Bayesian meta-analysis

Stanislaw

Hi there, I have a few questions regarding meta-analysis in JASP. Is there anyone who could help me with it?

EJ

Ask away, that's what the forum is for

E.J.

Stanislaw

Hi E.J.

thank you for your answer. I appreciate it.

First question refers to the ES.

1. What kind of ES should we think about? There are many different types of ES measures(Cohen's, Hedge's, eta squared, odds ratio, etc.) and they have different values and ranges.
2. Do we have to convert all ES to one type only?
3. If so, what is recommended?

Keep well

stan

EJ

Yes, in a meta-analysis you've have to use the same effect size measure for studies, or else you're comparing apples and oranges. What is recommended depends somewhat on the designs. For 2x2 contingency tables the log odds ratio is a popular measure. For t-tests, either Hedges g or Cohen's d. I've heard that there are arguments for Hedges over Cohen, but it likely does not matter much. Some people convert every ES to a correlation measure (I am not sure what the opinion of the field is on this practice). It is always good to study some meta-analyses in your own field and see what people do.

E.J.

Stanislaw

In one that was done previously (there are not to many of them on the topic) Cohen's d and F and t-ratio was used. "Effect sizes were calculated using Cohen's d as an index (..). When these statistics were not reported, the F and t ratios were used when available."

I am not sure if I understand what they mean by using F and t-ratios. Is there another way of calculating ES using only F and t values?

stan

EJ

The F and t values depend on sample size, whereas effect size does not.

E.J.

Stanislaw

Hi EJ

I have a question how can I deal with dependency in meta-analysis in JASP? I have a few outcomes and want them to be included in the analysis. How can I avoid dependency?

Only to mention that RevMan - one of the most popular meta-analysis software, does not have it.

The problem, as I read, is ubiquitous.

Perhaps see here:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4191743/

thank you

Stan

Stanislaw

Clarification: a few outcomes from one study.

EJ

I passed this on to our expert