USING ORDINAL REGRESSION MODELING TO

EVALUATE THE SATISFACTION OF JOMO KENYATTA

UNIVERSITY OF AGRICULTURE AND TECHNOLOGY
FACULTY OF SCIENCE STUDENTS

OMBUI G. MONARI

A RESEARCH PROJECT REPORT SUBMITTED TO THE

STATISTICS AND ACTUARIAL SCIENCES
DEPARTMENT IN PARTIAL FULFILLMENT OF THE
DEGREE OF POSTGRADUATE DIPLOMA IN APPLIED
STATISTICS

APRIL, 2010
 Declaration
I hereby declare that this is my original work and has not been presented
anywhere in any other university or institution either in whole or in part for
award of any degree, fellowship or any other similar title whatsoever.

Ombui M. Geofrey

Signature.................. Date ...............

This project report has been submitted for examination with my approval
as the supervisor.

Dr.Gichuhi A. Waititu

Signature.................. Date ...............

i
 ACKNOWLEDGMENTS

Thanks to Dr. Gichuhi A. Waititu who has been my supervisor and mentor
throughout my lifetime in the JKUAT Actuarial and Statistics department.
He has encouraged me, and had faith in me; and has been available at any
time that I needed his advice and knowledge. I could not have hoped for
a more supportive supervisor. I would also like to extend my thanks to all
the members of the JKUAT Actuarial and Statistics department who have
all helped and encouraged me during my time at JKUAT University in the
department.

I would also like to thank Dr. S. M. Mwalili of JKUAT Actuarial and

Statistics department.He has provided me with amazing support, encourage-
ment and numerous discussions about the issues discussed in this project.

Finally, a very special thank you to my parents, brothers and sisters who
have supported me throughout my academic studies from day one at school.
I cannot thank them enough for the support, love and encouragement and
financial assistance for this project.

ii
 Abstract

General students’ satisfaction of Jomo Kenyatta University of Agriculture

and Technology (J.K.U.A.T) faculty of science students is associated with
a combination of predictor variables that are a mixture of qualitative and
quantitative categories. Ordinal regression statistical technique was used to
model the relationship between the academic programs, facilities and ser-
vices and the outcome variable to determine the explanatory variables that
influence students’ satisfaction factors that will assist us advice the faculty
administrators on future service delivery improvements. Data description
was done using the frequency tables and interactive graphs while data analy-
sis was done using fitting statistics; information fitting that checks the pres-
ence of a relationship between the dependent variable and combination of
independent variables, goodness of fit that gives the information about how
many predicted cell frequencies differ from the observed frequencies, param-
eter of estimates that determines the factors that influence satisfaction and
the test of parallel lines assumption that makes judgment concerning the
model adequacy. The factors that we found to influence the satisfaction of
the J.K.U.A.T faculty of science were four, namely, Service delivery at the
department office, the library services, accommodation facilities inside the
university hostels and the accommodation facilities outside the university.

iii
Contents

1 INTRODUCTION AND LITERATURE REVIEW 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Statement of the problem . . . . . . . . . . . . . . . . . . . . 4
1.4 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.6 Justification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 MATERIALS AND METHOD 6

2.1 The model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.2 Definition of terms and concepts . . . . . . . . . . . . . 6
2.1.3 Ordinal Regression Model . . . . . . . . . . . . . . . . 7
2.2 Link function . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 The Complementary Log-log link function. . . . . . . . 9
2.3 The assumptions . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Data collection . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.5 Data questionnaire . . . . . . . . . . . . . . . . . . . . . . . . 11
2.6 Data sample and data analysis software package . . . . . . . . 11

3 RESULTS 12
3.1 Data description . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4 CONCLUSION AND RECOMMENDATIONS 51

4.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.2 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . 53

A Appendix I 56

iv
List of Figures

3.1 Year of study . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.2 Program of study . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.3 market demand . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.4 service delivery at the faculty office . . . . . . . . . . . . . . . 19
3.5 relationship between students and faculty sub-ordinate staff . 21
3.6 course promotion . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.7 computer skills . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.8 admission and registration process . . . . . . . . . . . . . . . . 27
3.9 library services . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.10 career counseling services . . . . . . . . . . . . . . . . . . . . . 31
3.11 JKUAT hospital facilities . . . . . . . . . . . . . . . . . . . . . 33
3.12 accommodation facilities inside JKUAT . . . . . . . . . . . . . 35
3.13 internet facilities in JKUAT . . . . . . . . . . . . . . . . . . . 37
3.14 inter-departmental sports events . . . . . . . . . . . . . . . . . 39

v
List of Tables

2.1 The five link functions . . . . . . . . . . . . . . . . . . . . . . 9

3.1 Departments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2 Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.3 Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.4 The lecturers’ service delivery . . . . . . . . . . . . . . . . . . 18
3.5 The Service delivery at the department office . . . . . . . . . . 20
3.6 The accessibility of the faculty office . . . . . . . . . . . . . . 22
3.7 The communication skills gained from your course . . . . . . . 24
3.8 The research skills gained from your course . . . . . . . . . . 26
3.9 The financial ability to pay school fees and meet personal needs 28
3.10 The tutorial services offered in your course . . . . . . . . . . . 30
3.11 The classroom facilities . . . . . . . . . . . . . . . . . . . . . . 32
3.12 The laboratory facilities . . . . . . . . . . . . . . . . . . . . . 34
3.13 The accommodation facilities outside JKUAT . . . . . . . . . 36
3.14 The student center faculties in JKUAT . . . . . . . . . . . . . 38
3.15 The general faculty of science service delivery . . . . . . . . . 40
3.16 Observed distribution of general question, participation by de-
partment, gender and program of study . . . . . . . . . . . . . 41
3.17 Model Fitting Information . . . . . . . . . . . . . . . . . . . . 42
3.18 goodness-of-Fit . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.19 Pseudo R-Square . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.20 Parameter estimates for Clog-log logistic regression . . . . . . 48
3.21 Test of parallel lines . . . . . . . . . . . . . . . . . . . . . . . 50

vi
Chapter 1

INTRODUCTION AND
LITERATURE REVIEW

1.1 Introduction
Students have always been investigated every end of semester using a twenty
seven questions questionnaire with gender, year of study, unit code, lec-
turer’s main name and programs of study measured using a nominal scale
while course objectives given at the beginning of the course, description of
course outline, appropriateness of course objectives, relevance of given refer-
ence materials,course coverage/completion, attendance of all schedule classes,
lecturer punctuality, lecturer duration, lecturer delivery as per the content,
use of examples and illustrations, use of teaching aids, communication skills,
notes/materials/handouts given, presentation sequence, students participa-
tion, lecturer’s motivation of students, availability of lecturer for consulta-
tion, lecturer’s mastery of content, administration of CAT’s and assignments
as scheduled, feedback on CATs and assignments(revision of cats)and lastly
relevance of CATs and assignments/labwork in relation to course outline is
measured using a five point ordinal scale to rate lecturers’ per unit as the
students have needs and rights to participate in quality and satisfactory ser-
vices. The survey reflects key issues as perceived by the Faculty of Science
administrators to timely plan for quality services majorly from lecturers to
students.

This research study was used to analyzed the students satisfaction factors
using ordinal regression statistical technique. Frequency tables and interac-
tive graphs were applied to detect the satisfaction factors regarding academic
programs, facilities and services. Ordinal regression method was useful in an-

1
alyzing the relationship between multiple explanatory variables and outcome
variable.

