Please help! I need to know which p-value to use to reject or accept the null hypothesis in my project.

[TheLadyDothReadTooMu]

There are two probability values EViews generates: the one for the independent variables regressed (written as Prob. next to the t-statistics column) and the one to determine the overall significance of the model (Prob(F-statistics) written below the table). To reject or accept the null hypothesis of my project, which one of those p-values do I compare with the level of significance? Note that I have only one independent variable.

[startz]

If you only have one independent variable, are the values different?

[TheLadyDothReadTooMu]

Yes, they are. I got 0.2712 for the Prob. of the t-statistic for the independent variable but the Prob(F-statistic) was 0.00001. I don't know which one to use.

[startz]

You've done something unusual, though not necessarily wrong. You'll need to post more details.

[TheLadyDothReadTooMu]

What details do you need?

[startz]

Perhaps a screenshot. Maybe the work file including the equation you estimated

[TheLadyDothReadTooMu]

Not to sound really stupid but how do I upload pictures here?

[startz]

Click on the Attachments tab below the reply field.

[TheLadyDothReadTooMu]

Alright. Thank you.

This is all I have of the results as I didn't run it myself. I used Panel Least Square for the regression analysis.

[startz]

It is hard to see what's going in here. I suspect that fixed effects were included, in which case you want the p-value next to the t-statistic.

[TheLadyDothReadTooMu]

I'm sorry but I don't understand what you mean by any of that. Isn't the p-value already next to the t-statistic?

[startz]

Probably should use the p-value next to the t statistic as opposed to the one near the F

[TheLadyDothReadTooMu]

I thought so too but my supervisor says otherwise. I thought the Prob(F-statistic) was only to test the overall significance of the model or the goodness of fit (as I've read in some journals) but my supervisor doesn't think so. If I had more than one independent variable, what p-value should I use? And what did you mean by fixed effects?

[startz]

The t test tests whether a given coefficient equals zero. The F test tests whether all the coefficients except the constant equals zero. So if there is only one coefficient the p-values should be identical. The output you have must have something unusual behind it, but there's no way to know what it is without knowing exactly how it was created.

[TheLadyDothReadTooMu]

Thank you so much for explaining this!