# Women in National Legislatures

A simple query to get data by Fabrizio Gilardi meant I was digging out old analyses from my article on the political representation of women in national legislatures. I put of running beta regressions on these data for too long, and now there was no reason not to.

Embarrassingly, in this context I realized that the coefficients in table 3 are incorrect — although marginally. It appears that in the process of changing the dependent variable from one that caters for the percentage of women in the population to ignoring it (makes it much easier to explain), I forgot to replace the entire table with new values (this was in the days before Sweave and odfWeave). Given that the results are essentially the same, I never noticed — but it still feels quite silly. Anyway, here are the corrected numbers, first as coefficient plot comparing the two models:

So the blue dots and lines use gender representation scores as dependent variable; the red dots and lines use the proportion of women in national legislatures as dependent variables. Below the corresponding table:

 Representation scores Proportion (Intercept) 0.653*** 0.154** (0.058) (0.058) PR/MMP 0.043* 0.052** (0.019) (0.019) Party Quotas 0.003 0.004 (0.005) (0.005) Statutory Quotas 0.045* 0.041 (0.021) (0.021) Political Rights -0.001 -0.004 (0.008) (0.008) Age Democracy 0.000 0.000 (0.000) (0.000) Professional Jobs 0.000 0.000 (0.001) (0.001) Nordic 0.147** 0.139** (0.044) (0.043) Eastern Europe -0.074 -0.062 (0.040) (0.040) Asia -0.105** -0.107** (0.036) (0.036) Middle East -0.097* -0.124** (0.044) (0.043) Sub-Saharan -0.031 -0.019 (0.036) (0.035) Latin -0.051 -0.048 (0.032) (0.032) R-squared 0.536 0.577 N 94 94

Beta regression, proportion of women as dependent variable:

 Proportion (Intercept) 0.689** (0.259) PR/MMP 0.204* (0.083) Party Quotas 0.015 (0.025) Statutory Quotas 0.219* (0.096) Political Rights -0.011 (0.035) Age Democracy 0.001 (0.001) Professional Jobs 0.001 (0.004) Nordic 0.988*** (0.255) Eastern Europe -0.380* (0.182) Asia -0.488** (0.163) Middle East -0.469* (0.193) Sub-Saharan -0.170 (0.162) Latin -0.253 (0.146) Pseudo R-squared 0.562 N 94

(Sorry, the arm package does not seem to support beta regressions at the moment, so no coefficient plot)

## 2 Replies to “Women in National Legislatures”

1. In case anyone should get the wrong impression, by “not supporting” beta regressions, I mean that you cannot throw a model at the coefplot() function; of course you can extract the coefficients and standard deviations and have these plotted!

This site uses Akismet to reduce spam. Learn how your comment data is processed.