In the study we investigated the following factors in relation to students

satisfaction : department, year of study, gender and Program of study were
measured using a nominal scale while age was measured using scale and the
independent variables course satisfaction in reference to the market demand,
lecturers’ service delivery, Service delivery at the faculty office, Service deliv-
ery at the department office, Relationship between students and the faculty
sub-ordinate, accessibility of the faculty office, course promotion by the fac-
ulty, communication skills gained from your course, computer skills gained
from your course, research skills gained from your course, Admission and
registration process in reference to time, Financial ability to pay school fees
and meet your needs, library services, tutorial services in your course, career
counseling services in the faculty, lecture room facilities, JKUAT hospital
facilities, course laboratory facilities, accommodation facilities in JKUAT,
accommodation facilities outside JKUAT, JKUAT internet facilities, stu-
dent center facilities, faculty inter-departmental sports events and lastly the
dependent variable generally on faculty of science service delivery that were
measured using a five scale ordinal scale in Statistical Package for the Social
Sciences (SPSS). The statistical regression technique was used to model the
satisfaction of sampled students from the Faculty of Science using ordinal
regression to identify explanatory variables related to academic programs,
facilities and services that contribute to the overall Students satisfaction.

1.2 Literature Review

Students end of semester questionnaires evaluate the students’ satisfaction
on faculty of science programs only from the lecturers’ delivery point of view.
To evaluate general students’ satisfaction factors and determine the students
satisfaction ratings on factors that influence their satisfaction in campus us-
ing ordinal regression model.

To obtain various satisfaction ratings, different statistical methods such

as descriptive statistics and ordinal regression techniques were used to ana-
lyze student satisfaction questionnaires. Descriptive statistics has extensively
been used to detect the satisfactory items that students have experienced
from their programs, facilities and services.

For instance, the mean responses of student satisfaction survey conducted

2
by Noel-Levitz Company revealed community college student satisfaction.
The survey respondents rated highest satisfaction on responsiveness to di-
verse populations, registration effectiveness, and academic services, while
rating the lowest satisfaction on admissions and financial aid, academic ad-
vising, and campus support services, [3].

Using percentages, means, modes, and qualitative written reports, stu-

dent satisfaction with the quality curriculum content, faculty involvement,
support service, facilities and recreation was rated using a five point ordinal
scale of ’very satisfactory’, ’satisfactory’, ’average ’, ’ unsatisfactory’ , ’very
unsatisfactory’. The rating depicted factors that influenced satisfaction, [4]
, [10].

Chi-square and linear regression techniques has been utilized to deter-

mine the association between the explanatory variables and the performance.
Cross tabulation and chi-square techniques were used,[2] to predict college
student retention based on satisfaction. A strong relationship between stu-
dent satisfaction and retention found on 40 of the 68 questions was 59%.

Using linear regression and decision tree analysis with the chi-squared
automatic interaction detector (CHAID) software program, a study by [8]
compared student satisfaction responses between academically and non aca-
demically oriented student groups. The research results demonstrated that
faculty preparedness, social integration, and pre-enrollment opinions emerged
as the most important variables contributing to student satisfaction for both
groups.

Linear regression methods were used to investigate the relationship be-

tween student satisfaction and medical school learning environment, [7]. The
study results provided evidence that curriculum structures, (e.g., timely feed-
back and promotion of critical thinking) were prominent explanatory vari-
ables.

Using a multilevel modeling technique to analyze survey data, a study

by [9] examined the impact that different departments have on student sat-
isfaction in a large research university. The research finding revealed that
characteristics of departments such as size, faculty contact with students, re-
search emphasis, and proportion of female students had a significant impact
on education satisfaction within major.

Utilizing an ordinal regression model, a newly implemented study,[5] was

3
used to estimating the probabilities of the four ordinal categories (”worse”,
”can’t tell”, ”better”, and ”much better”) of client improvement in a counsel-
ing center. The research findings showed that the five explanatory variables
significantly associated with the probability of an outcome category. These
variables included previous experience as a client; readiness to change; level
of symptomatic and interpersonal distress; pre-counseling clinical status; and
the number of counseling sessions in which a client might be involved.

Based on the literature review, one might conclude that descriptive statis-
tics (e.g., means, percentages, and frequency counts), chi-square (e.g., cross-
tabulation, Pearson’s chi-square test, decision tree with CHAIDS software
program), linear regression, and multilevel modeling approaches were increas-
ingly utilized to study student satisfaction in relation to various explanatory
variables. However, compared to these study , the ordinal regression method
seems to be the most suitable and practical techniques to analyze the effects
of multiple explanatory variables on the ordinal outcome that cannot be as-
sumed as continuous measure and normal distribution.

Researchers do not need to alter an ordinal outcome as binary or dichoto-

mous measure for logistic regression analysis, which may lead to the loss of
inherent information. The ordinal regression analysis is currently underused
in the field of education, several articles were found in the medical field,
which illustrated the foundation of the mathematical model.

1.3 Statement of the problem

The challenge facing the administrators is managing students’ satisfaction.

1.4 Objectives
General objective
To investigate which explanatory factors influence students’ satisfaction.
Specific objectives

1. To model the satisfaction of sampled students from the Faculty of Sci-

ence using ordinal regression.

2. To provide timely advice to administrators hence ensuring faculty con-

tinuity and timely development and modification of student facilities,
programs and receive quality services from the staff.

4
1.5 Hypothesis
Student satisfaction is influenced by the explanatory variables: department,
year of study, gender, Program of study, age, course satisfaction in refer-
ence to the market demand, lecturers’ service delivery, and Service delivery
at the faculty office, Service delivery at the department office, Relationship
between students and the faculty sub-ordinate, accessibility of the faculty
office, course promotion by the faculty, communication skills gained from
your course, computer skills gained from your course, research skills gained
from your course, Admission and registration process in reference to time,
Financial ability to pay school fees and meet your needs, library services,
tutorial services in your course, career counseling services in the faculty,
lecture room facilities, JKUAT hospital facilities, course laboratory facili-
ties, accommodation , facilities in JKUAT, accommodation facilities outside
JKUAT, JKUAT internet facilities, student center facilities and faculty inter-
departmental sports events.

1.6 Justification
End of semester questionnaires have always measured students solely on the
students’ satisfaction from the lecturer point of view. Thus the need to model
the general students’ satisfaction using ordinal regression.

5
Chapter 2

MATERIALS AND METHOD

2.1 The model

2.1.1 Introduction
Ordinal regression technique was used to model the behavior of dependent
variable with a set of independent variables. In ordinal regression, the de-
pendent variable is the order response category variable and the independent
variable was a mixture of qualitative and quantitative variables.

2.1.2 Definition of terms and concepts

Dependent variable: The dependent variable is ordinal. The first category
was considered as the lowest category and the last category was considered
as the highest category.
Covariate: Covariates are continuous independent variables e.g. age.
Factor: Factor is a categorically independent variable that must be coded
as numeric in SPSS e.g. department.
Complimentary Log-log (Clog-log)link function: clog-log link function
was used to predict the dependent variable category. Clog-log link function
was used in SPSS statistical package for ordinal regression modeling because
the data that was gathered for analysis had dependent ordinal variable with
equal category. The clog-log link function is of the form

f (X) = log (−log (1 − X)) (2.1)

Since the ordinal clog-log model is non linear, transformation was done on
the dependent variable, to make the model linear. Clog-log link function is
natural log of the odds and is good at linearizing the model. The technique

6
uses maximum likelihood, thus more cases than a similar Ordinary Least
Squares model (OLS) was required. The odds ratio was generated by the
clog-log, then probabilities were predicted from the model.

