Microsoft Word research method fm doc

Influence of some socio-demographic variables on the use of condom among 
commercial sex workers who had knowledge about condoms, Northwest Ethiopia, 
Dec. 1999. 
 
Helpful tips 
 
a) The X
♠ -test for linear trend should be considered if a significant finding is obtained by 
the ordinary X
♠ -test. 
b) Assess the association of variables (dependent and independent) by use of bivariate 
analysis before going to multivariate analysis.
Exercise 
A cross-sectional study to assess the knowledge, attitudes and practice of the population of 
Dembia district towards traditional harmful health practices was conducted in May 2001. The 
table below shows the age and sex distribution of the study population and individuals who 
underwent certain traditional harmful health practices in the last one year preceding the 
survey [G.D. Alene and M. Edris, Ethiop.J.Health DEV.2002;16(2):199-07]. 
Ever used condoms 
( n = 305) 
Variable 
Yes
No 
Odds 
Ratio 
P-value 
Age of CSWs (in year) 
15 - 24 
25 - 34 
35 - 49 
Educational status of CSWs 
Can't read and write 
Can read and write 
(informal schooling) 
Elementary school 
High School 
Occupation of partners 
Farmer (subsistence) 
Trader (Private business) 
Others 
55 
36 
8 
46 
14 
22 
17 
76 
15 
8 
77 
71 
58 
131 
21 
32 
22 
182 
13 
11 
1.00 
0.71 
0.19 
1.00 
1.90 
1.96 
2.20 
1.00 
2.76 
1.74 
<.001 
(X
♠ for 
linear 
trend) 
=.007 
(X
♠ for 
linear 
trend) 
=.023 
(X
♠ - test) 

---

Research methodology 
113
Variable 
study population 
( n = 6008) 
Number of persons 
who underwent 
THHP* (n = 368) 
Age group (in years) 
< 1 
1 - 4 
5 - 14 
15 - 44 
45 - 64 
65+ 
Sex 
Male 
Female 
194 
1255 
1829 
2381 
312 
37 
2992 
3016 
41 
56 
58 
184 
25 
4 
195 
173 
* = traditional harmful health practice 
Investigate the impact of age and sex on the experience of the study subjects 
towards THHPs. 
 
Epi6 performs a lot of tasks using simple statistical techniques. However, it should be noted 
that Epi6 does not have facilities to perform multivariate analysis, such as, logistic regression. 
The multiple linear regression technique contained by the analysis program (regress 
dependent variable = independent variable1, independent variable2, etc. ) lacks facilities to 
check whether the required assumptions are fulfilled. Therefore, it is not possible to control 
confounding effects with the Epi6 statistical package. In this regard, it will be advisable to 
consult higher versions of Epi Info and SPSS statistical packages.