) 0 0


- M eC
-4 -4


'a 4 M N N -



C '0 n cc O



N T0 C 4 C
N 0 ,- N


* E













o0 0







l C


ft





*


U,


4:
'-4
4:











the higher the tendency for academic integration (r=.360).

Peer-group interaction was positively correlated with

faculty-student interaction (r=.506) and both peer-group

interaction (r=.522) and faculty-student interaction (r-.512)

were correlated with social integration in a positive direction.

North Campus

Eleven of the 78 correlations involving North Campus

students were found to be significant at the .05 level (see Table

4). There was only one statistically significant correlation

involving dropout rate at this level. Students' perceptions of

dropout decisions were positively correlated with actual dropout

decision (r=.324). The remaining 10 correlations are presented in

the order in which they appear in Table 4. Older students tended

to have lower levels of background characteristics (r--.511) and

females appeared more committed (r-.379) than males. A human

relations type course was more likely to be taken by students

with more accumulated credit hours (r=.340). Those students with

higher peer-group interaction scores (ri-.384) and social

integration scores (r=-.399) were less likely to have taken a

human relations type course. Students with higher academic

integration had higher levels of faculty-student interaction

(r=.482) and social integration (r=.341). Peer-group interaction

was positively related to both faculty-student interaction

(r-.385) and social integration (r-.836). Faculty-student

interaction was also related to social integration in a positive

direction (r=.817).
























ml Go r> -4
O 0 M 0
0 0N 0 0




S0 -O
N 0

I I




44





S 0
cn







cI
cn


0



a
0













V
C.

0
U


4)

0









V
0
4















r.
0
t)


















0


0
Lli















0

















$4
0
14












kUl
0






0
a

>







4)
v





$4

0

L











0



04
0
10





0 4
E o
M,
|j
0
'M

LI
i-l
4J



Ij
u
*u




*a
o


(i ,

u (d
r\ )


- ccr-i g 4 ut


C C
0




.1 O





CO
u











0 0
m C0
LI 4



U U

Ca. U)


0
*.4



U 0





0 0
0 0
o o
$ $
a a


40 rN. co 0' -


IN.






40







U,


S4 0
(0 .14


3 4) 0
0 J -,4
S CU U


$0 0 0 ,
S $4 U $


4) 0 Q D U


S u u <
00 4 0 u 0 0 LI
<: In 0 3 : a C) n <


























NO O
0 0'

o M




0 0M

" 0







o o

t


I.1
II7


4 u-
- 4
o O
* *





N 0'

N 0
* *


C.,




N,


0 N

o I




4 N

* *
I I






- -
* *
I I




n O
ft 0
N -

I I





o 'a




0 0
O -4








I I
*t


4O m
* *






0 N
- 0

I I




' ir

* *







- -4



- O-


4 0
C. 0

I




- 0

o 0

I I


U) 0



S 4)

0 a. 6
0 a 4 U
M U U U ( 0
0 0 ji L 0

S 40 4) z1
4 u 4 4-4
4 0 p C a a


4 4 Q 4) U 0 I '

E C 00 P W '
X 0 u 50 2 5 ( 4 U
00 ( U ( (0 U 4)( 0


fl 0 -,r CN ,
r' >.o -4 'o T




S O N N 0 c
*
I I I




N 0 c N 0
'0 M 0 C1
o 0 N 0 -
























I*
S4 -







N










*


tfl



4:
>
'-.


4)




0
U


.0

(4


C:
0
.4.

. 4-


914 a

0 0
0 0


o o
0 a


-4 4











West Campus

West Campus had the highest number of statistically

significant correlations of the three campuses with 15, out of

78, being significant at the .05 level (see Table 5).

The only statistically significant correlation related to

the dependent variable was background characteristics. The

higher the student's level of background characteristics the

more likely a student was to persist (r-.288). The remaining 14

correlations are presented according to the frequency a

particular variable achieved statistical significance. Levels

of academic integration correlated positively with levels of

background characteristics (r=.233), age (ri.339), degree of

student commitment (r=.285), peer-group interaction (r-.325),

faculty-student interaction (r=.549), and social integration

(r-.545). Peer-group interaction correlated positively with

levels of student commitment (r=.237), faculty-student

interaction (r=.254), and social integration (r=.809). Social

integration was correlated positively with levels of commitment

(r-.287) and faculty-student interaction (r-.772). Number of

accumulated hours correlated positively with levels of

background characteristics (r=.270) and students who took a

human relations type course tended generally to have more

accumulated credit hours (r-.357). Female students had a higher

level of commitment than male students (r=.275).























