Fig 5. Correlations between rankings and labor market deficiencies.

The horizontal axis denotes the strength of the correlation between given labor market deficiency and the rankings. The vertical axis denotes the difference in constants from Model (3) between the rankings based on a method with and the rankings without a given type of adjustment. We report all estimated coefficients from the model, and mark those which pass the bar of 15% significance. Methods sensitive to selection have to have selection correction (see Table 2). Methods sensitive to segmentation have to adjust for industry, occupation or firm size. Methods sensitive to the existence of similar men and women are non-parametric and semi-parametric methods from Table 2. Data available at [http://grape.org.pl/data/gender-wage-gaps-around-eu-and-across-methods].