As a general thing, with no such line, if the papers he is talking about report p values or confidence intervals without being extremely clear about every step of the process that got them there, then they are lies.
-
-
Replying to @FrederickGuy4 @bradpwyble and
I should say that, like many researchers, I have come to recognize this after years of behaving otherwise. By this standard I have published lies, then - if not many lies, it was only due to my own low productivity. I wish I had operated in a research culture which forbade that.
1 reply 0 retweets 1 like -
Replying to @FrederickGuy4 @bradpwyble and
Within his overall rejection of the exploration/estimation divide, two specific places Shiffrin's advocacy of exploration undermines any claim that statistical "hits" are indeed hits: data cleaning, and the idiosyncratic nature of research settings.
1 reply 0 retweets 1 like -
Replying to @FrederickGuy4 @bradpwyble and
Yes data may need trimming & cleaning. & yes may be impossible to pre-register criteria for doing so. But implications different if code with all trimming steps + raw data are public, vs not.
2 replies 0 retweets 1 like -
Replying to @FrederickGuy4 @bradpwyble and
No mention of transparency in trimming from Sheffrin, but since he follows trimming immediately with his statement of "What is the harm... Scientists are skeptical and know many publications misstate and mislead", I take it that his support for transparency can't be assumed.
2 replies 0 retweets 1 like -
Replying to @FrederickGuy4 @economeager and
If you're going to make such assumptions about his positions and frame your arguments around those assumptions, then there isn't much point in having a discussion.
1 reply 0 retweets 2 likes -
Replying to @bradpwyble @economeager and
I'm just reading what he wrote as well as I can. If you know his work, tell me where I'm wrong in this interpretation.
2 replies 0 retweets 1 like -
Replying to @FrederickGuy4 @bradpwyble and
You cast scientific inquiry as resulting in hits and misses, as if those are the only possible outcomes. If your work consists only in producing p-values then you might have a point. But modelers don’t do this.
1 reply 1 retweet 2 likes -
Replying to @TrishaVZ @bradpwyble and
Not my term - was merely responding to a post which framed it that way. Though, if editors look at p-values, then that is world we work in like it or not, & thus important to be clear about context in which p-values are produced. Else they are lies.
1 reply 0 retweets 1 like -
Replying to @FrederickGuy4 @bradpwyble and
But Rich and I and other modelers don’t work in that world, which is the point.
1 reply 0 retweets 2 likes
Yes. Exactly.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.