Omitted variable due to Collinearity

[Pasindhu Mawalage]

Hi guys,

I am trying to find which coefficient beta 1 or 2 has a greater effect on child mortality. i have carried out the reparameterisation trick ( or lincom command in stata), but stata omitted the water variable due to collinearity. my null hypothesis is Ho : beta 1 = beta 2 and alternative is H1: beta 1 > beta 2. the following is my regression model:

INFMORTUND~5= β_0+β_(1 ) SANITATION+β_2 WATER+β_3 logGDPPC+U

INFMORTUND~5= 146.1188 - 0.506 SANITATION - 0.015 WATER- 8.453 logGDPPC

my question is : when stata omitted water due to is it because of perfect collinearity?

Thank you

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex. Being able to replicate your problem can be essential to helping you.

I can't understand your problem. I'm not sure why reparameteriszation is lincom related. Normally, we'd rescale parameters by dividing or multiplying the x by some number. I don't see how water is omitted due to collinearity if you report a parameter on water.

Comparing parameters is tricky. Any change in scale will change the comparison so measuring water one way will give you a different answer than measuring water another way. If the b's depend on scale, then it is not clear you can compare them meaningfully. Often, it doesn't make sense to compare parameters this way. There are approaches like J tests, but they're a bit more complex.