AMBO UNIVERSITY

COLLEGE OF NATURAL AND COMPUTATIONAL SCIENCES

DEPARTMENTS OF STATISTICS

SENIOR RESARCH REPORT

ON
ASSESSING THE FACTORS THAT AFFECT HEALTH STATUS OF STUDENTS IN AMBO
UNIVERSITY: A CASE OF 3rd YEAR NATURAL AND COMPUTATIONAL SCIENCES IN 2015

Prepared by

Tafese Ashine

ADVISOR: JEMAL AYALEW (MSc)

Submitted To: - Advisor

Submission Date: 01/06/2015
Ambo, Ethiopia

Approval sheet
This was to certify that the research entitled”Factors that Affect Health Status of students” in
Ambo University: in case of College of Natural and computational Science, submitted in partial
fulfillment of bachelor degree in Statistics of under graduate program at department of Statistics.
The assistance and the help received during the course of this investigation have been properly
and suitably acknowledged. Therefore, we recommend that it would be accepted as fulfilling the
[Link]. research project requirement.

Approved by:

Name of Student Signature Date

___________________ ________________ ______________

Name of Advisor Signature Date

___________________ ________________ ______________

Name of Dep’t Head Signature Date

_____________________ ____________ ________________

Name of Examiner Signature Date
_____________________ ____________ ________________

Acknowledgement
First, and foremost, i thank God for giving the opportunity to pursue Our graduate study at the
department of Statistics in Ambo University. I would like to say thanks to Almighty GOD, for
his generous and uncountable help in I life span and by his support in the ups and downs I have
had been passed through, gratitude and unfailing love I fall short of words to express to the
satisfaction of we study.

I have many thanks and uncompareable love in our heart for our advisor Mr Jemal Ayalew for
his invaluable comments, useful suggestions and materials like note that contributed to the
successful realization of our study and all our instructors.

Next I would also like to thank all our friends, families for their endless help and Pastors for
believing.

Table Contents page

CHAPTER ONE…………………………………………………………………………………………………………………………….1
1. INTRODUCTION………………………………………………………………………………………………………………………1
1.1 Background of the Study......................................................................................…..1
1.2 Statement of the problem...........................................................................................2
1.3 Objective of the Study................................................................................................3
1.4 General Objective......................................................................................................3
1.5 Specific Objectives....................................................................................................3
1.6 Scope of the Study.....................................................................................................3
CHAPTER TWO…………………………………………………………………………………………………………………………….5
2. LITERATUR REVIEW………………………………………………………………………………………………………………….5
2.1 Modern Public Health................................................................................................5
2.2 Applications in Health Status.....................................................................................5
2.3 Health Sectors in Ethiopia..........................................................................................6
2.4 Factors that Affect Health Status...............................................................................6
CHAPTER THREE………………………………………………………………………………………………………………………….7
3. DATA AND METHODOLGY……………………………………………………………………………………………………….7
3.1 Description of Study Area..........................................................................................7
3.2 Data............................................................................................................................7
3.2.1 Method of Data Collection......................................................................................7
3.2.2 Sampling design......................................................................................................7
3.3.3 Sample Size Determination.....................................................................................7
3.3 Variable of the study................................................................................................10
3.4 Statistical Model......................................................................................................11
3.4.1 Descriptive Statistics.............................................................................................11
3.4.2 Inferential Statistics...............................................................................................11
[Link] Chi-square Test Statistics...................................................................................12
[Link] Ordinal Logistic Regression...............................................................................13
[Link] The Model of Ordinal Logistic regression.........................................................14
[Link] Fitting an Ordinal Logit Model..........................................................................15
[Link] Parameter Estimates of ordinal logistic regression............................................16
[Link] Checking Model Adequacy................................................................................16
CHAPTER FOUR…………………………………………………………………………………………………………………………18
4. RESULTS AND DISCUSSIONS………………………………………………………………………………………………….18
4.1 Descriptive Statistics................................................................................................18
4.2 Bivariate Analysis....................................................................................................20
4.3 Ordinal Logistic Regression Analysis......................................................................22
CHAPTER FIVE…………………………………………………………………………………………………………………………..28
5. CONCLUSIONS AND RECOMMENDATIONS……………………………………………………………………………28
5.1 Conclusions..............................................................................................................28
5.2 Recommendations....................................................................................................29
REFERENCES...............................................................................................................30

Tables

Table 3.1 Dependent variable in coding form……….…………………..……..………..........9

Table 3.2 Independent variable in coding form……….…………………..……..………........9

Table 4.1 Percentage distribution of Health status of students …………………………...17

Table 4.2 Chi-Square Test Statistics ……………………………………….………………..19

Table 4.3 Model Fitting Information……………….…………………………….………20 Table

4.4 Goodness-of-Fit Statistics………….……………………………………..……….21

Table 4.5 Pseudo R-Square………………………………………………………………….21

Table 4.6 Test of parallel lines……………………………………………………….……...24
Table 4.7 Parameter Estimates……………………………………………………….……...22

Figures

Figure 4.1 Health Status of male and female students ……………………….……………...18

ABBR
EVIATIONS AND ACRONOMY
 AU……………...Ambo University

CHI ………………Consumer Health Informatics

HC ……………….Health Center

HEP ………………Health Extension Program

HIS……………….Hospital Information System

HSDP …………….Health Sector Development Plan

NCHS………… …National Cancer Health Sector

NGO …………….Non-Government Organization

OLR …………......Ordinal Logistic Regression

WHO……………World Health Organization

OR……………….Odds Ratio

PO……………….Proportional Odds

W LS ……………Weighted Least Squares

MLE…………… Maximum Likelihood Estimate

Abstract

This study was intended to investigate the factors that affect the health status of the students in

Ambo University in case of 3rd year College of Natural and computational Science 2015. The
self-health assessment was applied as an indicator of health status in this study. The sample for

this study was drawn from the students of 3 rd year College of Natural and computational Science

Ambo University. The information was collected by questionnaire from the students. The

dependent variables were health status of students. The independent variables included sex,

income, lack of hygiene of café, usage of medicine, smoking cigarette, alcohol consumption,

physical exercise, Environmental factor and bad smelling around the cafe. The statistical method

of analysis applied for testing the association and the factors affecting health status of the

students was chi-square test statistics and Ordinal Logistic regression. The results revealed that

sex, income, lack of hygiene of café, smoking cigarette, bad smelling around the café, Alcohol

consumption and Environmental factor are factors that affect health status of the students. In

general most of the variables included in the regression model had significant effects like sex,

low income (<200birr),lack of hygiene of café, alcohol consumption, smoking cigarette, bad

smelling around the café and Environmental factor are found to be more significantly

associated with health status of students.

