Kenya's Regression Results

Return to Previous Page

First scatterplot the WAZ against education and roofing to check for outliers and see the basic association between the variables.  This should look familiar.


wpe6.jpg (16047 bytes)


From this scatterplot (with an overlying regression line) it is clear that there is a positive association between better education and better nutritional status (WAZ score), as would be expected.  There are not any blatant outliers, therefore it should be safe to continue with the regression analysis using this variable. 


wpe5.jpg (15357 bytes)


The scatterplot shows that there is higher nutritional status associated with the better category of roofing (0=good roof; 1=bad roof) which is corrugated iron.  Again, no serious outliers are visible and the association is in the expected direction so it is now possible to continue to the regression analysis.

Try this exercise for the regression analysis:

1.  Open keast4j.sav

2.  Click on Statistics, Regression, Linear and enter waz in the Dependent variable list and enter dlowedn, dbadro, and roof_edn (the interaction term you create for roof and eduction) in the Independent variable list.

3.  Click on OK.

The results look like this:

wpe1.jpg (10469 bytes)


wpeB.jpg (8948 bytes)

wpeC.jpg (12148 bytes)

wpeD.jpg (17163 bytes)



The results include education, roof and the interaction variable of the two.  The interaction variable was included in this case because it might be expected that there would be some relationship between the socio-economic status of the family (estimated by roofing) and the level of education of the respondent that might lead to differences in the nutritional status of the child.  In order to control for this, the interaction variable was included.  There does not appear to be a significant level of interaction between roof and education, therefore the variable should be dropped from the equation and the regression should be run again and interpreted at that point.


REGRESSION without the INTERACTION (run the same one as shown above, but drop the variable ROOF_EDN from the model):

wpe7.jpg (8686 bytes)

       wpe8.jpg (12449 bytes)

       wpe9.jpg (14311 bytes)



The results of the regression add support to the argument that education does have an effect independent of SES (roofing) on nutritional status. The coefficient for roofing is somewhat small (B=-0.158) and the difference is not significant (p=0.108). For education level, on the other hand, there is a larger change for each unit increase in nutritional status. There is a -0.420 point change in poor education for each unit increase in nutritional status (WAZ score). This is also highly significant in the equation (P=0.000).  It is looking more likely that education of the mother does have a genuine effect on the nutritional status of the child.

The resulting equation is :

WAZ Score = - 0.987 + 0.420 (EDUCATION) + 0.158 (ROOF)

Return to Top of Page