Blog three of five, in a series by Dr Chrissi McCarthy.
TLDR: A meta-analysis pulls together many studies to reveal the overall pattern, giving a far more reliable picture than any single piece of research. Mor Barak et al. used this method to understand the true impact of diversity and inclusion on workplace outcomes.
Why Use a Meta-Analysis?
There is a lot of research on diversity, and, as we have discussed in earlier blogs, this research presents both positive and negative outcomes of diversity work. For that reason, Mor Barak et al. took an approach called a meta-analysis, which, put simply, pulls together existing work in an area. By doing this, the researcher can determine the overarching story.
This approach provides a much more trustworthy picture than relying on a single study, which might be unusually positive, negative, or flawed.
Step 1: Reviewing the Landscape
There are usually a couple of steps involved. First, review the overall literature to identify whether an overarching story is likely, which can help you determine the types of studies you need to look for. This isn’t about cherry-picking to suit your narrative. Instead, it’s about understanding that there is usually a lot of work out there in the fields, so you need to set some boundaries, else you will spend forever reading and never get the chance to start analysing.
It is like averaging many weather reports to understand the actual climate, rather than relying on a single day’s weather.
Step 2: Setting the Boundaries
Once you have done this, you set your bounds. In this case, Mor Barak et al decided only to select papers that had the following criteria.
- One diversity characteristic
- Work-related
- Human service organisation focused
- Published between 1990 and 2012
- Reported statistics
- Reported the size of the sample
Step 4: Coding and Comparing the Results
They then coded the studies to quickly determine which categories they fell into. Then they converted the statistical results to a common measure so they could make a standard comparison across the studies, because you cannot meaningfully compare percentages, odds ratios, and correlations unless they are translated into one shared format.
Step 5: Checking and Correcting the Statistics
After undertaking some statistical skill work, which corrected errors and checked results against context, they obtained a result that provided a generalisable picture of how diversity and inclusion relate to workplace outcomes.
The Big Picture
In short, they took many similar studies and worked out the main story about the effectiveness of diversity and inclusion work when you put them all together.
