Regression with Age Stratification

1.  Open keast4j.sav
2.  Click on Data, Select cases, If condition is satisfied..., and then on IF and a new table will appear and you should enter age >11 and age <=36 in the box.
3.  Click on Continue and OK.  (this select children 12 to 36 months)
4.  Now, click on Statistics, Regression, Linear and place waz in the Dependent variable box.
5.  Place the variables age, agesq, hmeasyn, htresp, dlowedn, meas_age in the Independent variable box.
6.  Click on Continue and OK

INTERPRETATION:

The regression results show a clear difference from the previous full regression. The coefficient for measles, which is of our great interest, is now much more believable (B=1.430) and it is significant (p=0.041).  The mother's nutrition status is significant as is the education of the mother, although it does not greatly affect the measles coefficient (you can run the model adding the variables for education and height one by one, if you would like to see the change in the coefficient for measles at each step).  The gives us more confidence that measles immunization does provide protection to a child's health and therefore better nutrition status for the child.

Now just to see the difference in another age strata, select the age group <11 months and run the analysis again.  The results will be quite different for the younger children.

Repeat the steps for the above exercise, but now select for age <11 in step 2.

INTERPRETATION:

Looking at the coefficient for measles shows that there is now a strong positive effect of measles immunization for young children (B=2.288), although it might be a function of the small sample of children that are immunized in that age group.  At least there is clearly a positive effect, as you would expect from immunization.  If a mother is immunizing her child at this young age, she is also likely accessing other clinic offerings.  This analysis for the younger age group is likely of less interest because most children do not receive immunization at this young age, but it provides a comparison for the previous age group 12-36 months.

DO NOT FORGET TO SELECT FOR ALL AGE GROUPS WHEN YOU FINISH, or DO NOT SAVE YOUR DATA WHEN YOU FINISH!!!!