2.1.3 Ordinal Regression Model

Generalized linear model is a very powerful class of model, which can be
used to provide solutions to a wide range of statistical questions. The basic
form of a generalized linear model is shown in the following equation.
link (γij ) = θj − [β1 Xi1 + β2 Xi2 + ......... + βK XiK ] (2.2)
where, link() -is the link function
γij -is the cumulative probability for the jth category for the ith case
θj - is the threshold for the jth category
k - is the number of regression coefficients
β1 ....βk - are the regression coefficients
Xi1 ....XiK are values of the predictors for the ith case
The important things to note on the generalized linear model are:
• The model is based on the notion that there is some latent continuous
outcome variable, and that the ordinal outcome variable arises from
discretizing the underlying continuum into ordered groups. The cutoff
values that define the categories are estimated by the thresholds. In
some cases, there is good theoretical justification for assuming such an
underlying distribution. However, even in cases in which there is no
theoretical concept that links to the latent variable, the model can still
perform quite well and give valid results.
• The thresholds or constants in the model (corresponding to the in-
tercept in linear regression models) depend only on which category’s
probability is being predicted. Values of the predictor (independent)
variables do not affect this part of the model.
• The prediction part of the model depends only on the predictors and
is independent of the outcome category. These first two properties
imply that the results were a set of parallel lines or planes-one for each
category of the outcome variable.
• Rather than predicting the actual cumulative probabilities, the model
predicts a function of those values. This function is called the link
function, and you choose the form of the link function when you build
the model. This allows you to choose a link function based on the
problem under consideration to optimize your results.

7
Three major components in ordinal regression model are:
Location component. The portion of the equation 2.2 includes the coeffi-
cients and predictor variables, is called the location component of the model.
The location is the ”meat” of the model. It uses the predictor variables
to calculate predicted probabilities of membership in the categories for each
case.
Scale component.The scale component is an optional modification to the
basic model to account for differences in variability for different values of the
predictor variables. For example, if certain groups have more variability than
others in their ratings, using a scale component to account for this improved
the model. The model with a scale component follows the form shown in 2.3.

θj − [β1 X1 + β2 X2 + ......... + βK XK ]
link (γj ) = (2.3)
exp (τ1 z1 + τ2 z2 + .......... + τm zm )
where,
τ1 ...τm are coefficients for the scale component
z1 ....zm are m predictor variables for the scale component.

2.2 Link function

The link function is a transformation of the cumulative probabilities that
allows estimation of the model.It defines what goes to the left side of the
equation. It’s the link between the random component on the left side of the
equation and the systematic component on the right.

Five link functions are available in the Ordinal Regression procedure. The
following are the link functions , form and typical application.

8
 Function Form
  Typical application
γ
Logit log 1−γ evenly distributed categories
CLog-log log (−log (1 − γ)) higher categories more probable
Negative log-log −log (−log (γ)) lower categories more probable
probit φ−1 (γ) latent variable is normally distributed
Cauchit (inverse cauchy) tan (π (γ − 0.5)) latent variable has many extreme values

Table 2.1: The five link functions

In ordinal regression analysis, we used the Clog-log link function to build

our model that is generally suitable for analyzing the ordered categorical
data with higher categories more probable among all categories.

2.2.1 The Complementary Log-log link function.

The Clog-log link function in ordinal regression modeling was used in ana-
lyzing the data is written in the form.

f (X) = log (−log (1 − X)) (2.4)

It is not a typo that there is a minus sign before the coefficients for the
predictor variables, instead of the customary plus sign.
Each Clog-log has its own θj term but the same coefficient β. That means
that the effect of the independent variable is the same for different Clog-log
functions. That’s an assumption that one has to check. That’s also the
reason the model is also called the proportional odds model. The θj terms,
called the threshold values, often are not of much interest. Their values do
not depend on the values of the independent variable for a particular case.
They are like the intercept in a linear regression, except that each Clog-log
has its own. They are used in the calculations of predicted values.

The coefficients in the Ordinal regression model depicts how much the
Clog-log changes based on the values of the predictor variables. Parameter
estimates from the output of SPSS computation where a table called ’param-
eter estimates’ appears were analyzed.

parameter estimates table: a variable named location variable gave

the coefficient for the independent variable for the specified link function in
ordinal regression.

9
 Factor Summary: Factor summary depicts that the general question
ordinal scale distribution in percentage on respondents.

Model Fitting Information: checks the presence of a relationship be-

tween the dependent variable and combination of independent variables was
based on the statistical significance of the final model.

Goodness of fit information: Pearson chi-square test that gives the

information about how many predicted cell frequencies differ from observed
frequencies.

R-square estimate: One cannot use simple r-square in ordinal regres-

sion when modeling categorical data.

Test of Parallel Lines: Test of parallel lines was designed to make

judgment concerning the model adequacy.

2.3 The assumptions

1. parallel lines. One of the assumptions underlying ordinal regression is
that the relationship between each pair of outcome groups is the same.
In other words, ordinal regression assumes that the coefficients that
describe the relationship between, say, the lowest versus all higher cat-
egories of the response variable are the same as those that describe the
relationship between the next lowest category and all higher categories,
etc. This is called the proportional odds assumption or the parallel
regression assumption. Because the relationship between all pairs of
groups is the same, there is only one set of coefficients. Thus, in order
to asses the appropriateness of our model, we evaluated whether the
proportional odds assumption is tenable. Statistical tests are available
in Statistical Package for the Social Sciences (SPSS) software version
12.0. However, these tests have been criticized for having a tendency
to reject the null hypothesis (that the sets of coefficients are the same),
and hence, indicate that there the parallel slopes assumption does not
hold, in cases where the assumption does hold, [6].

2. Adequate cell count: As per the rule of thumb, 80 % of cells must

have more than 5 counts. No cell should have Zero count as it is
considered as a missing value and excluded from the study. the large
percentage of cells with missing data could lead to a decrease of actual

10
 sample size from the model construction or an inaccurate Chi-square
test for the model fitting, since the model goodness-of-fit is usually de-
pendent of chi-square test result. The chi-square test normally depends
on the sample size. Hence, if the number of cells with a zero value is
large, the chi-square goodness of fit statistics may not be appropriate
[1].

2.4 Data collection

The random sample size was one hundred and twenty five students from the
faculty of science from each of the eight departments in the faculty of science
at equal capacity. The sample of the students was not Gender bias and
from any year of study however, more fourth years were sampled from either
program of study i.e Government Sponsored (GOK) or Alternative Degree
Program (ADP). Gender, department, year of study and program of study
were measure using a nominal scale as follows Department 1 = ’Physics’,
2 = ’ Medical Microbiology ’, 3 = ’Zoology’, 4 = ’Statistics and Actuarial
Sciences’, 5 = ’Pure and Applied Mathematics’, 6 = ’Biochemistry’, 7 =
’Chemistry’ and 8 = ’Botany’, Year of study 1 = ’First year’, 2 = ’Second
year’, 3 = ’Third year’ and 4 = ’Fourth year’, Gender 0 = ’Female’ and 1
= ’Male’, Program of study 0 = ’GOK’ and 1 = ’ADP’ and Age using scale
while a five point ordinal scale was used in rating the students satisfaction
levels with 5 = ’Very satisfactory’, 4 = ’satisfactory’, 3 = ’Average’, 2 =
’unsatisfactory’ and 1 = ’Very satisfactory’ for the rest of the explanatory
variables in SPSS

2.5 Data questionnaire

The questionnaire that was used to collect the data is attached in the ap-
pendix.

2.6 Data sample and data analysis software

package
The data sample was one hundred and twenty five students from the faculty
of science and Statistical Package for the Social Sciences (SPSS) software
version 12.0 was used to analyze the data.

