1/2: Consideration of intent and utility (psychology)may be illuminating for the issue of change-from-baseline (CfB) vs models for response conditional on baseline (mRcB). Humans intuitively construe effect as systematic change. But this intuition may make mischief for analysis.
-
-
-
2/2: Clearly the composite variable CfB is statistically inefficient (noisy)and confounds outcome with antecedents. If the ultimate objective of the evidence is to advise instrumental action, precise expression of outcome (what is actually observed) with mRcB is preferable.
- Još 6 drugih odgovora
Novi razgovor -
-
-
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
Frank's post here has some of his thinking on the subject:https://www.fharrell.com/post/errmed/#change …
-
There are admittedly some situations where using "change scores" versus "final score with adjustment for baseline" are actually identical
- Još 2 druga odgovora
Novi razgovor -
-
-
I cite this one in reviews a lot: https://deepblue.lib.umich.edu/handle/2027.42/73008 … Good for non experts and points to some solid work in the area.
-
That's my go-to ref too, Rich has given this topic a lot of thought over the years.
Kraj razgovora
Novi razgovor -
-
-
It's not always clear cut.
@MariaGlymour offers some great advice here and uses causal diagrams and arguments to see when one approach is appropriate over the otherhttps://pubmed.ncbi.nlm.nih.gov/15987729-when-is-baseline-adjustment-useful-in-analyses-of-change-an-example-with-education-and-cognitive-change/ … - Još 1 odgovor
Novi razgovor -
-
-
If you are concerned about regression to the mean (and most of us should be!), it is also possible to correct changes scores for this, instead of covarying baselines: https://link.springer.com/article/10.1007%2Fs10654-013-9845-4 … (formula inspired by the telomere literature, but generally applicable)
-
But the direct approach (ANCOVA) is much preferred, and it works for ordinal outcome where the difference between ordinal Y and ordinal X is no longer ordinal.
- Još 1 odgovor
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.