Int 4 O I co

N 0 -4 -
* *



0 -4 O ,.0

o 0 N
* *
I



, o >
N en-















I*
I s i














r>
in


4-
0r
0



4*

















0








Or
0)

0



0























V3
0








0
0















11










a <
E- &


c
0



0 0

jiC 4 u r --r
0 ua
j) c a
V I C C
0 C 0 1





z u 0 0

o0 4 w. -
1.4 4. 1 r 00 4
U 0 0 0 0 U
L4 u U o i




0 U ( 0 0


* 0
'. rN. c


0



00
4

0 CD0
V 4O
c U


0 o

00 O

i C
0 X U U 5
60 0 0 U 3
< N OS 4 U


C
0
,4.

a
1 1

0 u
0 U

.4 1

0 0
0 0
a a
o 0

* *

N c
-^ v-4




























I .


04 en
n
0 0





N 0'

o 0

o -





0 N

- 0

I I




N O
- me
0 O0

I I




N 0

- 0

I I





- '.
- O0

I


Um 0

U -4
)4 4J


4 41
411 U
4) m (c <
4( -4 0 H



m V U 4J (0 M 0 o
C4 (0 (0 3.4 A4 i


Q u 0 co u u 0 0

o u m 0 0 u

4En U c U.4 E n 0 0
)0 XO 1 o: 2 U 2 0
0 ( 0 U 4 0 3 3U

4JX C C.) < 0. L 0 0

-^ c M > >o r~ O O ^ '< -


'0 in c0
CO 0
- N
* *




n CIS
1 -
0 0







0 N Nd

o N 0






N CO 0

o N 0






N a csi

o O






( 0 0
04 r en
o M0
i*


On C N %D N r-
o Cl M -
- N N -N





0 0 ie, 4
- 0 -







N mn 0' N
o 4 0 N


o 4





N me, N







MC N
N M
* *








N
00
o)~ r


4C
I:


0
U
U

4)


F-
I3












As might be expected the peer-group interaction,

faculty-student interaction, and social integration variables

were all highly correlated. This is understandable since

peer-group interaction and faculty-student interaction variables

comprise the social integration variable. Only two variables

were significantly correlated with actual dropout rates. For

the total and West campuses the higher the student's background

characteristics the more likely the student would persist. East

Campus students with higher academic integration levels were

more likely to persist. The remaining statistically significant

correlations between the variables under study showed a variety

of relationships. These relationships indicated various degrees

of support for Tinto's inclusions of these variables in his

model.

Regression Analysis

Tinto's conceptual model considered in this study was

given in Figure 1 (see page 3).

Test of Hypotheses 1-4

Hypothesis 1. A significant proportion of the variation

in student dropout rates is not explained by

selected social system variables after controlling

for student background characteristics, student

commitments, and the academic system.

The test of the full model was a test of the regression

coefficients background characteristics, commitments, academic











2
system, and social system. The full model yielded an R2 of .022

with an F-statistic of .835 which was not significant at the .05

level. The test of this hypothesis was a test of the partial

regression coefficient for the social system variable given that

background characteristics, commitments, and academic system

were in the model. The results of the analysis are presented in

Table 6.

For the total campus when the dependent variable was

students' perceptions of dropout decision, the R2 with all

variables in the model was .022. Without social system in the

model it was .021. Thus adding the social system variable

explained .1% additional variation in perceived dropout. This

increase in variation resulted in a computed F-statistic of

.192, which did not exceed the critical F-statistic at the .05

level. Therefore the conclusion was to fail to reject

Hypothesis 1. The social system variable does not help predict

dropout rate.

The full model for East Campus resulted in an R of .185

with an F-statistic of 1.933, which was not significant at the

.05 level. The R2 increase and computed F-statistic for East

Campus for the partial regression coefficient were .062 and

2.569. This was not significant at the .05 level.
2
The full model for North Campus yielded an R of .029 with

a non-significant F-statistic of .253 at the .05 level. The

partial regression coefficient had an R increase of .024 and an
partial regression coefficient had an R increase of .024 and an













Table 6

2 2
R- R- Increase and Computed F Ratios for the Full Model and
Partial Regression Coefficents Relating Students' Perceptions of
Dropout Decisions With Social System Integration by Campus