Key Words: Factors that Affect Health Status of Students, Stratified Random Sampling, Ordinal
Logistic Regression, SPSS and Ambo University
CHAPTER ONE

[Link]

1.1 Background of the Study

Health is the level of functional or metabolic efficiency of a living being. In humans, it is the
general condition of a person's mind, body, usually meaning to be free from illness, injury or
pain (as in "good health" or "healthy"). Health is the presence of a positive capacity to lead
energetic satisfying and productive life. Society as a whole benefits from peoples good health
just as individual do. For that reason many government and voluntary agencies strive preserve
and improve the health of all the people. The World Health Organization (WHO) defined health
in its broader sense in 1946 as "a state of complete physical, mental and social well-being and
not only the absence of disease or infirmity"(King and Motulsky, 2002).

WHO, an agency of the United Nations work to promote better health through the world. Health
services include all services dealing with the diagnosis and treatment of disease, or the
promotion, maintenance and restoration of health (Grad, Frank P. 2002). Mental illness is
described as 'the spectrum of cognitive, emotional, and behavioral conditions that interfere with
social and emotional well-being and the lives and productivity of people. Having a mental illness
can seriously impair, temporarily or permanently, the mental functioning of a person. Other
terms include: 'mental health problem', 'illnesses, ‘disorder', 'and dysfunction’ (Hungerford et al.
2012).

Ethiopia experiences a heavy burden of disease mainly attributed to communicable infectious

diseases and nutritional deficiencies. However, there has been encouraging improvements in the
coverage and utilization of the health services over the periods of implementation of the Health
Sector Development Plan (HSDP).HSDP constitutes the health chapter of the national poverty
reduction strategy and aims to increase immunization coverage and decrease under-five mortality
at large. The health service currently reaches about 72% of the population and the Federal
Ministry of Health aims to reach 85% of the population by 2009 through the Health Extension
Program (HEP).
The HEP is designed to deliver health promotion, immunization and other disease prevention
measures along with a limited number of high-impact curative interventions. Health sector
development plan health facilities in Ethiopia are too few because of large population size and
incorrect distribution of resources, lack of money, basic drugs and equipment. The major health
problems of the country remain largely preventable communicable diseases and nutritional
disorders. The health care system represents a social determinant of health as well as it
influences other determining factors (Pankhurst, Richard,Gezahegn, Yoseph, 2004).

Health standard of Ethiopia is too few when compared to other because of several reasons like in
sufficient economy and improper use of materials in hand. Ethiopia faces the difficult problems
that improved health status in both vital for growth and development and expensive as well, this
and many other problems placed very sever burden on development in general and health
resources in particular in more ways than [Link] absence of regulatory systems to monitor the
quality, safety, and efficacy of Medicines can compromise the overall effectiveness of health
care services and endanger the public health. A strong regulatory system is considered an
essential component of a health system. In Ethiopia, the Food, Medicine and Health Care
Administration and Authority (FMHACA), formerly known as the Drug Administration and
Control Authority, regulates the country’s pharmaceutical sector in an environment vulnerable to
drug smuggling and the circulation of substandard medicines (Yehulu Denekew, 2014).

1.2 Statement of the problem

Health service forms part of the basic need of the society. Health is factor in the development of
endeavor of country hence it determine the future of nation by affecting the wellbeing of its
population for these realization of its human potential. Ethiopia one of the developing countries
have a shortage of major public health service and facilities such as clean water supply,
education and others are adequate.

Becoming unhealthy was a crucial issue in Ethiopia. A research has been conducted in Dilla
University using logistic regression model (MeleseAbebe, 2013/2014) on this topic. But the
outcomes of health status are ordinal on their nature. Therefore, assessing the determinant factors
of health statutes using ordinal model is mandatory. As a result, we are eager to use ordinal
logistic regression model to see the problem. In line with this, the following research questions
were addressed in this study.
These are:-

 Does ordinal logistic regression model suitable for health status data?
 What are the determinant factors of health status at University level in particular Ambo
University?
 Are the students satisfied in the service provided by the university clinic?

1.3 Objective of the Study

1.3.1 General Objective

 The main objective of this study was to investigate the factors affecting health status of
3rd year natural and computational science students in Ambo University in 2015.

1.3.2 Specific Objectives

Under the heading of the main objective this study tried to address the following specific
objectives. These are:-

• To apply ordinal logistic regression model for health status data.

• To determine the most prevalent factor(s) of health status in Ambo University.

• To know which sex group most affected by health status.

1.4 Scope of the Study

The study focus on health status and investigation of the major factors related to health status of
students in Ambo University 3rd year Natural and computational Science. The study deals about
the major factors that affect health status students.
1.5 Significance of the Study

This study will be expected to show clearly the possible of obtaining better utilization of the
clinic by identifying the key factors that affected the health status of students. Besides, the
finding of this study wii be expected to solve the factors that affected the health of student in this
university.

1.6 Limitation of the Research

This study purposes to target all students of Ambo University who study in College of Natural
and computational Science in 2015. There are a number of problems and difficulties that should
be suggested during the accomplishment of this paper.

Some of the limitations are:-

• The students were not volunteer to give full information about health status.

• Lack of detailed information in Ambo University clinic about health status

CHAPTER TWO

2. LITERATUR REVIEW

2.1 Modern Public Health

Literature search applications for healthcare professionals facilitate searching biomedical
literature databases to find and display medical reference information. With onset of the
epidemiological transition and as the prevalence of infection disease decreased through the
20thcentery. Public health began to put more focus on chronic disease such as cancer and heart
disease. For instance in the United states public health work [Link] Josephine Backer
established many programs to help the poor in New York keep their infants healthy, leading
teams of nurses in to the crowded neighborhoods of hell Kitchen and teaching mothers how to
dress, feed and bath their babies. Studying data from the United States, where minimum illegal
drinking increased over time, Dee and Evans (1997) used an instrumental variable approach to
minimize within state variations in alcohol availability to estimate the impact of drinking on
health status. The WHO reports lack of exclusive breast feeding during the first six months of
life contributed as to over a million avoidable child deaths each year. There is recognition that
our health is affected by many factors including where we live, Genetics, our income, our
education states and our social relationships. There are known as “social determinants of
health.”The new public health seeks to address those health inequalities by adducting for
population, based policies that improve health in an equitable manner (WHO, 2008)