11
Chapter 3

RESULTS

3.1 Data description

Department
The data was collected from the eight faculty of science departments and the
respondents per department were as follows; fifteen from the department of
Physics, fourteen from the department of Medical Microbiology, fifteen from
the department of Zoology, seventeen from the department of Statistics and
Actuarial sciences, twenty one from the department of Pure and Applied
Mathematics, sixteen from the department of Biochemistry, thirteen from
the department of Chemistry and finally fourteen from the department of
Botany. The questionnaires were administered at random thus the difference
in proportions per department.

Table 3.1: Departments

Department Count Percentage(%)

Physics 15 12.0%
Medical Microbiology 14 11.2%
Zoology 15 12.0%
Statistics and Actuarial Sciences 17 13.6%
Pure and Applied Mathematics 21 16.8%
Biochemistry 16 12.8%
Chemistry 13 10.4%
Botany 14 11.2%
Total 125 100.0%

12
 Figure 3.1: Year of study

The respondents’ counts from different years of study were as follows;

53.6% were fourth years, 39.2% was distributed between the second and
third years while first years were 7.2%. Majority of the respondents were
fourth year students as they were thought to have more experience on the
university operations, compared to the other years’ students like first years
who may have not even known where their departments offices.

13
 Gender
The frequency of the of the interviewees with respect to their gender were
forty nine and seventy six female and male respectively i.e. 39.2% and 60.8%
female and male respectively. There are more male to female respondents
that reflect the sex ratio of female to male in the faculty.

Table 3.2: Gender

Gender Count Percentage(%)
Female 49 39.2%
Male 76 60.8%
Total 125 100.0%

14
 Figure 3.2: Program of study

Program of study
The interviewees were from the Self Sponsored students (ADP) and the
Government of Kenya sponsored students (GOK) programs. There were
41.6% interviewees from GOK and 58.4% from the ADP program of study.
The ADP respondents are more than the GOK respondents because some
departments like Statistics and Actuarial Sciences many ADP students.

15
 Age
The respondents were spread in age from seventeen to thirty three years of
age as follows; twenty one to twenty four years of age had 78.4% respon-
dents, seventeen to twenty years of age and 11.2% repondents and 10%of the
repondents were between twenty five years of age thirty three years of age.

Table 3.3: Age

Respondents’ Age Count Percentage(%)

17 1 0.8%
19 3 2.4%
20 10 8.0%
21 20 16.0%
22 24 19.2%
23 34 27.2%
24 20 16.0%
25 6 4.8%
26 1 0.8%
27 2 1.6%
28 1 0.8%
29 1 0 .8%
30 1 0.8%
33 1 0.8%
Total 125 100.0%

16
 Figure 3.3: market demand

The course satisfaction in reference to the market demand was as follows;

80.8% rated average and satisfactory, 12% were either very unsatisfactory or
unsatisfactory and 7.2% rated very satisfactory.

17
 Lecturers’ service
The lecturer’s service delivery had its frequency as follows; average had high-
est count at sixty one, followed by satisfactory at a frequency of fifty three,
then unsatisfactory at a count of seven and finally very satisfactory and very
unsatisfactory at a count of two respectively.

Table 3.4: The lecturers’ service delivery

Lecturers’ service Count Percentage(%)

delivery categories
very unsatisfactory 2 1.6%
Unsatisfactory 7 5.6%
Average 61 48.8%
Satisfactory 53 42.4%
very satisfactory 2 1.6%
Total 125 100.0%

18
 Figure 3.4: service delivery at the faculty office

The count of service delivery at the faculty office was as follows; average
had highest count at fifty one, followed by satisfactory at a frequency of thirty
three, then unsatisfactory at a count of twenty three, very unsatisfactory at
a count of thirteen and finally very satisfactory at a count of five respondents
respectively.

19
 Service delivery at the department
The service delivery at the department office count was as follows; satisfac-
tory had highest count at forty eight, followed by average at a frequency
of thirty eight, then unsatisfactory at a count of nineteen, very satisfactory
at a frequency of fifteen and finally very unsatisfactory at a count of five
respondents.

Table 3.5: The Service delivery at the department office

Service delivery at the Count Percentage(%)

department categories
very unsatisfactory 5 4.0%
Unsatisfactory 19 15.2%
Average 38 30.4%
Satisfactory 48 38.4%
very satisfactory 15 12.0%
Total 125 100.0%

20
 Figure 3.5: relationship between students and faculty sub-ordinate staff

The count of the relationship between students and faculty sub-ordinate

staff was recorded as follows; average had highest frequency at fifty six, fol-
lowed by satisfactory at a frequency of thirty four, then unsatisfactory at a
frequency of twenty one, very unsatisfactory at a count of nine and finally
very satisfactory at a count of five respondents respectively.

21
 Faculty accessibility
The accessibility of the faculty office frequency was as follows; average had
highest count at fifty six, followed by satisfactory at a frequency of thirty
seven, then unsatisfactory at a count of fifteen, very satisfactory at a fre-
quency of nine and finally very unsatisfactory at a count of eight respondents
respectively.

Table 3.6: The accessibility of the faculty office

Faculty accessibility Count Percentage(%)

categories
very unsatisfactory 8 6.4%
Unsatisfactory 15 12.0%
Average 56 44.8%
Satisfactory 37 29.6%
very satisfactory 9 7.2%
Total 125 100.0%

22
 Figure 3.6: course promotion

The course promotion by the faculty was rated by the interviewees with
highest frequency at average of fifty eight, unsatisfactory at a frequency of
twenty seven, satisfactory at a count of twenty, then very satisfactory at
eleven and lastly very unsatisfactory at nine.

23
 Communication skills
The communication skills gained from the respective respondents’ course was
rated with highest frequency at satisfactory of forty eight, average at the
frequency of forty five respondents, unsatisfactory at a frequency of twenty
three, very unsatisfactory at a count of six and finally very satisfactory at
three.

Table 3.7: The communication skills gained from your course

Communication skills Count Percentage(%)

gained categories
very unsatisfactory 6 4.8%
Unsatisfactory 23 18.4%
Average 45 36.0%
Satisfactory 48 38.4%
very satisfactory 3 2.4%
Total 125 100.0%

24
 Figure 3.7: computer skills

The counts on the computer skills gained from the respective courses of
the interviewees was rated with highest count of respondents rating average
at forty four, very unsatisfactory was second with thirty respondents, unsat-
isfactory third with twenty six respondents, satisfactory fourth with twenty
three respondents and finally very unsatisfactory with only two respondents.

25
 Research skills
The research skills gained from the respondents’ course was rated as follows
with satisfactory being rated highest with a frequency of forty interviewees,
thirty eight average, twenty one very unsatisfactory, nineteen unsatisfactory
and lastly seven very satisfactory.

Table 3.8: The research skills gained from your course

Research skills Count Percentage(%)

gained categories
very unsatisfactory 21 16.8%
Unsatisfactory 19 15.2%
Average 38 30.4%
Satisfactory 40 32.0%
very satisfactory 7 5.6%
Total 125 100.0%

26
 Figure 3.8: admission and registration process

The counts on the admission and registration process in reference to time

was rated with highest frequency of respondents rating average at fifty three,
unsatisfactory was second with thirty six respondents, then satisfactory was
third with twenty one respondents, very unsatisfactory fourth with eleven
respondents and lastly very satisfactory with only four respondents.

27
 Financial ability to pay school fees and meet personal needs
The financial ability to pay school fees and meet personal needs by the re-
spective respondents was rated with highest frequency of respondents rating
average at forty three, unsatisfactory was second at thirty, then satisfactory
was third with twenty six respondents, very unsatisfactory fourth with twenty
three respondents and finally very satisfactory with just three respondents.

