Crucial idea, first letter in our GULP acronym. Means what the word says: are the findings in a report generalisable?
Details #
Crucial idea underplayed, referred to rather tokenistically or frankly completely ignored in many (most?) research reports in our field. When ignored it feeds into the idea “these are the facts [for everyone, everywhere]”.
Addressing generalisability well is complicated and has to start with “where and for whom is generalisability important?” Then the question becomes “to what extent is it reasonable, plausible, to argue that these findings generalise there, to those people?” Addressing that question, once you have chosen your area of important possible generalisability, is complex. The key issues include the following.
* The dataset size: single case studies and short qualitative case series generally don’t aspire to high and definite generalisability but they can have conceptual generalisability: a single report can change how we see a phenomenon or uncover a new one.
* With quantitative data the dataset size influences how precisely we have estimated a summary statistic looking at it as indicating a population value. Be wary of generalisation from work that can only very imprecisely have estimated any population value because of a small sample. (Which is not to say that small samples, particularly of rarer phenomena, can’t be valuable.)
* Are the measures and methods going hit issues of generalisability to routine practice? The language and cultural issues are obvious but just spending 90 minutes filling in questionnaires for a research study may bear zero relationship with routine service delivery.
* Was the participant group similar to the people to whom you would like to generalise? How might its formation impact on generalisability?
These are just good starting questions. The key is to recognise that absolute generalisability is a fantasy, what matters is to be wise about what can be argued and to look at what further work, formal research or routine data collection, can tell us more about the generalisability of any findings.
Try also #
Confidence intervals
Estimation
Generalisability
GULP
Sample frames and sampling
Chapters #
Idea runs throughout the book but is key in Chapter 3.
Online resources #
None?
Dates #
First created 28.xi.23.
