New answers tagged

0

If the dependent measure has multiple categories that you want treated as ordinal (like low, medium, and high compliance) then look at the PLUM procedure in IBM SPSS Statistics in the Base module.


0

After thinking about this more, the best notation for adjacent accuracy that I have come up with is $$\mathrm{AA}(y,\hat{y}) = \frac{1}{N} \sum_{i=0}^{N-1} 1(\hat{y}_i \approx y_i) ,$$ followed by a statement that $\hat{y}_i \approx y_i$ holds true if the predicted class $\hat{y}_i$ is off by no more than one class from the true class $y_i$. The ...


1

I don't see any reason to use option B. There's no reason why a 5-point scale should be more valid than a 3-point one, and re-scaling is somewhat arbitrary because you could also put point 2-4 in the medium category. What I cannot guarantee is the validity of option A, but this doesn't have to do with whether your scale has 3 or 5 points (and is therefore ...


1

1) Is method 1 or method 2 better? In my opinion this depends on the background and what you want. Method 1 gives more differentiated information, method 2 focuses on the decision problem and forces people to not sit on the fence (by giving 2 or 3 versions the same score). This is a subject matter/psychological issue, not a statistical one. 2) In both cases ...


Top 50 recent answers are included