Consider a prespecified secondary RCT endpoint that is missing 50% of the time. Is it okay to not report it *at all*?
-
-
Replying to @venkmurthy
Question. Does reporting "50% are missing, therefore we decided not to report statistics from these data" count as reporting? This impacts my answer a lot.
1 reply 0 retweets 1 like -
Replying to @NoahHaber
That is not reporting the result, but it is providing an explanation.
1 reply 0 retweets 1 like -
Replying to @venkmurthy
Great, then I am a firm yes. There do exist situations in which you would not report the data with that much missing (although you should make those data available if possible). E.g. If the mechanism of the missing data is highly endogenous, it can be more misleading to report.
3 replies 0 retweets 3 likes -
Replying to @NoahHaber @venkmurthy
I'm actually in favor of this approach. I think it's far more transparent to report the fact that it's missing, why it's missing, and why you don't think the numbers are usable than to include them without all that explanation
1 reply 0 retweets 2 likes -
Replying to @GidMK @NoahHaber
Exactly. I would go so far as to say, its okay to not calculate sig tests (if you believe that is right), report alongside sensitivity analyses and to relegate to a supplement with lots of caveats and limitations around it. But it isn't okay to not do it at all.
1 reply 0 retweets 1 like -
Replying to @venkmurthy @GidMK
Yeah, this is another scenario. If the you powered the study and set your alpha such that significance/non-significance corresponds to an actual decision, your data are now underpowered relative to that decision, and it might be fair to scrap the result.
1 reply 0 retweets 1 like -
Just kidding no one actually does this.
#plessthanpointfivelife1 reply 0 retweets 2 likes
To be fair, there's a massive disincentive to doing it. Might be better for SCIENCE, but it's definitely worse for individual researchers
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.