2.2 Applications in Health Status

The main stakeholders in the healthcare process were health status consumers (patients).
Consumer oriented care, where patients directly involved in the process of care, will greatly
improve the healthcare process. Technology can play key role in consumer-oriented healthcare
(for example, making information accessible to consumers, integrating consumers’ preferences
into HISs, remote monitoring, communication, etc.), which studied in branch of medical
informatics called Consumer Health Informatics (CHI). The management of diseases with
chronic conditions is very costly (Morris CollenTribute, 2012).
2.3 Health Sectors in Ethiopia
Matrices of health in Ethiopia were among the world’s worst. According to the US government
Ethiopians health care system is wholly inadequate, even often improvements. Ethiopian
experiences a heavily border of disease mainly attributed to communicable infectious disease
nutritional deficiencies, shortage and high turnover of human resources and in adequacy of
essential drugs and supplies have also contributed to the Burden. However there has been
encouraging improvements in the coverage and utilization of the health services over the period
of implementation of health sector development plan (KesetebirhanAdmassu, 2014).

2.4 Factors that Affect Health Status

According to Probst,et al (2002)34% of African and Native Americans along with 25% of
Hispanics communities are poor, or live below the low income level. This poverty causes
minorities to delay necessary medical care and affects the amount of health status resources
available within community. Probst, Samuels, Jespersen, et al (2002) stated that 70% of African
Americans, 84% of Hispanics and 67% of Native Americans live in communities that are Health
Professions Shortage of water. In addition, Ellickson et al. (2004) found that early smokers were
affected at least three times more than non-smokers to regularly use marijuana and hard drugs.
Students who binge drank were affected than both non-drinkers and drinkers who did not binge
to report involvement in other risky health behaviors according to Cutler (Miller and Norton,
2007).

Most health determinants like lifestyle background, economic and social conditions are cause for
the illness of Ambo people. Disease like typhoid and food hygiene are usual disease that affect
Ambo peoples as well as Ambo University students, because of factors like whether change,
food hygiene ,sanitation problem dusts are usually cause common called most students(Teka. G.
E, 2004; Baraki N, 2004). In this study the model we use was Ordinal logistic regression. Ordinal
logistic regression in which the means a type of logistic regression analysis that when the
response variable was categorize more than two with having natural order or rank. we can rank
the values, but the real distance between categories was unknown. Diseases are grade on scales
from unhealthy to healthy.
CHAPTER THREE

3. DATA AND METHODOLGY

3.1 Description of Study Area
The study was conducted at Ambo University main campus which was one of the higher
education institutions in Ethiopia. It was found in which are located western Shoa of Oromia. It
was created as agricultural college in 1939. Ambo University was situated about 114 km for
from west of Addis Ababa. It becomes one of the higher institutions in 1999 E.C and training
with the help of ministry of education. Now a day, the University has three additional branches,
namely Awaro, wolliso and guder campus.

3.2 Data
The Method of Data Collection in this study was primary source using questionnaire as
instruments of data collection. The questionnaire was administered and distributed to the
students under the direct supervision of the researcher.

3.2.1 Method of Data Collection

3.2.2 Sampling design

Sampling design was a system of taking small ratio of observation from a large population to get
information of that large population from the sample observation by using some statistical
design. In this study, stratified random sampling were used to determine the sample size of the
target population the study population were heterogeneous.

3.2.3 Sample Size Determination

To determine the sample size, degree of freedom and margin of errors would be considered.

In this study, stratified random sampling was applied to determine the sample size of the students
from the whole population. The total number of samples ’n’ was calculated as follows:-

Where, Nh=total number of students from each department.

N1 = 50 Number of students in biology, N2= 43 Number of students in chemistry,

N3 = 32Number of students in physics, N4 = 44 Number of students in sport science,

N5 = 41Number of students in mathematics and N6 = 57 Number of students in statistics.

N=N1+N2+N3+N4+N5+N6= 50+43+32+44+41+57 = 267

College of Natural
and computational
sciences

sport
BIOLOGY Chemistry Physics Statistics
sciences Mathematics

Year III Year III Year III Year III year III Year III

Now, the sample size was determined by using the formula for sample size determination for the
population. This is.
(Z α )2 p(1− p)
2
n 0=
d2
Z α - Critical value at 95% confidence level of certainty (1.96)
2

P = probability of success, q=probability of failure.

Based on the previous study p=0.75(Teka.G.E, 2004)

(1.96)2 0.75 (1−0.75)

n 0=
0.052
=289
Note, if a Finite population correlation (FPC) is >5%, there is a satisfactory sample size
approximation to the required, so we used the correlation formula of sample size.
 n0
FPC = = 289/267=1.082 > 0.05, so the sample fraction was greater than 0.05 so, we can do a
N
n0
n=
sample size n0
1+
N
289
n= =139
289
1+
267
Where, n=total number of samples, α=the significance level=0.05.

d = margin of error =0.05 and = Z0.025=1.96.

For stratum h (each department), number of sample was calculated by using proportional
allocation; based on this the following results were obtained.

Nh
nh = N n , h=1 , 2, 3 , 4 ,5 , 6 , where nh is sample size of the hth stratum (department)

N1 50 N2 43
n1 = n= 139=¿26 n2 = n= 139=¿22
N 267 N 267

N3 32 N4 44
n3 = n= 139 = 17 n4 = n= 139 = 23
N 267 N 267

N5 41 N6 57
n5 = n= 139 = 21 n6 = n= 139 = 30
N 267 N 267

Therefore, sample size (n) = n1 + n2 + n3 + n4 + n5 + n6= 26+22+17+23+21+30 = 139 is

calculated.

3.3 Variable of the study

The dependent variable was Health status of the students, which was categorized as Unhealthy
(illness often), Moderate and Healthy (strong)).

Table 3.1 Nature of the dependent variable with the corresponding code

Categories How often did you visit Ambo university clinic Code
due to sickness for the last 4 months?
Unhealthy >=5 times 0
 Moderate 1-4 times 1
Healthy None 2
The independent variables that used in this study were sex, income, and hygiene of café, usage of
medicine, smoking cigarette, alcohol consumption, physical exercise, environmental factor &bad
smelling around the cafe.

