Timeline for What does observation mean in meta-analysis?
Current License: CC BY-SA 3.0
16 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 23, 2017 at 12:39 | history | edited | CommunityBot |
replaced http://stackoverflow.com/ with https://stackoverflow.com/
|
|
| Mar 18, 2016 at 6:59 | comment | added | Karthik Sharma | Thanks, I will look into it. I have one last doubt, regarding missing values. How to replace missing values with mean? I mean I did google, but I got confused regarding it, so I am asking for some explanation. | |
| Mar 16, 2016 at 13:09 | comment | added | Giuseppe Biondi-Zoccai | I recommend you to use the meta R package: metacont for continuous outcomes and metabin for dichotomous variables. You find plenty of examples here: cran.r-project.org/web/packages/meta/meta.pdf. | |
| Mar 16, 2016 at 4:33 | comment | added | Karthik Sharma | Okay. What are the best tools you can advise on to do it? | |
| Mar 15, 2016 at 23:13 | comment | added | Giuseppe Biondi-Zoccai | I still don't follow. Break down your goals into several simple steps. First identify 1 dependent variable, and then 1 independent variable. You need a precision estimate for the dependent variable as well (might be inverse of sample size, but the standard error of the dependent variable would be better). Then run this simple bivariate approach separately for all your potential moderators. If you want then to do meta-regression with multiple moderators simultaneously, you theorically can do it, but usually it is fraught with overfitting. | |
| Mar 15, 2016 at 12:43 | comment | added | Karthik Sharma | TF1, TF2, OF1, OF2, EF1 EF2 are my moderators. The primary outcome is to do hypothesis checking by these moderators. Like TF1 moderates the relationship between IT variables and IT Adoption. For calculating precision feature I have issue with sample size. As each paper has around 10 studies as I have multiple estimates from same study which are correlated. So how do I solve it is my question? | |
| Mar 15, 2016 at 8:42 | comment | added | Giuseppe Biondi-Zoccai | I am not sure what you need or want to do. Give me some additional context. Also clarify better what is the primary outcome, which are the secondary outcomes, and which are the moderators. You need a precision feature (eg standard error) for each unit of the analysis for the outcomes, but you may skip it for the moderators. | |
| Mar 14, 2016 at 22:26 | history | edited | Giuseppe Biondi-Zoccai | CC BY-SA 3.0 |
Integrated with additional content
|
| Mar 14, 2016 at 22:21 | comment | added | Giuseppe Biondi-Zoccai | If you don't have the standard errors of the correlation coefficients (most likely your outcome of interest), then you have to back calculate them relying on the sample size. See edited version of my entry. | |
| Mar 14, 2016 at 7:34 | comment | added | Karthik Sharma | I have edited the image, can you tell me. The effect size and size of the study are given. I am still stuck at entering values into my database. Like my only confusion is for the correlation values. How do I enter into the sheet? To make it simple I have totally 6 factors to be considered. So where should I enter each of the correlation value in the table is what I am confused about. | |
| Mar 14, 2016 at 5:29 | comment | added | Karthik Sharma | Once we have the effect sizes, how do we calculate the measure of precision or weighing factor? | |
| Mar 13, 2016 at 22:29 | history | edited | Giuseppe Biondi-Zoccai | CC BY-SA 3.0 |
Improved accuracy of my entry (focusing on precision rather than effect size)
|
| Mar 13, 2016 at 22:27 | comment | added | Giuseppe Biondi-Zoccai | I have edited my entry. The effect of interest might indeed be there as correlation factor. What is missing is a measure of precision / weighting factor (e.g. its standard error). | |
| Mar 13, 2016 at 14:41 | comment | added | Karthik Sharma | Effect size is the correlation i.e r value. Correlation value is the effect size right? | |
| Mar 13, 2016 at 14:22 | comment | added | Karthik Sharma | Regarding the example its a sample help I got. Its random structure and I am confused how to enter data or the value of r into my table. Observation means size of the sample right? | |
| Mar 13, 2016 at 13:43 | history | answered | Giuseppe Biondi-Zoccai | CC BY-SA 3.0 |