Table 3.9: The financial ability to pay school fees and meet personal needs

Financial ability to pay Count Percentage(%)

school fees and meet
personal needs categories
very unsatisfactory 23 18.4%
Unsatisfactory 30 24.0%
Average 43 34.4%
Satisfactory 26 20.8%
very satisfactory 3 2.4%
Total 125 100.0%

28
 Figure 3.9: library services

The respondents frequencies on the rates of the library services was as

follows; highest frequency of respondents rated average at fifty, unsatisfactory
was second at thirty, then followed closely by satisfactory at twenty eight
respondents, very unsatisfactory fourth with fifteen respondents and finally
very satisfactory with only two respondents.

29
 Tutorial services
The respondents’ frequencies on the rates of the tutorial services with respect
to their respective courses average had highest frequency of respondents at
forty seven, unsatisfactory was second at thirty nine, then followed by very
unsatisfactory at twenty two interviewees, satisfactory was rated fourth with
thirteen respondents and lastly very satisfactory with four respondents.

Table 3.10: The tutorial services offered in your course

Tutorial services Count Percentage(%)

categories
very unsatisfactory 22 17.6%
unsatisfactory 39 31.2%
average 47 37.6%
satisfactory 13 10.4%
very satisfactory 4 3.2%
Total 125 100.0%

30
 Figure 3.10: career counseling services

The respondents rating frequencies on the career counseling services in

the faculty with respect to their courses had highest frequency of respondents
at average with thirty nine, unsatisfactory recorded a frequency of thirty
four, then followed closely by very unsatisfactory at thirty one respondents,
satisfactory fourteen respondents and finally very satisfactory with seven
respondents.

31
 Classroom facilities
The respondents rates on classroom facilities had frequencies as follows; high-
est frequency of respondents rated average at forty eight , unsatisfactory
second with a frequency of thirty seven, then followed by satisfactory at
twenty four respondents, very unsatisfactory ten respondents and lastly very
satisfactory with six respondents.

Table 3.11: The classroom facilities

Classroom facilities Count Percentage(%)

categories
very unsatisfactory 10 8.0%
Unsatisfactory 37 29.6%
Average 48 38.4%
Satisfactory 24 19.2%
very satisfactory 6 4.8%
Total 125 100.0%

32
 Figure 3.11: JKUAT hospital facilities

The JKUAT hospital facilities rates by respondents were as follows; high-

est count of respondents were average at sixty one, unsatisfactory at twenty
three, satisfactory eighteen respondents, then followed closely by very unsat-
isfactory at sixteen, and finally very satisfactory with seven respondents.

33
 Laboratory facilities
The JKUAT laboratory facilities with respect to the respondent courses rates
were as follows; highest count of respondents were unsatisfactory at forty
four, followed by average at thirty eight, satisfactory twenty two respondents,
then very unsatisfactory at sixteen, and finally very satisfactory with five
respondents.

Table 3.12: The laboratory facilities

Laboratory facilities Count Percentage(%)

categories
very unsatisfactory 16 12.8%
Unsatisfactory 44 35.2%
Average 38 30.4%
Satisfactory 22 17.6%
very satisfactory 5 4.0%
Total 125 100.0%

34
 Figure 3.12: accommodation facilities inside JKUAT

The accommodation facilities inside JKUAT rates by GOK students were

as follows; highest frequency of respondents was twenty six at average, fol-
lowed by satisfactory at twelve, unsatisfactory eight interviewees, then very
satisfactory at five, very unsatisfactory with one respondent and finally not
applicable as rated by the ADP students was at seventy three.

35
 Accommodation outside JKUAT
The accommodation facilities outside JKUAT rates as rated by ADP students
were as follows; highest frequency of respondents were average at thirty six,
twelve unsatisfactory, followed closely by ten interviewees satisfactory and
nine interviewees very unsatisfactory, then very satisfactory with six respon-
dents and finally fifty two respondents not applicable that as rated by the
GOK students.

Table 3.13: The accommodation facilities outside JKUAT

Accommodation outside Count Percentage(%)

JKUAT categories
very unsatisfactory 9 7.2%
Unsatisfactory 12 9.6%
Average 36 28.8%
Satisfactory 10 8.0%
very satisfactory 6 4.8%
not applicable 52 41.6%
Total 125 100.0%

36
 Figure 3.13: internet facilities in JKUAT

The internet facilities in JKUAT were rated by the interviewees as follows

with highest frequency of respondents unsatisfactory with counts at forty five,
followed by thirty seven very unsatisfactory respondents, thirty two respon-
dents’ average, eight interviewees satisfactory and lastly three respondents
very satisfactory.

37
 Student center
The student center facilities were rated by the respondents as follows with
highest frequency of respondents unsatisfactory with frequency of fifty three,
followed by forty two respondents very unsatisfactory, twenty one intervie-
wees average, seven respondents satisfactory and finally two respondents very
satisfactory.

Table 3.14: The student center faculties in JKUAT

Student center Count Percentage(%)

categories
very unsatisfactory 42 33.6%
Unsatisfactory 53 42.4%
Average 21 16.8%
Satisfactory 7 5.6%
very satisfactory 2 1.6%
Total 125 100.0%

38
 Figure 3.14: inter-departmental sports events

The frequencies for the faculty inter-departmental sports events were

rated average at fifty, followed by thirty seven interviewees unsatisfactory,
twenty respondents very unsatisfactory, then fifteen respondents rated satis-
factory and finally three interviewees rated it very satisfactory.

39
 General faculty of science service delivery
The counts generally on the faculty of science service delivery were rated
by the interviewees were sixty two respondents average, twenty seven re-
spondents satisfactory, twenty four interviewees unsatisfactory, followed by
seven respondents very unsatisfactory and eventually five respondents very
satisfactory.

Table 3.15: The general faculty of science service delivery

General faculty of science Count Percentage(%)

service delivery categories
very unsatisfactory 7 5.6%
Unsatisfactory 24 19.2%
Average 62 49.6%
Satisfactory 27 21.6%
very satisfactory 5 4.0%
Total 125 100.0%

3.2 Data analysis

Factors’ Summary
This implies that over 75% of the students are above averagely satisfied of
these 25% are either satisfactory or very satisfactory, but 25% are below av-
erage satisfaction.Overall all departments participated in equal capacity.

The ratio of female to male respondents was approximately 39% to 61%.

The big difference depicts the general female to male ratio in the university.

The factor program of study had Government of Kenya (GOK) spon-

sored students at 41.6% and Alternative Degree Program (ADP) students at
58.4% the difference in respondents in percentages was because some of the
departments like Statistics and Actuarial Sciences have more ADP students.

40
Table 3.16: Observed distribution of general question, participation by de-
partment, gender and program of study

Variables n Percentages
General question Very unsatisfactory 7 5.6%
unsatisfactory 24 19.2%
average 62 49.6%
satisfactory 27 21.6%
Very satisfactory 5 4.0%
Department Physics 15 12.0 %
Medical Micro-Biology 14 11.2%
Zoology 15 12.0 %
Statistics and Actuarial Sciences 17 13.6%
Pure and Applied Mathematics 21 16.8%
Biochemistry 16 12.8%
Chemistry 13 10.4%
Botany 14 11.2%
Gender female 49 39.2%
Male 76 60.8%
Program of study GOK 52 41.6%
ADP 73 58.4%
Valid 125 100.0%
Missing 0
Total 125

Missing data variables were truncated hence the model has no missing
variables.

Model Fitting Information

The results from model fitting in the section provide results of ordinal logistic
regression versus reduced model(intercept) with complimentary log-log link
function. The presence of a relationship between the dependent variable and
combination of independent variables is based on the statistical significance
of the final model. From Table 1.17, the -2LL of the model with only inter-
cept is 321.455 while the -2LL of the model with intercept and independent
variables is 0.000. That is the difference (Chi-square statistics) is 321.455-
0.000 = 321.455 which is significant at 0.05 since P=0.000<0.05. We can
conclude that there is the association between the dependent and indepen-
dent variable(s)in complimentary Log-log link function

41
 Table 3.17: Model Fitting Information

Model -2Log Likelihood Chi-Square df Sig.

