st: Conceptual clarity in Instruments validity

[Abekah Nkrumah <ankrumah@gmail.com>]

Dear All,

I tried something for the IV and seem to get good results but am a bit
skeptical of the procedure . Below is what I did.

I used the following as instruments for a composite  empowerment
Index. The composite  empowerment index is made up of 4 sub-indices
all constructed via polychoric pca. The variables used in constructing
the 4th sub-index are:

age difference between couples
number of other wives for a husband, an indication of the level of polygamy
 age at first marriage.

The instruments used for the composite empowerment variables are as follows:

The ratio of wife to husband's age interacted with the proportion of
households in the cluster that have monogamous marriages.
The non-self cluster difference in the number of death amongst
daughters and sons.

My skepticism is however in the fact that the two variables making the
first instruments can directly be traced to the first and second
variables used in computing the 4th sub-index s above (i.e. age
difference between couples and number of other wives). Can i argue
that the instruments are entirely different from the two variables
used in computing the sub-index and so can be used as valid
instruments?

Thanks very much


Gordon

*
*   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/