Dependent Variable: Students' Perceptions of Dropout Decisions


2 2
Campus R full F R increase F partial


Total .022 .835 .001 .192

East .185 1.933 .062 2.569

North .029 .253 .024 .837

West .042 .748 .006 .448


* p<.05




Table 7

2 2
R-2 R- Increase and Computed F Ratios for the Full Model and
Partial Regression Coefficients Relating Students'Actual Dropout
Decisions With Social System Integration by Campus


Dependent Variable: Students' Actual Dropout Decisions


2 2
Campus R full F R increase F partial


Total .055 2.152 .004 .686

East .258 2.962* .058 2.651

North .041 .364 .001 .009

West .124 2.435 .034 2.704


*<-.05












F-statistic of .837. This was also non-significant at the .05

level.

2
The full model for West Campus resulted in an R of .042

with an F-statistic of .748, which was non-significant at the

.05 level. The partial regression coefficient also was

2
non-significant at the .05 level with an R increase of .006 and

an F-statistic of .448.

Table 7 contains the other half of the results to

Hypothesis 1 with students' actual dropout decisions as the

2
dependent variable. The test of the full model yielded an R2 of

.055 with an F-statistic of 2.152, which was not significant at

the .05 level. Without social system in the model, the R2 was
2
.051, which resulted in an R increase of .004. This increase

in variation resulted in a computed F-statistic of .686, which

did not exceed the critical F-statistic at the .05 level.

Therefore the conclusion was fail to reject at the .05 level for

Hypothesis 1.

2
The full model for East Campus had an R of .258 with an

F-statistic of 2.962, which was significant at the .05 level.

The R2 for the partial regression coefficient for East Campus
2
was .200, which resulted in an R increase of .058 and produced

a computed F-statistic of 2.651. This did not exceed the

critical F-statistic at the .05 level and the decision was not

to reject Hypothesis 1. Social system integration did not

improve prediction of dropping out. The full model for North











2
Campus had an R of .041 and an F-statistic of .364. West

Campus had an R2 of .124 and an F-statistic of 2.435. Both were

non-significant at the .05 level. The R2 increase for both North

Campus of .001 and West Campus of .034, with computed

F-statistics of .009 and 2.704 respectively, did not exceed the

critical F-statistic at the .05 level, therefore resulting in a

failure to reject decision for Hypothesis 1.

Hypothesis 1 was tested on four different populations,

total campus, East Campus, North Campus, and West Campus. Two

different dependent variables were used, students' perceptions

of dropout decisions and students' actual dropout rate. Of the

eight separate tests of Hypothesis 1, one was found to

statistically significant at the .05 level.

Hypothesis 2. A significant proportion of the variation

in student dropout rates is not explained by peer-group

interaction after controlling for student background

characteristics, student commitment, the academic system,

and faculty-student interaction.

The test of this hypothesis was a test of the partial

regression coefficient for peer-group interaction given that

background characteristics, commitments, academic system, and

faculty-student interaction were in the model. The results of

the analysis are presented in Table 8 with students' perceptions

of dropout decisions as the dependent variable.

2
The R with all variables in the full model, for total

campus was .035 with an F-statistic of 1.065. Without peer













Table 8

2 2
R-_ R- Increase and Computed F Ratios for the Full Model and
Partial Regression Coefficients Relating Students' Perceptions of
Dropout Decisions With Peer-Group Interaction by Campus


Dependent Variable: Students' Perceptions of Dropout Decisions


2 2
Campus R full F R increase F partial


Total .035 1.065 .009 1.420

East .158 1.181 .016 .623

North .054 .374 .049 1.551

West .050 .715 .015 1.037


*2<.05



Table 9

2 2
R-;R- Increase and Computed F Ratios for the Full Model and
Partial Regression Coefficients Relating Students' Actual Dropout
Decisions With Peer-Group Interaction by Campus


Dependent Variable: Students' Actual Dropout Decisions


2 2
Campus R full F R increase F partial


Total .060 1.875 .009 1.333

East .334 3.315* .112 5.558*

North .242 2.112 .109 4.728*

West .131 2.055 .035 2.773


*p<.05












group interaction in the model, it was .026. The peer-group

interaction variable therefore explained an additional .9

percent of the variation in dropout rate. This increase in

variation resulted in a computed F-statistic of 1.420, which did

not exceed the critical F-statistic at the .05 level. The

conclusion was to fail to reject Hypothesis 2. Peer-group

interaction did not improve prediction of dropping out.