Table 3.2 List of independent variables with respective name and category codes

Variable designation Description Value labels

X1 Sex 0=”Female”, 1=”Male”
X2 Average income of student 0=”<200”, 1=”200-600”, 2=”600-1000”,
3=”>1000”
X3 Lack of hygiene of café 0=”Good”, 1=”fair”,2=”bad”
X4 Usage of medicine 0=”No”, 1=”Yes”
X5 Smoking cigarette 0=”No”, 1=”Yes”
X6 Alcohol consumption 0=”No”, 1=”Yes”
X7 Physical exercise 0=”never”,1=”1-3days”,2=” >3 days”
X8 Environmental factor 0=”have not influence” ,1=”Not such
much”,2=”Strict factor”
X9 Bad smelling around the café 0=”No”, 1=”Yes”

Demographic and socioeconomic

characteristics
Health status of the
• Sex
Students (healthy, moderate
• Average income
and unhealthy)
• Lack of hygiene of café
• Environmental factor
• Bad smelling around the café
Lifestyle of the students
• Usage of medicine
 • Smoking cigarette
• Alcohol consumption
• Physical exercise

3.5 Statistical Model

3.5.1Descriptive Statistics
Statistical analysis that would use in this study was descriptive statistics such as a table, bar-charts to
describe the frequency distribution and percentage.

3.5.2 Inferential Statistics

Inferential statistics is the procedure by which we reach a conclusion about a population based
on the information contain in the sample drawn from that population. Therefore among
inferential statistics, we would use chi-square test and ordinal logistic regression.

Chi-square Test Statistics

It was used to analysis the association between dependent and independent variables. Hypothesis
testing: Ho: There was significant association between dependent variable and independent
variables Vs H1: Not Ho

To test the null hypothesis we can compare X 2cal with X 2tab is given by

∑ ∑ ( Oij−Eij )
2

X
2
cal = (i=1, 2.....n, j=1, 2,3,…m)Where, Eij is expected frequency
Eij
corresponding to (ij)th ∧Oijbe the observed frequency. From this calculated and tabulated values
with degree freedom (n-1) (m-1) we decide about the rejection of Ho that say there was no
significant association between two variables.

We reject Ho if X 2cal greater than X 2tab or if p-value less than level of significance, otherwise we
do not reject Ho.

Steps of chi-squares:-

1. State hypothesis (i.e. null and alternative hypothesis)

 2. Select desired level of significance (alpha level of significance)
3. Check the statistics
4. Compute the calculated value¿)
5. Obtain the critical (tabulated) value ( X 2tab )
6. Interpret the test

Assumptions of chi-squares

[Link] sample must be randomly selected from the population.

[Link] population must be normally distributed for the variable under study.

[Link] observation must be independent of each other.

[Link] cell (each data entry) should contain at least five observations.

3.5.3 Ordinal Logistic Regression

Logistic regression may be useful when we are trying to model a categorical dependent variable
as a function of one or more independent variables being the dependent variable has two
outcomes. Ordinal logistic regression (OLR) is a type of logistic regression analysis when the
response variable has more than two categorizes with having natural order or rank.

In statistics, the ordered logit model (also ordered logistic regression or proportional odds
model), is a regression model for ordinaldependent variable. It is natural to consider methods for
more categorical responses having more than two possible values. The most well-known of these
ordinal logistic regression methods is called the proportional odds model. The basic idea
underlying the proportional odds model is re-expressing the categorical variable in terms of a
number of tertiary variables based on internal cut-points in the ordinal scale.

We can consider the 3 tertiary logistic models corresponding to regressing each of the

logit [ p ( y ≥ 3 ) ] =α j + β 1 X 1 + β 2 X 2 +. … … ..+ β k X k

Separately againstthe X ' s .many variables of interest are ordinal. That is, we can rank the values,
but the real distance between categories is unknown. Diseases are grade on scales from
unhealthy to healthy.
Assumptions of ordinal regression

• The dependent variable should be measured at the ordinal level.

• Ordinal independent variables must be either continuous or categorical.

• There is no multicollinearity.

• Each independent variable has an identical effect at each cumulative split of the
ordinal dependent variable.

• The relationship between each pair of outcome groups is the same.

• The effects of any explanatory variables are consistent or proportional across the
different thresholds.

3.6The Model of Ordinal Logistic regression

ln ( θ j )=α j−[ β X 1+ β2 X2 +.… … ..+β 9 X 9 ]¿
1
¿

Where j goes from 1 to the number of categories minus1.

This was the part we really want to find out. We have been thinking of attitudinal responses to
abortion as a set of three unordered responses, but there is a very clear and intentional ordering to
these responses like as unhealthy, moderate, and healthy. If we know that a category is ordinal
then there are special models that tell us how independent variables relate to someone being
higher or lower on the scale.

We could use the ordered logit model so that we could use the categorized directly as our
dependent variable. Different links lead to proportional odds models or ordered probit models.
The model cannot be consistently estimated using ordinary least squares; it is usually estimated
using maximum likelihood.

The /proportional odds model and the /partial proportional odds modelare special cases of the
cumulative logit model. This is a cumulative logit model that assumes that the odds of response
below a given response level are constant regardless of which level we picked. This model
allows separated intercepts for the cumulative logit, but restricted the parameter sets for the
predictors to be the same across all logits. A proportional odds model that constrains some
predictors to have common parameters and leaves other predictors free to have separate
parameters is called a partial proportional odds model.

The basic form of the generalized linear model is:

θ j −[ β 1 X 1+ β 2 X 2 +… .+ β k X k ]
Link ( γ j )=
exp ⁡( τ 1 Z 1+ τ 2 Z 2+ …+ τ m Z m)

Where,γ j is the cumulative probability for the j t h category,θ j is the threshold for the j t hcategory,
β1, … , βk are the regression coefficients, X1, …, Xk are the predictor variables, and k is the
number of predictors. The numerator on the right side determined the location of the model.
Thedenominator of the equation specifies the scale. The τ 1,…,τm are coefficients for the scale
component and Z1,…,Zm are mpredictor variables for the scale component (chosen from the same
set of variables as the X’s).

3.6.1 Fitting an Ordinal Logit Model

The quantity to the left of the equal sign is called a logit. If the spacing between levels of the
ordinal response scale is known, so that numerical scores can reasonably be assigned to the
response levels, then a /mean response model can be fit. When we have more than two events,
we can extend the ordinal logistic regression model. Suppose the underlying relationship to be
characterized is: yi = x'β+ε, ( Agresti, 2002; Greene, 2003), where y i is the exact but un observed
dependent variable (perhaps the exact level of improvement by the patient); x is the vector of
independent variables, and β is the vector of regression coefficients which we wish to estimate.

The Model fitting Information gives the -2 log-likelihood (-2LL) values for the baseline and the
final model and a chi-square to test the difference between the -2LL for the two models.

