EViews.com

Hi,

I want to estimate a meta-regression by OLS (where the dependent variable is the t-estatistic from diferent articles and the indepedent variables are dummy variables that characterize these articles). So, because of that the program shows the mensage: "Near singular matrix. Regressors may be perfectly collinear". In fact, this is true, but I know that the EViews allows to correct the bias that comes from having data from the same study/article.

Anyone know how I can do that?

Thanks for your help,

Regards,

ES

I would guess you've fallen into the dummy variable trap.

Thank your for the answner.

Probabily it has. But in various articles in the literature, the authors can estimate a regression of this type. For that reason, I know that EViews have an option that allows to correct the bias that comes from having data from the same article. Do you know how I do that?

Thanks for for your attention,

Regards,

ES

Thank your for the answner.

Probabily it has. But in various articles in the literature, the authors can estimate a regression of this type. For that reason, I know that EViews have an option that allows to correct the bias that comes from having data from the same article. Do you know how I do that?

Thanks for for your attention,

Regards,

ES

Trust Gareth. Look up dummy variable trap.