WikiJournalClub talk:Conventions
Are we gonna call it WikiJournalClub, Wiki Journal Club, or some combination of those with variances in capitalizations?
- I think it should always be written as <blink>WiKiJoUrNaLcLuB</blink>
- Damn.
- I'm fond of "Wiki Journal Club", all separate. The CamelCase stuff popular in earlier wikis was, well, popular in earlier wikis and is kinda ugly methinks. --Diberri 23:27, 21 November 2011 (PST)
Stats conventions
1) I like that we are setting conventions for statistics. What is considered "non significant"? If a trial uses testing for P<0.05, then P=0.06 is considered nonsignificant. Also if a trial specifies a P value of 0.78, do we call that P=0.78 or P=NS. At what point do we want to call it P=NS. Do we want to call it NS only if P value is not specified?
2) Also what do we about P values that are "not specified?" I suppose in most cases if the CI includes 1.0, then it is P=NS even if it is not stated, by definition. If a P-value is "not specified", I'm inclined to leave the phrase "P not specified" out because it looks a little redundant to me. I think the absence of the phrase "P not specified" already implies that the P is not specified.
3) Also do we want to set conventions for P-values <0.00000000000000000000000000001? If it crosses a certain boundary, should we truncate it to P <0.0001?
Thoughts? --mlam 12:47, 17 December 2012 (PST)
- My preference is to avoid creating our own arbitrary thresholds, so that we don't introduce our own definitions of statistical significance. Instead, we should simply report the primary data. Addressing your points in order:
- When the numeric P value isn't provided in the text, we use "P=NS".
- When a P value isn't specified at all, we use "P not specified".
- When a P value is unwieldy, we can use scientific notation, like P<10-7.
- --Dave Iberri 15:11, 17 December 2012 (PST)