Reduced 321.455
Final 0.000 321.455 34 0.000

Goodness-of-Fit
The table tests for consistency between the observed data and the fitted
model. The null hypothesis states that the observed data are consistent with
the fitted model. The null hypothesis is accepted and one concludes that the
observed data were consistent with the estimated values in the fitted model
since the P-value was insignificance p=1.00>0.05. Using complementary Log-
log Link function.

Table 3.18: goodness-of-Fit

Measure Chi-Square df Sig.

Pearson 299.192 462 1.00
Deviance 238.120 462 1.00

Pseudo R-Square
In ordinal regression models, these measures were based on likelihood ratios
rather than raw residuals. There are several measures intended to mimic the
R-squared analysis, but none of them are an R-squared. The interpretation is
not the same, but they can be interpreted as an approximate variance in the
outcome. The three different methods were used to estimate the coefficient
of determination.McFadden’s r-squared (McFadden, 1974) is based on
the log-likelihood kernels for the intercept-only model and the full estimated
model.Cox and Snell’s r-squared (Cox and Snell, 1989)is a generalization
of the usual measure designed to apply when maximum likelihood estimation
is used, as with ordinal regression. However, with categorical outcomes, it
has a theoretical maximum value of less than 1.0. For this reason, Nagelk-
erke (Nagelkerke, 1991) proposed a modification that allows the index to
take values in the full zero-to-one range. From ”Model Fitting Information”
table McFadden R2 (aka pseudo R2) is Pseudo R2 = Model L2/DEV0 =
321.455/321.455 = 1.000. using CLog-log link function.

42
 Table 3.19: Pseudo R-Square

Cox and Snell 0.924

Nagelkerke 1.000
McFadden 1.000

Parameter Estimates
Estimate
While direct interpretation of the coefficients in this model is difficult due to
the nature of the link function, the signs of the coefficients can give impor-
tant insights into the effects of the predictors in the model.

Sig.
These are the p-values of the coefficients that, within a given model, the null
hypothesis that a particular predictor’s regression coefficient zero given that
the rest of the predictors variables are in the model.

Threshold
The response category 1=very unsatisfactory for the general question had
p-value=0.800>0.05, we fail to reject the null hypothesis and concluded that
the regression coefficient for response category 1 for the general question was
zero in the estimation.

The response category 2= unsatisfactory for the general question had p-

value=0.270>0.05, we fail to reject the null hypothesis and concluded that
the regression coefficient for response category 2 for the general question was
zero in the estimation.

The response category 3= average for the outcome variable had p-value=0.032<0.05,
we reject the null hypothesis and concluded that the regression coefficient for
response category 3 for the general question was found to be statistically dif-
ferent from zero in the estimation.

The response category 4= satisfactory for the general question had p-

value=0.006<0.05, reject the null hypothesis and concluded that the regres-
sion coefficient for response category 4 for the general question was found to
be statistically different from zero in the estimation.

Location parameters
The predictor age in years had p-value=0.098>0.05, we fail to reject the null

43
hypothesis and concluded that the regression coefficient for age in years of
the respondent was zero in estimating the general question given that the
other predictor variables were in the model.

The independent variable reference to market demand had P-value=0.286

>0.05, we fail to reject the null hypothesis and concluded that the regression
coefficient for reference to the market demand of the respondent course was
zero in estimating the general question given that the other predictor vari-
ables were in the model.

Lecturer service delivery had a P-value=0.249>0.05, we fail to reject the

null hypothesis and concluded that the regression coefficient for lecturer ser-
vice delivery to the interviewee was zero in estimating the outcome variable
given that the other predictor variables are in the model.

The predictor variable service delivery at faculty office to the students

P-Value=0.033<0.05, we reject the null hypothesis and concluded that the
regression coefficient for service delivery at faculty office was found to be
statistically different from zero in estimating the general question given the
other independent variable are in the model.

The predictor variable service delivery at department office P-value=0.984

> 0.05, we fail to reject the null hypothesis and concluded that the regres-
sion coefficient for service delivery at the department office to the students
was zero in estimating the response variable given that the other predictor
variables are in the model.

Relationship between the students and the faculty subordinate staff had
P-value=0.712>0.05, we fail to reject the null hypothesis and concluded that
the regression coefficient for relationship between the students and the fac-
ulty subordinate staff was zero in estimating the outcome variable given that
the other predictor variables are in the model.

The predictor variable faculty office’s accessibility P-value=0.880>0.05,

we fail to reject the null hypothesis and concluded that the regression co-
efficient for faculty office’s accessibility was zero in estimating the general
question given that the other predictor variables are in the model.

The predictor variable course promotion p-value=0.112>0.05, we fail to

reject the null hypothesis and concluded that the regression coefficient for
course promotion was zero in estimating the response variable given that the

44
other predictor variables are in the model.

Communication skills gained from course of study of the respondent inde-

pendent variable had P-value=0.622>0.05, we fail to reject the null hypoth-
esis and concluded that the regression coefficient for communication skills
gained from course of study of the respondent was zero in estimating the
outcome variable given that the other predictor variables are in the model.

The predictor variable computer skills gained from course of study of the
respondent P-value=0.364>0.05, we fail to reject the null hypothesis and con-
cluded that the regression coefficient for computer skills gained from course
of study of the respondent was zero in estimating the response variable given
that the other predictor variables are in the model.

The predictor variable research skills gained from course of study of the
respondent P-value=0.542>0.05, we fail to reject the null hypothesis and con-
cluded that the regression coefficient for research skills gained from course
of study of the respondent was zero in estimating the general question given
that the other predictor variables are in the model.

The predictor variable admission and registration process in reference to

time had P-value=0.114>0.05, we fail to reject the null hypothesis and con-
cluded that the regression coefficient for admission and registration process
in reference to time was zero in estimating the response variable given that
the other predictor variables are in the model.

Financial ability to pay tuition fees and meet personal needs had P-
value=0.130>0.05, we fail to reject the null hypothesis and concluded that
the regression coefficient for financial ability to pay tuition fees and meet
personal needs was zero in estimating the outcome variable given that the
other predictor variables were in the model.

The predictor library service’s significance P-value=0.030<0.05, we re-

ject the null hypothesis and concluded that the regression coefficient for the
library service was found to be statistically different from zero in estimating
the general question given the other independent variables are in the model.

The predictor variable tutorial Service in the respondents’ course of study

p-value=0.652>0.05, we fail to reject the null hypothesis and concluded that
the regression coefficient for tutorial Service in the respondents’ course of
study was zero in estimating the response variable given that the other pre-

45
dictor variables are in the model.

The predictor variable career counseling service in the faculty of science

P-value=0.325>0.05, we fail to reject the null hypothesis and concluded that
the regression coefficient for career counseling service in the faculty of sci-
ence zero in estimating the general question given that the other predictor
variables are in the model.

Classroom facilities had P-value=0.404>0.05, we fail to reject the null hy-

pothesis and concluded that the regression coefficient for classroom facilities
was zero in estimating the outcome variable given that the other predictor
variables are in the model.

The predictor JKUAT hospital facilities P-value=0.668>0.05, we fail to

reject the null hypothesis and concluded that the regression coefficient for
JKUAT hospital facilities was zero in estimating the response variable given
that the other predictor variables are in the model.

Course laboratory facilities predictor variable had p-value=0.142>0.05,

we fail to reject the null hypothesis and concluded that the regression coef-
ficient for respondent course laboratory facilities was zero in estimating the
outcome variable given that the other predictor variables are in the model.

