Timeline for How to compare contingency tables for a specific pattern?
Current License: CC BY-SA 3.0
16 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 18, 2015 at 20:28 | vote | accept | Fabio | ||
| Jul 16, 2015 at 22:05 | comment | added | Fabio | Also, MCC for example is described as giving a much more equilibrated value (respect to other 2x2 table statistics) regarding to the two rows, and seems to especially work well with classes of different size. At least in the few references I've seen it used. | |
| Jul 16, 2015 at 22:02 | comment | added | Fabio | True, I'm being ambiguous about the third condition. It was in fact one of the sub-questions in the original question: how do I take effect size into account in a statistically sound way? Your formula does include that, but I cannot use current research for the cutoff threshold. For example, with fisher test I could use a "reasonable" p < 0.05. With MCC, I can use Pearson's correlation "reasonable values", such as > 0.4 is a moderate correlation. ERD is highly (linearly) correlated to MCC (0.95 r) so I could use the same. With an hand-crafted method, how do I choose a cut-off threshold? | |
| Jul 16, 2015 at 15:21 | comment | added | John | You're going to have to figure out the distribution of scores for whatever measurement and a constrained one will have tail issues and maybe not allow some math (e.g. correlations don't add). It would be nice to have probability density plots. If I was you I'd next compare ranks of the measures and see if what measure coincides with how you interpret good and bad patterns. (and given your figures probably do stats on the ranks in any event) | |
| Jul 16, 2015 at 15:19 | comment | added | John | Your original two measures don't take into account your third condition (two cells greater than 0). As far as I can see they only take into account the proportion on a row which would cover the first two conditions. My formula includes the third condition. You need to decide on the importance of that. And further, artificially constraining your measurements between 0 and 1 is almost never a good idea. Have you really thought about why you're doing that? | |
| Jul 16, 2015 at 12:44 | history | edited | Fabio | CC BY-SA 3.0 |
added 972 characters in body
|
| Jul 15, 2015 at 19:45 | history | tweeted | twitter.com/#!/StackStats/status/621405204206997504 | ||
| Jul 15, 2015 at 18:42 | answer | added | John | timeline score: 1 | |
| Jul 15, 2015 at 15:00 | comment | added | Fabio | In the original answer I spoke of both "good tables" and "bad tables", and differentiated the cases. This got a bit confusing. The ones respecting the property A >> B are "perfectly good" tables - but I'm not looking for those. In the above, there is no ambiguity: those tables are the ones I need to identify and order by "extremeness", following the pattern I specified. I hope it's clearer. Thanks. | |
| Jul 14, 2015 at 22:34 | comment | added | John | I thought before you also said that A should be larger than B. | |
| Jul 14, 2015 at 13:13 | comment | added | Fabio | I've tried to shorten it as much as I can. Thanks for the tip. | |
| Jul 14, 2015 at 13:12 | history | edited | Fabio | CC BY-SA 3.0 |
rewrote to be as concise as possible and to present only one aspect of the problem. Thanks John for the tip.
|
| Jul 10, 2015 at 14:40 | comment | added | John | This looks like two questions so it's broader than is usually desired here (moreso than you realize). Edit this so that it's just about how to solve your fitting problem. If, after you get an answer you still want to ask about p-values as measures of fit then do that after. And also edit the problem because there's lots of superfluous information that's in some ways a little misleading. All you need is the contingency table and the statement A >> B is good but it must have z >> 0. That would satisfy the whole thing I think. Could x be small? Anyway, be clearer on what you need to fit. | |
| Jul 9, 2015 at 15:07 | history | edited | Fabio | CC BY-SA 3.0 |
cleanup
|
| Jul 9, 2015 at 11:54 | review | First posts | |||
| Jul 9, 2015 at 11:55 | |||||
| Jul 9, 2015 at 11:51 | history | asked | Fabio | CC BY-SA 3.0 |