The goodness of fit statistics indicated that the model fits much better than the location only
model. From the observed and expected frequencies, we could compute the usual Pearson and
Deviance goodness-of-fit measures.

Oij−Eij 2
The Pearson goodness of fit statistic is
2
= ( Eij ) .

Oij
The deviance measure is D = 2Oijln( ).
Eij
Both of the goodness-of-fit statistics should be used only for models that have reasonably large
expected values in each cell. If we have a continuous independent variable or many categorical
predictors or some predictors with many values, we may have many cells with small expected
values. There are several R2like statistics that can be used to measure the strength of the
association between the dependent variable and the predictor variables. They are not as useful as
the R2 statistic in regression, since their interpretation is not straight forward.

Three commonly used statistics are:-

2
1)Cox and Snell R2 = 1 – ( L ¿ ¿)
n

Cox∧Snell R 2
2) Nagelkerke’s R = 2
2 and
1−L(β (0))
n

L( β)
3) McFadden’s R2 = 1 – ( ). Where, L (β) is the log-likelihood
L(β(0) )
function for the model with the estimated parametersand L ( β(0) ) is the log-likelihood with just
the thresholds, and nis the number of cases(sum of all weights).

3.6.2 Parameter Estimates of ordinal logistic regression

The model is usually estimates using maximum likelihood. The method of maximum likelihood
corresponds too many well-known estimation methods in statistics. The proportional odds model
assumes that the true β values are the same in all three models, so that the only difference in
models is the intercept terms, α Cc=1, 2, 3. This means that the estimates from the three ordinal
models could be pooled to provide just one set of β estimates. By exponentiation the pooled
estimate relative to a given predictor, i.e. taking, e β j we obtained an estimate of the common odds
ratio (OR) that described the relative odds for y ≥ 3 for values of X jdiffering by one unit.

The maximum likelihood estimate (MLE) of θ is that value of θ that maximizedlik (θ): it is the
value that made the observed data the “most probable”. If the X i are iid, then the likelihood
n
simplified to: lik (θ) =∏ f (xi /θ).
i=1
Rather than maximizing this product which could be quite tedious, we often used the fact that the
logarithm is an increasing function so it would be equivalent to maximized the log likelihood:-

𝑙(θ) =∑ log ⁡(f ( xi|θ )).

n

i=1

3.6.3 Checking Model Adequacy

Goodness of Fit and the Deviance

A most general way to detected lack of fit searches for any way the model fails. A goodness of
fit test compared the model fit with the data. This approach regards the data as representing the
fit of the most complex model possible the saturated model, which has a separated parameter for
each observation. In certain cases, this test statistic has a large-sample chi-squared null
distribution.

The likelihood-ratio statistic for this test is the deviance of the model and Pearson of the model
Oij−Eij 2
The Pearson statistics for testing goodness of fit is2= ( ) .
Eij

Oij
The corresponding likelihood-ratio (deviance) statistics is = 2Oijln( ). Then, 2and G2have
Eij
approximated chi-squared null distributions.G2 follows2distribution with k degree of freedom.
Where, k is the number predicated variable.

If G2 is larger that is L1>L0, then reject H0 and conclude that the model is good fit or variables are
significantly affected the respondent or dependent variable.
Pseudo-R2 statistics would be use in ordinal regression to estimate the variance explained by the
independent variable.
This would be made through the following:-

Cox∧Snell R 2
Nagelkerke’s R = 2
2
1−L(β (0))
n
Testing of parallel lines

The statistical hypotheses can be stated as follows:

HO: β 1=β 2=…=β k versus H1: at least one is different.

For a POM to be valid, the assumption that all the logit surfaces are parallel must be tested. The
test of parallelism makes use of -2 log-likelihoods for the constrained model (the model that
assumes the planes or surfaces are parallel) and for the general model (the model that assumes

planes or surfaces are separated). The chi-square statistic is the difference between of -2 log-
likelihoods of the two models. There is some debate about the adequacy of the proportional odds
test as it may be too sensitive to sample size (Scott et al., 1997). A non-significant test statistic is
usually taken as sufficient evidence that the POM is valid whereas a significant test statistic
might be misleading. If the test is significant, some have proposed additional, often graphical,
methods for determining whether the logit surfaces are actually parallel. Other solutions include
fitting a partial proportional odds model, or dichotomizing the ordinal outcome.
CHAPTER 4

[Link] AND DISCUSSIONS

4.1 Descriptive statistics

A total of 139 questionnaires were administered and data were collected from 3 rd year natural and
computational sciences college students at Ambo University. The administered questionnaires
were successfully responded with 100% response rate. The frequency and the percentage
distribution of the characteristics of respondents were summarized in Table 4.1.

Table 4.1 Frequency (Percentage distribution) of Health status of students due to Unhealthy,
moderate & Healthy with their frequencies in each categories are described as table instead:
Variables category Health Status of Students %(n)
Unhealthy Moderate Healthy Total%(n)
Sex Male 35 (14) 62.5(35) 79.1 (34) 59.71(83)
Female 65(26) 37.5 (21) 20.9 (9) 40.29 (56)
Income <200 72.5(29) 46.4 (26) 41.9 (18) 52.52(73)
200-600 17.5 (7) 41.1 (23) 34.9 (15) 32.37 (45
)
>600 10 (4) 12.5 (7) 23.2 (10) 15.11 (21)
Hygiene of café Good 42.5 (17) 25 (14) 20.9 (9) 28.78 (40)
Fair 55 (22) 53.6 (30) 37.2 (16) 48.92 (68)
Bad 2.5 (1) 21.4 (12) 41.9 (18) 22.30 (31)
Medicine No 40 (16) 41.1(23) 46.5(20) 42.45(59)
Yes 60(24) 58.9 (33) 53.5(23) 57.55 (80)
Smoking No 100 (40) 98.2(55) 81.4 (35) 93.52(130)
Yes 0(0) 1.8 (1) 18.6 (8) 6.48 (9)
Alcohol No 52.5 (21) 82.1(46) 69.8(30) 69.78(97)
Yes 47.5(19) 17.9 (10) 30.2 (13) 30.22 (42)
Exercise Never 35 (14) 25 (14) 32.6 (14) 30.22 (42)
1-3 days 45 (18) 53.6 (30) 46.5 (20) 48.92(68)
>3 days 20 (8) 21.4 (12) 20.9 (9) 20.86(29)
Environmental Have no 0(0) 33.9 (19) 25.6 (11) 21.58 (30)
influence
Not such 85 (34) 33.9 (19) 27.9 (12) 46.76 (65)
much
Strict 15 (6) 32.1 (18) 46.5 (20) 31.66(44)
factor
Smelling No 57.5 (23) 39.3(22) 20.9 (9) 38.85(54)
Yes 42.5(17) 60.7 (34) 79.1 (34) 61.15 (85)

As per Table 4.1, out of 139 respondents 59.7% were males and 40.3% were females. This
percentage decreases slightly for unhealthy status with 35 percent males and increases for
healthy with 79 percent male students. Whereas the percentage increases slightly for unhealthy
status with 65 percent females and decreases for healthy with 20.9 percent female students.