The predictor variable accommodation facilities inside JKUAT hostels

P-value=0.009<0.05,we reject the null hypothesis and concluded that the
regression coefficient for accommodation facilities inside JKUAT hostels was
found to be statistically different from zero in estimating the general question
given the other independent variable are in the model.

The predictor variable accommodation facilities outside JKUAT had P-

value=0.028<0.05, we reject the null hypothesis and concluded that the re-
gression coefficient for accommodation facilities outside JKUAT was found
to be statistically different from zero in estimating the general question given
the other independent variable are in the model.

The predictor variable internet facilities in JKUAT P-value=0.862>0.05,

we fail to reject the null hypothesis and concluded that the regression coef-
ficient for internet facilities in JKUAT was zero in estimating the response
variable given that the other predictor variables are in the model.

Student center facilities had P-value=0.905>0.05, we fail to reject the null

46
hypothesis and concluded that the regression coefficient for student center
facilities was zero in estimating the outcome variable given that the other
predictor variables are in the model.

Inter-departmental sport events had P-value=0.942>0.05, we fail to reject

the null hypothesis and concluded that the regression coefficient for inter-
departmental sport events was zero in estimating the outcome variable given
that the other predictor variables were in the model.

The predictor variable year of study of the respondent P-value=0.551>0.05,

we fail to reject the null hypothesis and concluded that the regression coef-
ficient for year of study independent variable was zero in estimating the
response variable given that the other predictor variables are in the model.

The predictor variables of the eight departments P-values are 0.897, 0.296,
0.610, 0.203, 0.169, 0.886 and 0.261 for Physics, Medical Micro-Biology, Zool-
ogy, Statistics and Actuarial Sciences, Pure and Applied Mathematics, Bio-
chemistry, Chemistry and Botany respectively on setting alpha level to 0.05,
we fail to reject the null hypothesis and concluded that the regression coeffi-
cient for the eight departments location variable was zero in estimating the
general question in controlling the other predictor variables are in the model.

The interpretation for a dichotomous variable such as gender parallels

that of a continuous variable: the observed difference on 0=’females’ P-
value=0.200>0.05, we fail to reject the null hypothesis and concluded that
the regression coefficient for the gender independent variable was zero in es-
timating the general question in controlling the other predictor variables are
in the model.

Program of study predictor variable 0=’GOK’ P-value=0.282>0.05, we

fail to reject the null hypothesis and concluded that the regression coeffi-
cient for the program of study predictor variable was zero in estimating the
outcome variable in controlling the other independent variables are in the
model.
CLog-log link function was used in the estimation of the parameters

47
 Table 3.20: Parameter estimates for Clog-log logistic regression
Parameters Esti- Std. Wald df Sig. 95% Cofide-
mate Error nce Interval
Lower Upper
Bound Bound
Threshold Very unsatisfactory 0.545 2.152 0.064 1 0.800 -3.672 4.763
unsatisfactory 2.345 2.125 1.218 1 0.27 -1.820 6.511
Average 4.588 2.145 4.575 1 0.032 0.384 8.792
satisfactory 6.066 2.200 7.605 1 0.006 1.755 10.377
Location Age in years -0.136 0.082 2.733 1 0.098 -0.0297 0.025
Reference to the -0.173 0.162 1.140 1 0.286 -0.491 0.145
market demand
Lecturer service 0.248 0.215 1.329 1 0.249 -0.174 0.670
delivery
Service delivery 0.391 0.183 4.561 1 0.033 0.032 0.749
faculty office
Service delivery 0.003 0.156 0.000 1 0.984 -0.303 0.309
department office
Relationship 0.081 0.219 0.136 1 0.712 -0.348 0.510
between students
faculty subor-
dinate staff
Faculty office -0.029 0.193 0.023 1 0.880 -0.407 0.349
accessibility
Course 0.256 0.162 2.520 1 0.112 -0.060 0.573
promotion
Communication -0.090 0.183 0.244 1 0.622 -0.449 0.268
skills gained
Computer Skills 0.150 0.165 0.825 1 0.364 -0.173 0.473
gained
Research skills -0.094 0.155 0.373 1 0.542 -0.397 0.209
gained
Admission regi- 0.242 0.153 2.498 1 0.114 -0.058 0.542
stration process
reference time
Financial ability -0.205 0.135 2.291 1 0.130 -0.470 0.060
to pay tuition
fee and meet needs
Library service -0.358 0.164 4.730 1 0.030 -0.680 -0.035
Tutorial Service 0.066 0.145 0.204 1 0.652 -0.219 0.350
in course
48
 Parameters Esti- Std. Wald df Sig. 95% Cofide-
mate Error nce Interval
Lower Upper
Bound Bound
Location Career counseling 0.143 0.145 0.970 1 0.325 -0.142 0.428
services
Classroom facilities 0.122 0.146 0.697 1 0.404 -0.164 0.409
JKUAT hospital -0.065 0.152 0.184 1 0.668 -0.363 0.233
facilities
Course laboratory 0.220 0.150 2.157 1 0.142 -0.074 0.513
facilities
Accommodation 0.592 0.22 6.773 1 0.009 0.146 1.038
facilities
in JKUAT
Accommodation 0.334 0.152 4.818 1 0.028 0.036 0.631
facilities outside
JKUAT
Internet faci- -0.032 0.184 0.030 1 0.862 -0.394 0.329
lities in JKUAT
Student center -0.019 0.161 0.014 1 0.905 -0.334 0.296
facilities
Inter-department 0.010 0.141 0.005 1 0.942 -0.267 0.288
sports
Year of study 0.114 0.191 0.355 1 0.551 -.261 0.489
[Department=1] -0.071 0.549 0.017 1 0.897 -1.146 1.005
[Department=2] 0.640 0.612 1.093 1 0.296 -0.560 1.840
[Department=3] -0.252 0.494 0.260 1 0.610 -1.220 0.716
[Department=4] -0.717 0.563 1.621 1 0.203 -1.821 0.387
[Department=5] -0.692 0.503 1.893 1 0.169 -1.678 0.294
[Department=6] -0.078 0.541 0.021 1 0.886 -1.137 0.982
[Department=7] -0.624 0.554 1.265 1 0.261 -1.710 0.463
[Department=9] 0(a) 0. . 0 0. . .
[Gender=0] -0.385 0.301 1.640 1 0.200 -0.974 0.204
[Gender=1] 0(a) . . 0 . . .
[Program of study=0] 0.939 0.873 1.157 1 9.282 -0.772 2.650
[Program of study=1] 0(a) . . 0 . . .
a. This parameter is set to zero because it is redundant

49
 Test of Parallel Lines
Test of parallel lines was designed to make judgment concerning the model
adequacy. SPSS tests the proportional odds assumption that is commonly
referred to as the test of parallel lines. The model null hypothesis states
that the slope coefficients in the model are the same across the response
categories. Since the significance P-Value=1.000>0.05 indicated that there
was no significant difference for the corresponding slope coefficients across
the response categories, suggesting that the model assumption of parallel
lines was not violated in the model with the Complementary Log-log link.

Table 3.21: Test of parallel lines

Model -2Log Likelihood Chi-Square df Sig.

Null Hypothesis 0.000
General 0.000(a) 0.000 102 1.000

1. The log-likelihood value is practically zero. There may be a complete

separation in the data. The maximum likelihood estimates do not exist.

50
Chapter 4

CONCLUSION AND
RECOMMENDATIONS

4.1 Conclusion
This project demonstrates the use of ordinal regression statistical technique
to model students’ satisfaction ratings data.This is a statistical tool that used
when the outcome is categorical with a natural ordering. Ordinal regression
allows for predicted probabilities of success to be calculated for each level of
the response. The data for this project was collected from J.K.U.A.T fac-
ulty of science’s one hundred and twenty five students of first to fourth year
picked at random from the eight departments. The data contains most of
the student satisfaction evaluations factors.

