Hi everyone,
I'm seeking advice on selecting the most suitable statistical method in Stata for analyzing my panel data covering the years 1990-2020.
I aim to identify factors influencing whether occupations experienced above-average or below-average wage changes over this period compared to the total of other occupations.
1. Linear (Fixed-Effects) Regression for Panel Data
I could go for a linear regression here and choose the relative deviation of wage change of an occupation compared to the average wage change of all occupations as the dependent variable.
While probably feasible, this method leads to less intuitive interpretations. For example a 10% increase in unionization might result in a 2.5% increase in relative wage change, which can be difficult to grasp.
2. Ordinal Logistic Regression for Panel Data (RE + FE)
As an alternative, I could calculate an ordinal logistic regression for panel data and use the "xtologit" command in Stata (for random effects) or the user-written "feologit" command (for fixed effects).
To do this, I would have to convert the continuous percentage changes in wage into categorical variables (e.g., 1 = '0-1% above-average', 2 = '1-2% above-average', etc.). This method should simplify interpretation but result in a perhaps significant loss of detailed information.
Could you advise me on the most appropriate regression method to use, or suggest another possible method in Stata for addressing my research question?
Thank you very much in advance!
I'm seeking advice on selecting the most suitable statistical method in Stata for analyzing my panel data covering the years 1990-2020.
I aim to identify factors influencing whether occupations experienced above-average or below-average wage changes over this period compared to the total of other occupations.
- Data Description: I have panel data with percentage changes in wages for different occupations relative to the base year 1990. The wage changes are expressed as continuous metric variables.
- Research Goal: I want to examine which factors determine the relative wage change of an occupation compared to the average wage change of all occupations.
1. Linear (Fixed-Effects) Regression for Panel Data
I could go for a linear regression here and choose the relative deviation of wage change of an occupation compared to the average wage change of all occupations as the dependent variable.
While probably feasible, this method leads to less intuitive interpretations. For example a 10% increase in unionization might result in a 2.5% increase in relative wage change, which can be difficult to grasp.
2. Ordinal Logistic Regression for Panel Data (RE + FE)
As an alternative, I could calculate an ordinal logistic regression for panel data and use the "xtologit" command in Stata (for random effects) or the user-written "feologit" command (for fixed effects).
To do this, I would have to convert the continuous percentage changes in wage into categorical variables (e.g., 1 = '0-1% above-average', 2 = '1-2% above-average', etc.). This method should simplify interpretation but result in a perhaps significant loss of detailed information.
Could you advise me on the most appropriate regression method to use, or suggest another possible method in Stata for addressing my research question?
Thank you very much in advance!
Comment