In terms of monthly income of the total respondents 51.58(30) of the students has less than 200
birr, 32.37 (45) from 200 – 600 birr and 15.11 (21) more than 600 birr. Among those income less
than 200 birr, most of the respondents 72.5(29) were belongs to unhealthy conditions. On the
other hand majority of the respondents 41.1 (23) from income category 200 – 600 were belongs
to moderate conditions and 20.9(9) from those income greater than 600 were under healthy
conditions.

In terms of Environmental factor of the total respondents 52.52(73) of the students have no
influence, 46.76 (65) from have not such much and 31.66 (44) from strictly factor. Among those
have no influence, most of the respondents 52.52(73) were belongs to unhealthy conditions. On
the other hand majority of the respondents 46.76 (65) from have not such much, were belongs to
moderate conditions and 31.66 (44) from those from strictly factor were under healthy
conditions.

Status of health problem of students because of bad smelling around café (61.15%) is exceeding
the students in which their health status is not determined by those problems (38.85%).

We can also illustrate the sex category by using bar-chartas below as a counter example.

Bar chart
 80.0% Healthy status of
students
unhealthy
moderate
healthy

60.0%
Percent

40.0% 79.07%

65.0%
62.5%

20.0% 37.5%
35.0%

20.93%

0.0%
female male

Sex of students

Figure 4.1 Health Status of male and female students

From Figure 4.1we can understood that the number of females that are more exposed to health
problems are larger than males of the same status.

4.2 Bivariate Analysis

Bivariate analysis is one of the simplest forms of the quantitative (statistical) analysis. It involves
the analysis of two variables (often denoted as X, Y), for the purpose of determining the
empirical relation/association between them. Since the dependent variable in this study is ordinal
the type of relation is association. To determine the association in the bivariate analysis the
appropriate test statistic is Chi-Square Test. Chi-square is simply an extension of a cross-
tabulation that gives us more information about the association.
The bivariate analysis between the dependent variable health status and the expected independent
variables using chi-square were presented in table 4.2.

Table 4.2 Bivariate analysis using chi-square test statistic

Variables Pearson Chi-Square df Sig.

Sex 17.033 1 0.000*
Income 12.962 3 0.044*
Lack of hygiene of café 19.998 2 0.000*
Usage of medicine 0.432 1 0.806
Smoking cigarette 15.250 1 0.000*
Alcohol consumption 9.724 1 0.008*
Physical exercise 1.333 2 0.856
Environmental factor 37.162 2 0.000*
Bad smelling around the café 11.673 1 0.003*
 (*) Indicate the statistically significant.

Table 4.2 shows that among 9 independent variables included in the bivariate analysis 7 of them
have statistically significant associationwith health status of the studentsnamely sex of students,
bad smelling around the cafe, income, Lack of hygiene of cafe, smoking cigarette, Alcohol
consumption and Environmental factor. However,Physical exercise and Usage of medicine have
no significantassociation with Health status of the student.

The Pearson chi-square with p-value <0.05means there is significant association between health
status and the aforementioned explanatory variables those have a p-value of less than 0.05.
Among the significant variables with the modest p- values less than 0.2 (Agresti, 2002 ) were
included for the final model (multiple ordinal logistic regression model).

4.3Ordinal Logistic Regression Analysis

Under Ordinal Logistic Regression Analysis we can deal Model Fitting Information,Goodness-
of-Fit, Pseudo R-Square, Parameter Estimates and Test of parallel lines.

Logitlink function is used in the analysis because it is evenly distributed categories and is
reasonable choices when the changes in the cumulative probabilities are gradual andlogit
involves all levels of the response and dichotomizes the response scale.

Model Fitting Information

The Model Fitting Information table were displayed in table 4.3, which gives the -2 log
likelihood for the intercept only and final models, can be used in comparisons of nested models.
The statistically significantchi-square statistic (p<0.05) indicates that the Final model gives a
significant improvement over the baseline intercept-only model. This tells us that the model
gives better predictions than if we just guessed based on the marginal probabilities for the
outcome categories. Therefore, the Full model (with factors that affect health status as a
predictor) is significantly better than the ‘health status’ model.

Table 4.3 Model Fitting Information

-2 Log Chi-
Model Likelihood Square df Sig.

Intercept
295.439
Only

Final 201.745 93.695 14 .000

Goodness-of-Fit
The goodness fit test were displayed in table 4.4. The result of Table 4.4 suggest that the model
does fit very well (p>0.05)(i.e. fail to reject the null hypothesis depending on the observed
data). Means that the model fits the data adequately.

Table 4.4Goodness-of-Fit Statistics

Chi-Square df Sig.

Pearson 241.204 246 .574

Deviance 194.813 246 .993

Link function: Logit

Pseudo R-Square

For logistic and ordinal regression models,it not possible to compute the same R 2 statistic as in
linear regression so, three approximations are computed instead.

Table 4.5 Pseudo R-Square

Cox and Snell .490

Nagelkerke .553

McFadden .310

What constitutes a “good” R2 value depends upon the nature of the outcome and the explanatory
variablesHere, the pseudo R2 values (e.g. Nagelkerke = 55.3%) indicates that there is relatively
small proportion of the variation in health status between students . This is just as we would
expect because there are numerous factors that affect health status of student. The results support
the conclusion that the final model fit the data well.

Final Model Parameter Estimates and Test of parallel lines

A stepwise multiple logistic regression analysis was carried out to select the most important
covariates among the 7 covariates provided from the bivariate logistic regression analyses with
the modest p-value less than 0.25. As a result 7 of the variables were found to be significant
using the stepwise likelihood ratio test of the multiple logistic regressions procedure at
significance level of 0.05(see Table 4.7). Before applying the final multiple ordinal logistic
regression models with 7 covariates for the intended purpose, it has to be assessed test of parallel
lines assumption.

