# Fixed effects of independent variables highly correlated with the intercept. Is it an issue?

In a linear mixed model run with lmer() I got this output for the correlation between fixed effects. My independent variable is almost perfectly correlated (negatively) with the intercept. Is this a problem?.

I am not sure what this actually means and if I need to actually drop this variable from the model.

Correlation of Fixed Effects:
(Intr) Age  Status Tarsus Quality
Age       -0.061
Status2    0.115 -0.829
Tarsus    -0.991  0.041 -0.138
Quality   -0.049 -0.087  0.134 -0.033
GroupSize -0.067  0.024  0.145 -0.023  0.163


Correlation between the intercept and slope (more generally, the parameter estimate for the effect of a continuous predictor) is not specific to mixed models. It typically means that the mean of your predictor variable is far from zero (a classic example is when people use year as a continuous predictor variable).

It's probably not a problem. If you mean-center your variables, though, you will make your intercept term more interpretable; rather than indicating the expected response of an individual with tarsus length equal to zero, it will be the expected response of an individual with the mean tarsus length (see Schielzeth 2010 Simple means to improve the interpretability of regression coefficients).