When you say "weighted mean difference",  I assume you mean the "raw mean difference". All study effect sizes are weighted in a meta-analysis regardless to whether they are standardized mean differences, raw mean differences, odds ratios, risk ratios, correlations, etc.. The raw mean difference is preferred when all studies use the same outcome (a continuous one) and unit of measure. For example, studies on the effect of a drug on blood pressure will almost all use mmHg as the unit of measure and if they don't, the reported value can be easily converted to mmHg. On the other hand, the standardized mean difference is used when the studies don't use the exact same outcome measure. An example would be if you were interested in examining the effect of something on college board exam scores. You have the SAT and the ACT as the two possible exams but they have greatly different scales. Here you are forced to use the standardized mean difference.

For interpreting the standardized mean difference value, Cohen developed some benchmarks about 30 years ago and a lot of people still use them. Standardized mean differences of 0.2, 0.5, and 0.8 are equated to effect sizes of small, medium, and large. However, these labels should probably be specific to the area of research.

Very little to add to Gordon's excellent answer above. I'd just like to add a strong endorsement of the advantage of reporting anything in meaningful units of the original measure wherever possible. In general Standardised differences are hard to interpret whereas the original units generally have direct meaning: weight, height, time etc. By standardising them we render them hard to interpret. there does seem to be a tendency to reported SMD in preference even when means of the original unit could be used. There is no advantage. The only proviso is when the meaning of units on the original scale are hard to interpret - e.g. a quality of life 'score'. Here it could be that there is marginal advantage in reporting a SMD as they can be equated to standard effect sizes whereas a difference of (say) 2 points has no clear meaning. But still I'd try to stick to the original units and find some way of giving them meaning, for example if there is evidence that a 2 point difference equates to the change from immediate post surgery to full recovery...

Peter

Dear Gordon L Warren  and 

Peter Donald Griffiths,

Thank you very much for yours summarized and helpful answers

Useful links from Cochrane Handbook for Systematic Reviews of Interventions.

http://handbook.cochrane.org/chapter_9/9_2_3_1_the_mean_difference_or_difference_in_means.htm

http://handbook.cochrane.org/chapter_9/9_2_3_2_the_standardized_mean_difference.htm

Can we use SMD if some studies report raw data and some report log transformed data? 

The mean difference (more correctly, ‘difference in means’) is a standard statistic that measures the absolute difference between the mean value in two groups in a clinical trial. It estimates the amount by which the experimental intervention changes the outcome on average compared with the control. It can be used as a summary statistic in meta-analysis when outcome measurements in all studies are made on the same scale.

Aside: Analyses based on this effect measure have historically been termed weighted mean difference (WMD) analyses in the Cochrane Database of Systematic Reviews (CDSR). This name is potentially confusing: although the meta-analysis computes a weighted average of these differences in means, no weighting is involved in calculation of a statistical summary of a single study. Furthermore, all meta-analyses involve a weighted combination of estimates, yet we do not use the word ‘weighted’ when referring to other methods.

The standardized mean difference is used as a summary statistic in meta-analysis when the studies all assess the same outcome but measure it in a variety of ways (for example, all studies measure depression but they use different psychometric scales). In this circumstance it is necessary to standardize the results of the studies to a uniform scale

before they can be combined. The standardized mean difference expresses the size of the intervention effect in each study relative to the variability observed in that study.

Hi.

Can I combine two results, one in SMD and another in mean difference???

Hi Dhara Shah , you can use both. Look this paper Article A meta-analysis on the effect of dietary application of exog...

Thank you so much for great explanation

I am actually writing an assignment on this topic.

Thank you all for your contributions.

However, I am still wondering if there's more to their differences? Can't seem to find anything else 😕