st: RE: RE: El-Badry correction help

[Nick Cox <n.j.cox@durham.ac.uk>]

Just repeating a question (once) is allowed on Statalist, but it's better to wonder why no one replied first time round. 

One possible reason is widespread ignorance about what the El-Badry correction is. If you gave a word explanation, or better a formula, then those like me who've never heard of it might look at your code to see whether your code matched the recipe. 

If the beast is just too complicated to explain, then clearly that would not be your fault, but you're limiting your readership to those who know already, on the evidence to date an empty set. 

Nick 
n.j.cox@durham.ac.uk 

Nanlesta Pilgrim (a.k.a. Terri) 

I'm sending again in hopes that someone might have an answer.

What am I missing? I am trying to apply the El-Badry correction to census data.

I began by creating a counting variable within each age category (15-49).
Then I did the following:

regress pnsborn pzeroborn  /*restricting the analysis to the first observation*/
predict yhat /*predicted values from the regression model*/

*Produced a scatter plot with regression line added
graph twoway lfit pnsborn pzeroborn || scatter pnsborn pzeroborn

*Compute residuals, create new variable residuals:
predict residuals, resid

*Produce a residual plot by age groups with horizontal line at 0
*Should be no pattern
twoway (connected residuals age3, sort), ytitle(residuals) yline(0)
xlabel(#12, labels valuelabel)

The graph is actually showing a pattern- when there should be none.
What have I either done wrong or failed to add?


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/