You may hit this term in systematic review and meta-analyses. It’s the latest, and important, iteration in systematic reviewing. Instead of collating the summary statistics across the studies included in the review, the authors obtained pseudonymised data from the authors of the original studies and collate these original data for analyses. In principle it yields much richer analyses but inevitably it will have a bias from the variable availability of the original data.
Details #
This is not my expert are but good examples of IPD analyses will use multi-level modelling taking into account that the data are nested within studies. I would hope that IPD analyses will lead to much more nuanced explorations of the issues in the review but my (limited) sense of what I’m seeing so far is that it still falls pretty much into answering the “does this/doesn’t this support x over y” binary questions that dominant systematic reviewing and meta-analyses rather than using the extra statistical power and breadth of data that is (in principle) provided to explore more predictors and look at the diversities of individual participants’ data.
Try also #
Meta-analysis
Multi-level models (MLM)
Nested data
Systematic reviewing
Chapters #
Not discussed in the book.
Online resources #
None currently.
Dates #
First created 21.ii.24.
