Non-standard default p-values estout/esttab. Why?

[Fabio Schanaider]

Hey stata users, i’ve been creating regression tables using estout and esttab and just recently noticied that the three default p-values are *0.05, **0.01 and ***0.001. Traditionally I have mostly used/seen in empirical paper tables *0.1, **0.05, ***0.01, so I was wondering if anybody knows precisely why these table packages are more rigorous with there p-value default.

[William Lisowski]

Perhaps you are reading in a different field than I am, or in older literature. I was surprised when I returned to research to see that the tighter p-values are the most commonly seen in the current literature in the area in which I am working, and apparently they are also common in the area in which the author of the community contributed esttab command is working.

Since the default p-values can easily be changed, I would not attach too much significance to the choice of significance levels in esttab.

[Fabio Schanaider]

You are probably right, I was just wondering if anybody could track the specific area. My field is applied econ, and for instance, in the last two editions of AER (top journal in the field) i've found published papers with the standard p-value reporting I'am used to. Obviously sample size, id strategy, assumptions, etc. are all relevant factors in deciding how to report tables. For me it is mostly a visual tool to guide the reader, and I know that there are even serious proposals to abolish it entirely from published articles, but what I meant is that I am not aware of a consensus shift in p-value standard reporting, at least in my area, and I was just wondering where the author(s) of the command decided that their default was more "conventional".

Thank you answering!

[Andrew Musau]

AER in #3 refers to the American Economic Review. As this is a multidisciplinary forum, you should avoid using discipline-specific abbreviations as most people will have no clue what you are referring to. Ben Jann, the author of esttab, is a sociologist and has contributed quite a number of programs that have benefited many in the Stata community. See the following paper that discusses the evolution of "alphas and asterisks" in this field.

https://www.jstor.org/stable/3598292...o_tab_contents

[Fabio Schanaider]

Thank you! Apologies for the abbreviation will keep this in mind when posting in the future.