Test of Parallel Lines

One of the assumptions underlying ordinal logistic regression is that the relationship between
each pair of outcome groups is the same. This is commonly referred to as the test of parallel lines
because the null hypothesis states that the slope coefficients in the model are the same across
response categories (and lines of the same slope are parallel). If we fail to reject the null
hypothesis, we conclude that the assumption holds. As a result the parallel line assumption test
was displayed in table 4.6. For our model, the proportional odds assumption appears to have held
because our significance of our Chi-Square statistic is .364 > .05.

Table 4.6 Test of parallel lines

-2Log Chi-
Model Likelihood Square df Sig.

Null
201.745
Hypothesis

General 186.531(a) 15.214(b) 14 .364

Table 4.6 shows parallel line test for general model with chi square value 15.214 and p-
value=0.364 which is greater than the 5% level of significance,fail to reject the null hypothesis.
Therefore, there is no enough evidence to saylocation parameters (slope coefficients) are not the
same across response categories. Means that the parallel line assumption is held.

The model fitting information, goodness fit and parallel line tests were satisfied for the given
data. Then the multiple ordinal logistic regression coefficients can be estimated using the
maximum likelihood estimation method implemented in the SPSS package. The results are
displayed in Table 4.7 consistsof the coefficients, their standard errors, the Wald test and
associated p-values (Sig.), the 95% confidence interval of the coefficients and odds ratios. The
thresholds are shown at the top of the parameter estimates output, and they indicate where the
latent variable is cut to make the three groups that we observe in our data. Based on Table 4.7 the
appropriate interpretation were made as well.

Table 4.7 Parameter Estimates of the Final Ordinal Logistic Regression

Estimate Std. Wald df Sig. 95% Confidence Interval
Error Lower Upper Bound
Bound
[Healthy = 0] -7.523 1.534 24.057 1 .000 -10.529 -4.517
[Healthy = 1] -4.593 1.439 10.189 1 .001 -7.413 -1.773
[Sex=0] -1.560 .415 14.155 1 .000* -2.372 -.747
[Sex=1] 0a . . 0 . . .
[Income=0] -2.207 .610 13.082 1 .000* -3.402 -1.011
[Income=1] -.830 .602 1.901 1 .168 -2.009 .350
[Income=2] 0a . . 0 . . .
[cafe=0] -1.728 .615 7.901 1 .005* -2.932 -.523
[cafe=1] -1.577 .535 8.686 1 .003* -2.625 -.528
[cafe=2] 0a . . 0 . . .
[smoking=0] -2.893 1.207 5.744 1 .017* -5.260 -.527
[smoking=1] 0a . . 0 . . .
[Alcohol=0] .967 .455 4.516 1 .034* .075 1.858
[Alcohol=1] 0a . . 0 . . .
[Factor=0] 1.148 .550 4.354 1 .037* .070 2.227
[Factor=1] -1.005 .458 4.813 1 .028* -1.903 -.107
[Factor=2] 0a . . 0 . . .
[smelling=0] -1.166 .420 7.698 1 .006* -1.989 -.342
[smelling=1] 0a . . 0 . . .

 (*) Indicate the statistically significant.

Out of the considered explanatory variables, Environmental factor, sex, low income (200) , bad
smelling around the café, Lack of hygiene of café, smoking cigarette and Alcohol consumption
have a significant relationship with that of health status of students.
The threshold coefficients are representing the intercepts, specifically the point (in terms of a
logit) where health status might be predicted into the three categories.

ln ( θ 1) =−7.523−[−1.560 X 1−2.207 X 2−1.728 X 31−1.577 X 32−2.893 X 5+ 0.967 X 6 +1.148 X 81−1.005 X 82−1.166

ln ( θ 2) =−4.593−[−1.560 X 1−2.207 X 2−1.728 X 31−1.577 X 32−2.893 X 5 +0.967 X 6+ 1.148 X 81−1.005 X 82−1.166

In this coefficient the negative sign indicates that those variables have negative effects on health
status of students are sex, low income, lack of hygiene of café, smoking cigarette and bad
smelling around café. Conclusion, the findings indicate that Health status of a student is
significance associated with sex, low income, alcohol consumption, lack of hygiene of cafe, bad
smelling around the café, smoking cigarette and Environmental [Link] also the independent
variables that have no significance association with Health status of a student are usage of
medicine and physical exercise.

Interpretations of the Significant Coefficients in the Final Model

Female: The odds for females being in healthy versus moderate and unhealthyis 1.560 less than
males when the other variables in the model are held constant. Likewise, the odds for females
being inthe combinedhealthy andmoderate versusunhealthyis 1.560 less than males when the
other variables in the model are held constant.

Alcohol consumption- The odds for not used alcohol consumption being in healthy versus
moderate and unhealthy is 0.967 less than have used alcohol consumption when the other
variables in the model are held constant. Likewise, the odds for not used alcohol consumption
being in the combined healthy and moderate versus unhealthy is 0.967 less than have used
alcohol consumption when the other variables in the model are held constant.

Bad smelling around café- The odds for not Bad smelling around café being in healthy versus
moderate and unhealthy is 1.166 less than have Bad smelling around café when the other
variables in the model are held constant. Likewise, the odds for not Bad smelling around café
being in the combined healthy and moderate versus unhealthy is 1.166 less than have Bad
smelling around café when the other variables in the model are held constant.
Lack of hygiene of café - The odds for good hygiene of café being in healthy versus moderate
and unhealthy is 1.728 less than bad hygiene of café when the other variables in the model are
held constant. Likewise, the odds for good hygiene of café being in the combined healthy and
moderate versus unhealthy is 1.728 less than bad hygiene of café when the other variables in the
model are held constant. The odds for fair hygiene of café being in healthy versus moderate and
unhealthy is 1.577 less than bad hygiene of café when the other variables in the model are held
constant. Likewise, the odds for fair hygiene of café being in the combined healthy and moderate
versus unhealthy is 1.577 less than bad hygiene of café when the other variables in the model are
held constant.

