Module 24 ยท What is coverage (the score)?
Coverage: how much of the outcome does the pattern explain?
- Define coverage as a model-fit measure.
- Keep CNA-coverage and vaccination-coverage straight.
- Set the cov threshold sensibly.
The second score is coverage, an evaluation measure that you should not confuse with vaccination coverage. It answers: of all the facilities that had the outcome, what share does this pattern account for?
Coverage is "how much of the problem this pattern captures"
If 90 of 100 low-coverage facilities fit your pattern, the coverage score is 0.90. A lower score isn't automatically bad. It often just means other roads to the same problem exist, which is equifinality saying hello. You set the threshold with the cov argument, and its default is also 1.
"Coverage" in CNA output is a model-fit statistic. "Coverage" in your program is the share of children vaccinated. Same word, two completely different jobs, which is mildly unfair of the universe. In reports, say "the pattern's coverage score" versus "DTP3 coverage" so nobody gets lost.
# require patterns that are both reliable AND broadly explanatory cna(immunization, outcome = "LowCov", con = 0.9, cov = 0.9)
Mind the two meanings of 'coverage'.
- In your own writing, how will you avoid confusing the two?
- If a pathway has low coverage-score, what might that tell you about other pathways?