0

I am attempting an multilevel analysis looking at political trust across different countries. For my level 2 contextual independent variables I will be using macroeconomic data such as GDP growth, inflation, unemployment, GDP per capita. I look to see how much macroeconomic performance can explain variance in political trust between countries. I am unsure on how I should treat these variables. I have read literature saying you shouldn't center these variables, other literature says you should. Right now the data is in its raw form, e.g. GDP per capita ranges from \$17000 for one country to \$85,000 for another country. GDP growth ranges from 0.9% to 11%. Unemployment ranges from 2% to 15%. When I run the regression with the raw figures in Rstudio I get this message

Warning: Some predictor variables are on very different scales: consider rescaling

What would be the best way to treat these variables, would it be to rescale them? Standardize them or center them? I am using Rstudio for this analysis and would appreciate recommendations on r functions to use as I am new to stats.

Nick Cox
  • 48,377
  • 8
  • 110
  • 156
user18325005
  • 1
  • Have you seen https://stats.stackexchange.com/questions/29781/when-conducting-multiple-regression-when-should-you-center-your-predictor-varia/29783#29783 – mdewey Mar 01 '22 at 14:31
  • 1
    It depends on your methods: GDP per capita measured in dollars is several thousands times unemployment measured as percentages. This may not matter with simple linear regression but may matter with methods such as penalised regression or nearest neighbour classification – Henry Mar 01 '22 at 16:36
  • What about for a linear mixed model? @Henry – user18325005 Mar 01 '22 at 18:07

0 Answers0