Interpretations of the Significant odd ratio in the Final Model

By taking the exponent of the pooled estimate relative to a given predictor, i.e. taking e βj , we
obtain an estimate of the common odds ratio that describes the relative odds for X j differing by
one unit. Values greater than one indicate that the variable in question increases the odds of
being Unhealthy student and values between 0 and 1 indicate a decrease in the odds of being
Unhealthy student.
Thus, the above table demonstrates the sex, low income (<200), lack of hygiene of café, smoking
cigarette and alcohol consumption and smoking cigarette increases the odds of an individual
student being Unhealthy is just the complement of the odds of being level healthy. exp ( ^β), is a
factor by which the odds of being Unhealthy of individual change when i th independent variable
increases by one unit (i.e. from 0 to 2). For instance from Table 4.6 the value of the odd ratio
exp ( β^ ) =¿ 0.178, for good hygiene of café andexp ( β^ ) =¿ 0.2066, for fair hygiene of café
indicates that students having good hygiene of café have 0.178 times chances to be Unhealthy
than those having fair hygiene of café and exp ( β^ )= 0.2101, for sex indicates that female students
have 0.2101 times chances to be Unhealthy than male student.
CHAPTER FIVE

CONCLUSIONS AND RECOMMENDATIONS

Conclusions
The main objective of this study was to investigate the factors affecting health status of students
in Ambo [Link] descriptive analysis findings shown that among the respondents 28.8 %
of the students were unhealthy (visit the university clinic more than 5 times for the consecutive
four months due to sickness), 40.1% were moderate (visit the university clinic 1 to 5 times) and
30.1% were under healthy (does not visit the university clinic) condition. The study indicates that
students which getting lower income (72.5%) per-month are exposed unhealthy status and more
female students are unhealthy (65%) when compared to male students (35%) with their health
status.

The current study also shows that the factors that affect health status of students were sex, low
income(<200), lack of hygiene of café, bad smelling around café, alcohol consumption, smoking
cigarette and Environmental factor are found to be more significantly associated with health
status of students.

In this analysis we have looked at regression models that can be applied when our outcome is
represented by an ordinal [Link] we have seen how to evaluate the ordinal
Proportional odds model by completing a series of tertiary logistic regressions at each of the
cumulative splits in the data, and how this can allow us to directly evaluate the consistency in
Odds ratios across an ordinal outcome. Where the Proportional odds assumption is justified
ordinal regression models can be a powerful means of summarizing relationships that utilizes all
the information present in the ordinal outcome.

The Health status of a student is strongly associated with sex of students, low income(<200),
smoking cigarette, lack of hygiene of café, Environmental factor, alcohol consumption and bad
smelling around café. Perhaps due to the biological factors as well as the prevailing low socio
economic status females for maintaining Health status is mandatory.

The independent variables that have no significance association with Health status of a student
are lack of usage of medicine and physical exercise.
Recommendations

Based on the results of this study the following recommendations were made. These are

 The university should give emphasis on factors that affect health status of students.
 Female students have some differences on health status when compared to male students.
In order to address this deference the university should give special treatment for female.
 The concerned bodies should work on minimizing lack of hygiene of café and bad
smelling around café in the university.
 The Dean of Ambo University should give attention to problems and report for the
concerned body to solve those problems.
REFERENCES

[1]. Agresti, A (1996). An Introduction to categorical data Analysis. John Wiley and Sons,
Inc., New York.
[2]. Agresti, A. (2002) Categorical Data Analysis, Second Edition. Hoboken, New Jersey: John
Wiley & Sons, Inc.

[3]. Agresti, A. (2010), Analysis of Ordinal Categorical Data, 2nd Edition,and New York: John
Wiley & Sons.

[4]. Alcohol Alert (National Institute on Alcohol Abuse and Alcoholism)(30 PH 359). October
1995. Retrieved 1 Nov 2013.

[5]. Applied Logistic Regression (Second Edition). David W. Hosmer and Stanley Lemeshow.

[6]. CAMH Healthy Aging Project. (2008). Improving Our Response to Older Adults with

Substance Use, Mental Health and Gambling Problems: A Guide for Supervisors, Managers, and

Clinical Staff. Toronto: Centre for Addiction and Mental Health.

[7]. Ethiopian Health and Health related Indicators, [Link]

[8]. Greene, William H., Econometric Analysis (fifth edition), Prentice Hall, 2003, 736-740.

[9]. Healthcare in Ethiopia, [Link] [Link]

[10]. Health Sector Development Plan; [Link]

[11]. Long, J. S. and Freese, J. (2006) Regression Models for Categorical and Limited Dependent
Variables Using Stata, Second Edition. College Station, Texas: Stata Press.

[12]. Liao, T. F. (1994) Interpreting Probability Models: Logit, Probit, and Other Generalized
Linear Models. Thousand Oaks, CA: Sage Publications, Inc.

[13]. Ordinal Regression using SPSS for social scientists (second edition).

[14]. Powers, D. and Xie, Yu. Statistical Methods for Categorical Data Analysis. Bingley, UK:
Emerald Group Publishing Limited.
Appendix Ι

AMBO UNIVERSITY

COLLEGE OF NATURAL AND COMPUTATIONAL SCIENCE

DEPARTMENT OF STATISTICS

Questionnaire

Dear Respondents: The objective of this study is design to gather information about factors that
affect health Status of students in Ambo University 3 rd Year Natural and Computational Science.
The confidentiality of your response will be kept and only use for research purpose.

Instruction: Please respond to question by marking “X” in the box. You are kindly request to fill
the questionnaire clearly and neatly, and you are not request to write your name.

1. Sex: A) Male B) Female

2. Age: A) 16-20 B) 21-25 C) 26-30 D) above 30
3. Department? A) biology B)chemistry C) physics D) sport science
E) Mathematics F) statistics
4. Average income per month (in birr):
A) <200 B) 200-600 C) 600-1000 D)>1000
5. How often did you visit Ambo University clinic due to sickness for the last 4 months?
A) >5 times B) 1-4 times C) None

6. How do you see the Cleannessof your cafe? A) Good B) Fair C) Bad

7. What do you suggest about the safety of service given from the clinic?

A) More satisfactory B) Moderate satisfactory C) not satisfactory

8. Do you use the medicine prescribed by Doctor/Nurse properly?

A) Yes B) No

9. Do you smoke cigarette? A) Yes B) No

10. Do you use (drink) alcohol?A) Yes B) No

11. Do you think bad smelling around the café influences on your health? A) Yes B) No

12. How much do you think that environmental factors affect your health status?

A) have not influence B) Not such much C) Strict factor

13. How much time do you do physical exercise per week?

A) Never B) 1-3 days C) >3 days

Thank you in advance for your cooperativeness!!!

DECLARATION

This is to declare that the project report entitled “Assessing the factors that affect health status
of students ” submitted by us in partial fulfillment of requirements for the award of the degree
of Bachelor of Sciences, in Statistics, Ambo university, is an honest record of the work carried
out by us under the Advisor and guidance of Jemal Ayalew.