Clog-log link became the best model based on the screening criteria the
credibility of model assumption, the fitting statistics i.e. fitting Informa-
tion, goodness of fit information, and the stability of parameter estimation.
Therefore, needless to say, major research findings and implications should
be drawn from the best model.

The explanatory variables related to the satisfaction of faculty involve-

ment is Service delivery at the department office it was identified in the
best model. Student satisfaction with faculty involvement significantly con-
tributes to the probability of students expressing satisfaction with the gen-
erally on faculty of science service delivery. It is evident that a the faculty of
science is one of the largest faculties within the university, thus higher stu-
dent satisfaction rating regarding faculty involvement provides compelling
evidence that faculty members have played a significant role in creating a

51
pleasant environment influenced on student satisfaction for the generally on
faculty of science service delivery.

Furthermore, the library services ware significantly associated with the

satisfaction of generally on faculty of science service delivery. It may provide
evidence that improved service delivery at the library has addressed the needs
of Faculty of science students and contributed to the fulfillment of university
goal, e.g., booking of and reserving of library books online a success.

The study suggested that the accommodation facilities inside and outside
the university that can be improved further by providing services like inter-
net in the hostels and provision of free transport to and from the university
for the students residing outside the university and providing power backups
in the rooms of residence

Overall, this study should be viewed as an important first step for the fac-
ulty of science to explore the relationship between the generally on faculty
of science service delivery satisfaction and multiple explanatory variables
concerning academic programs, facilities and services in the faculty. The
knowledge gained from this study would be beneficial to the faculty of sci-
ence and its students.

The goal was to obtain information from students to establish the ex-
planatory variables that influence satisfaction that could be helpful to deci-
sion makers in faculty of science for improving academic programs, facilities
and services in the faculty. For example the administrators could ensure
that the faculty students could ensure themselves participate in the quality
of academic programs supported by the faculty capacity and facilities and
services. Model assumption of parallel lines was checked to ensure model
adequacy and it was fulfilled by the model, assuring the model goodness of
fit, fitting Information and parameter estimation stability.

Clearly, the ordinal regression modeling is a unique statistical technique

in that the ordinal outcome variable is frequently encountered in the field
of educational research and the model assumption of parallel lines is easily
assumed and verified.

52
4.2 Recommendations
Further research to be done on the signs of the regression coefficients.

In this study the data collected was from the Jomo Kenyatta University
of Agriculture and Technology faculty of science only. The questionnaire
should be rolled out to all the JKUAT faculties, schools and institutes.

The same questionnaire or otherwise could be used in other public and

chartered universities to have a larger sample size for analysis.

As stated earlier random sampling method for collected data. We rec-

ommend that stratified random sampling be used since it is a method of
sampling, which involves the division of a population into smaller groups,
known as strata. In stratified random sampling, the strata are formed based
on their members sharing a specific attribute or characteristic which could be
a sample from each province in Kenya. A random sample from each stratum
is taken, in a number proportional to the stratum’s size when compared to
the population. These subsets of the strata are then pooled to form a ran-
dom sample. The main advantage with stratified sampling will be how it will
capture key population characteristics in the sample. Similar to a weighted
average, this method of sampling produces characteristics in the sample that
will be proportional to the overall population. Stratified sampling works well
for populations with a variety of attributes.

53
Bibliography

[1] A. Agresti. Categorical Data Analysis. 2nd rev-ed., New York: John
Wiley & Sons, 2002.

[2] L. Bailey, Brenda, Curtis Bauman, and A. Lata, Kimberly. Student

Retention and Satisfaction: The Evolution of a Predictive Model., 1998.

[3] Frank Cooney. A Review of the Results and Methodology in the 1999
noel Levitz Student Satisfaction Survey at Salt Lake Community Col-
lege. Salt Lake City,. Utah: Salt Lake Community College., (Eric No:
ED443482), 2000.

[4] K. Damminger, Joanne. Student Satisfaction with Quality of Academic

Advising Offered by Integrated Department of Academic Advising and
Career Life Planning,. Glassboro, (Eric No: ED453769), 2001.

[5] J. Hummel, T. and W. Lichtenberg, J. Predicting Categories of Im-

provement Among Counseling Center Clients. 2001.

[6] A. O’Connell, A. Methods for modeling ordinal outcome variables. Mea-

surement and Evaluation in Counseling and Development, 33(3):170–
193, 2000.

[7] S. Robins, Lynne, D. Gruppen, Larry, L. Alexander, Gwen, C. Fantone,

Joseph, and Wayne Davis. A Prediction Model of Student Satisfaction
with the Medical School Learning Environment. Academic Medicine,,
72(2), 1997.

[8] H. Thomas, Emily and Nora Galambos. What Satisfies Students? Min-
ing Student-Opinion Data with Regression and Decision-Tree Analysis.
Stony Brook, 2002.

[9] D. Umbach, Paul and R. Porter, Stephen. How Do Academic Depart-

ments Impact Student Satisfaction? Understanding the Contextual Ef-
fects of Departments., 2001.

54
[10] Nancy Wild. Rogue Community College Student Satisfaction Survey,
Management. (Eric No: ED448831), 2000.

55
Appendix A

Appendix I

Student Satisfaction Questionnaire

The questionnaire will be used to determine the students’ satisfaction in
relation to their performance.
Please take a few minutes of your time to choose the response which best
describes your opinion in the following statements. Please consider all the
courses you have already taken and currently taking as you formulate your
response. Your response will be considered confidential.

PART I a) Personal Details

1. Department

2. Year of study First Second Third Fourth

3. Gender Female Male

4. Age

5. Program of study
GOK Alternative degree program (ADP)

56
b) faculty involvement
In a scale of 1 - 5 where
1 = Very unsatisfactory
2 = unsatisfactory
3 = Average
4 = satisfactory
5 = Very satisfactory
please rate by ticking in the appropriate box.

1. Course satisfaction in reference to the market demand

1 2 3 4 5

2. Lecturers’ service delivery

1 2 3 4 5

3. Service delivery at the faculty office

1 2 3 4 5

4. Service delivery at the department office

1 2 3 4 5

5. Relationship between students and the faculty sub-ordinate

1 2 3 4 5

6. Accessibility of the faculty office

1 2 3 4 5

PART II
curriculum contents and incorporated psychological factors

1. Course promotion by the faculty

1 2 3 4 5

2. The communication skills gained from your course

1 2 3 4 5

3. The computer skills gained from your course

1 2 3 4 5

4. The research skills gained from your course

1 2 3 4 5

PART III Support services

57
1. Admission and registration process in reference to time
1 2 3 4 5
2. Financial ability to pay school fees and meet your needs
1 2 3 4 5
3. The library services
1 2 3 4 5
4. The tutorial services in your course
1 2 3 4 5
5. Career counseling services in the faculty
1 2 3 4 5
PART IV Facilities
1. The lecture room facilities
1 2 3 4 5
2. The JKUAT hospital facilities
1 2 3 4 5
3. Course laboratory facilities
1 2 3 4 5
4. The accommodation facilities in JKUAT
1 2 3 4 5
5. The accommodation facilities outside JKUAT
1 2 3 4 5
6. The JKUAT internet facilities
1 2 3 4 5
7. The student center facilities
1 2 3 4 5
PART V Recreation activities
1. The faculty inter-departmental sports events
1 2 3 4 5
PART VI General question
1. Generally on faculty of science service delivery
1 2 3 4 5

58