The full models for East, North, and West campuses

resulted in R2's of .158, .054, and .050, with F-statistics of

1.181, .374, and .715. All three were non-significant at the

.05 level. The R2 increase and computed F-statistic for the

partial regression coefficients for East Campus were .016 and

.623, for North Campus .049 and 1.551, and for West Campus .015

and 1.037. In each of the three campuses the computed

F-statistic did not exceed the critical F-statistic at the .05

level. All three campuses resulted in a failure to reject

decision for Hypothesis 2.

With actual drop out as the dependent variable (Table 9),
2
the full model for total campus yielded an R of .060 with a

F-statistic of 1.875. This was not significant at the .05

level. The partial regression coefficient variable for total
2
campus had an R increase of .009. With a computed F-statistic

of 1.333, the decision was to fail to reject Hypothesis 2, since

the critical F-statistic was not exceeded at the .05 level.

East Campus full model was significant at the .05 level with an












R2 of .334 and a F-statistic Of 3.315. The partial regression

coefficient of -.199 resulted in an R2 increase of .112 and a

F-statistic of 5.558. Since this did exceed the critical

F-statistic at the .05 level the decision was to reject

Hypothesis 2. The higher the peer-group interaction score the

more likely a student would drop out.

The full model for North Campus had an R2 of .242 with an

F-statistic of 2.112 and West Campus had an R2 of .131 with an

F-statistic of 2.055. Both were non-significant at the .05

level. The partial regression coefficient of .157 for North

Campus resulted in an R2 increase of .109 and a computed

F-statistic of 4.728, which exceeded the critical F-statistic at

the .05 level. This resulted in the conclusion to reject

Hypothesis 2. The higher the peer-group interaction score the

more likely the student will remain in school. West Campus had

an R2 increase of .035 that resulted in computed F-statistics of

2.773 This did not exceed the critical F-statistic at the .05

level resulting in a fail to reject conclusion for Hypothesis 2.

Hypothesis 2 was tested on four different populations,

total, East, North, and West campuses. Two different dependent

variables were used, students' perceptions of dropout decisions

and students' actual dropout rate. Of the eight separate tests

of Hypothesis 2 only two were found to be statistically

significant. On the East and North campuses, with students'

actual dropout rate as the dependent variable, Hypothesis 2 was

found to be statistically significant.












Hypothesis 3. A significant proportion of the variation

in student dropout rate is not explained by

faculty-student interaction after controlling for student

background characteristics, student commitments, the

academic system, and peer-group interaction.

The test of the hypothesis was a test of the partial

regression coefficient for the faculty-student interaction given

that background characteristics, commitments, academic system,

and peer-group interaction were in the model. The results of

the analysis are presented in Table 10 with students'

perceptions of dropout decisions as the dependent variable.

The test of the full model with all the variables in

yielded an R2 of .035 and an F-statistic of 1.065. This was

found to be non-significant at the .05 level. The partial re-

gression coefficient for the total campus was .034. The faculty

student interaction variable therefore explained an additional

.1 percent of the variation in dropout rate. This increase in

variation resulted in a computed F-statistic of .268, which did

not exceed the critical F-statistic at the .05 level. Therefore

the conclusion was to fail to reject Hypothesis 3. Faculty

student interaction did not improve prediction of perceived

dropouts.

The full models for the East, North, and West campuses re-

sulted in R2's of .152, .054, and .050 with F-statistics of

1.181, .374, and .715. All were non-significant at the .05

level. The R2 increase and computed F-statistic for East Campus













Table 10

2 2
R- R- Increase and Computed F Ratios for the Full Model and
Partial Regrssion Coefficients Relating Students' Perceptions of
Dropout Decisions With Faculty-Student Interaction by Campus
---------------------------------

Dependent Variable: Students' Perceptions of Dropout Decisions


2 2
Campus R full F R increase F partial

Total .035 1.065 .001 .268



East .152 1.181 .025 .978



North .054 .374 .006 .192



West .050 .715 .001 .031


*E<.05



Table 11

2 2
R-, R- Increase and Computed F ratios for the Full Model and
Partial Regresson Coefficients Relating Students' Actual Dropout
Decisions With Faculty-Student Interaction by Campus



Dependent Variable: Students' Actual Dropout Decisions


2 2
Campus R F R increase F partial



Total .063 1.875 .001 .002

East .334 3.305* .086 4.453*



North .242 2.112 .167 7.284*

West .131 2.055 .005 .356


*7<.05