regression analysis： bootstrap red x

[Teresa Wang]

Hi everyone, I'm having this problem with my classmates and can't find a solution! When I use bootstrap to test the significance of the coefficients of my regression model, I get little red x because I have dummy variables （UWrepu , Adrepu , year, industry）in my control variables? (I find that when I remove the time and industry fixed effects it's not a red “x” anymore) Please help me to answer! Thank you!

code：bootstrap, reps(1000): reg Underprice SSEr size roa lev Issuerate IPOduration UWrepu Adrepu rd IH F M i.year i.ind,noheader

[Carlo Lazzaro]

Teresa:
please see https://www.statalist.org/forums/help#adviceextras #1 and 4. Thanks.

[Clyde Schechter]

Carlo gives good advice.

Nevertheless, this is the third posting in about 5 minutes of the very same question. It is asked and answered at https://www.statalist.org/forums/for...ootstrap-red-x.

It is completely unreasonable to expect answers within minutes of posting. It occasionally happens by chance. But remember, this is not a paid help line. This is a user forum; everybody is here when they care to be and answers whatever they like. Some questions never get answered because nobody cares, nobody knows, or the question is too unclear and nobody can figure out what is wanted. It is not reasonable to repost a question in less than 24 hours. Even then, don't just repost the same question: if it goes that long with no response it is likely because the question has not been clearly asked, or is missing key pieces of information. So instead of "bump"ing or posting the same thing over again, read the Forum FAQ for advice on why your question might be going unanswered, and then post a revised version that remedies those problems on the same thread.

[Joseph Coveney]

I get little red x because I have dummy variables

It's because in particular bootstrap samples, it just so happens that there isn't any variation in one or more indicator variables, i.e., they're all zeroes or all ones in that randomly selected sample.

There's not much that you can do about that to my knowledge. If you don't have too many red xs, then you might just go ahead and accept the results as they are.

I use bootstrap to test the significance of the coefficients of my regression model . . .

I don't think that that is especially recommended. Bootstrapping is used more for constructing confidence intervals and not so much for significance testing, per se. (Yeah, they're related, but bootstrapping is considered inefficient for significance testing compared to, say, permutation tests.)

Clyde, she might having trouble with the slowness of the forum's software. I often find that it takes forever for a post to appear. I'm guessing that she has the same problem and just assumed that the website ate her earlier attempts, and she's trying to re-post it.

[Joseph Coveney]

It's because in particular bootstrap samples, it just so happens that there isn't any variation in one or more indicator variables, i.e., they're all zeroes or all ones in that randomly selected sample.

I don't think that that's the cause—regress will more likely just omit the predictor, proceed to fit the model and feed bootstrap what it needs.

Do those predictors have substantial proportions of missing values? You might be seeing the consequences where the particular sample doesn't have sufficient complete observations for the number of predictors.