2
$\begingroup$

I am running a pooled OLS regression for my thesis. My adjusted r square is 0.74 which is quite high compared to other published studies. I have checked for normality of residuals, serial correlation and homoscedasticity. All the 3 tests meet the conditions for OLS. Should I be concerned about the high adjusted r square? Is there any test I could do to know whether the regression model has no other flaws?

$\endgroup$
  • 1
    $\begingroup$ Have you generated graphical diagnostic plots for the regression? If not, that's what you should do next. Look in particular for outlying regressor values (e.g., those with large mahalanobis distance from the mean): a single such outlier can create an artificially inflated $R^2$ (which will inflate the adjusted $R^2$). $\endgroup$ – whuber Nov 27 '16 at 17:57
0
$\begingroup$

The adjusted R square is what it is. Differences with other published studies cold be due to differences in the data and/or the predictor variables.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.