I am doing a meta-analysis using the Open Meta software. The findings being meta-analyzed are simply the proportion of those with disease X who also have disease Y for each study. I am given an estimate of the overall proportion, the lower and upper bounds of the CI, and the significance value. I am not really sure how to interpret the significance value in this case. I have attached a photo. 
$\begingroup$
$\endgroup$
5
-
$\begingroup$ Rather than being necessarily a pure statistics question, your inquiry may simply call for checking the Open Meta documentation to see what null hypothesis is being tested via their version of this procedure. (You won't find many instances of "binary random effects model" on this Stack Exchange site.) But at any rate I'm guessing what will matter to you is not any p, not any significance test result, but the 0.122 and the 0.012. $\endgroup$– rolando2Commented Mar 8, 2017 at 22:17
-
$\begingroup$ @rolando2 Thank you for your response. Their help section is not helpful. But yes, you are correct those values will be more useful to me. $\endgroup$– aspire94Commented Mar 8, 2017 at 23:25
-
$\begingroup$ @aspire94 what is the objective of your meta'analysis. $\endgroup$– user10619Commented Mar 23, 2017 at 10:39
-
$\begingroup$ Shall apprciate if I get a brief about binary random effects model. $\endgroup$– user10619Commented Mar 23, 2017 at 13:27
-
$\begingroup$ The question is vague. need more details. $\endgroup$– user10619Commented Mar 23, 2017 at 13:30
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
4
it is significant.
reject the null hypothesis, so there is a significant relationship with your variables
-
$\begingroup$ Welcome to Cross Validated! Please take a moment to view our tour. $\endgroup$– TavrockCommented Mar 8, 2017 at 18:07
-
$\begingroup$ I understand it means it is significant. But WHAT is significant since I am only looking at the proportion in each meta-analysis? $\endgroup$– aspire94Commented Mar 8, 2017 at 18:10
-
$\begingroup$ What null hypothesis does it reject? Note that there is no moderator variable supplied for it to relate to either. $\endgroup$– mdeweyCommented Mar 8, 2017 at 18:27
-
$\begingroup$ the null hypothesis is there is no relationship $\endgroup$ Commented Mar 10, 2017 at 15:15
$\begingroup$
$\endgroup$
The only thing it can be testing is whether the estimated proportion could be zero which it clearly is not. Whether that hypothesis makes any sense in your scientific situation is up to you but I would have thought it rarely, if ever, does.
If the documentation for this program is so poor that it does not tell you I would suggest you change to another program as it does not seem state of the art.
