Regression Table Interpretation

[Ayse Uygun]

Hello everyone, I'm new in Stata. I have this table to interpret but my question is about constant values and values in bracket? Are they significant to interpret or what should I do to interpret?

[Clyde Schechter]

The values in parentheses (brackets) are the standard errors of the corresponding effects. As for the constant terms themselves, they are the expected value of the outcome variables of the regressions when all of the independent variables are zero. Whether these constant terms are of any use depends on the way the regression was set up. In many cases, the situation of all of the independent variables being zero is not even a possibility in the real world--so the constant terms are of no interest and should be ignored. In many other cases, such a situation is possible, but it isn't a situation that has any particular importance or relevance, one might still choose to ignore them.

But if the situation where all of the independent variables are zero corresponds to some meaningful or important case, then the constant terms, being the corresponding expected value of the outcome, might be of interest and you would report it. In doing so, it is important to remember that it is not the effect of anything: it is just the expected value of the outcome in this special case. It's "statistical significance" would represent a test of the hypothesis that the expected value of the outcome in that all-predictors-zero situation is, itself also zero. Even where the expected outcome in this case is of interest, it is truly rare for there to be any interest in testing whether that outcome is also zero, so, unless the research goals specify interest in this hypothesis, you would report the constant term and its standard error, but you would ignore the "statistical significance."

[Ayse Uygun]

The values in parentheses (brackets) are the standard errors of the corresponding effects. As for the constant terms themselves, they are the expected value of the outcome variables of the regressions when all of the independent variables are zero. Whether these constant terms are of any use depends on the way the regression was set up. In many cases, the situation of all of the independent variables being zero is not even a possibility in the real world--so the constant terms are of no interest and should be ignored. In many other cases, such a situation is possible, but it isn't a situation that has any particular importance or relevance, one might still choose to ignore them.

But if the situation where all of the independent variables are zero corresponds to some meaningful or important case, then the constant terms, being the corresponding expected value of the outcome, might be of interest and you would report it. In doing so, it is important to remember that it is not the effect of anything: it is just the expected value of the outcome in this special case. It's "statistical significance" would represent a test of the hypothesis that the expected value of the outcome in that all-predictors-zero situation is, itself also zero. Even where the expected outcome in this case is of interest, it is truly rare for there to be any interest in testing whether that outcome is also zero, so, unless the research goals specify interest in this hypothesis, you would report the constant term and its standard error, but you would ignore the "statistical significance."

Thank you so much sir for your answer. It helped really well. I have one more question. When I want to examine my data week by week, my weeks' p values are always zero. I don't understand why. Do you think why, in your mind?

[George Ford]

The t-statistics are all very large (the pvalue is 0 only due to rounding; they are very small).

And, these coefficients measure the difference between the constant for week1 (the excluded week).

[Carlo Lazzaro]

Ayse:
as an aside, you should take a look at the R_sq between=0.
It seems that there's no variation in your observations.

[Ayse Uygun]

Ayse:
as an aside, you should take a look at the R_sq between=0.
It seems that there's no variation in your observations.

thank you so much Mr. Lazzaro

[Ayse Uygun]

The t-statistics are all very large (the pvalue is 0 only due to rounding; they are very small).

And, these coefficients measure the difference between the constant for week1 (the excluded week).

Thank you so muh Mr. Ford