Ordinal Logistic Fit/Marginal Prediction/CI

JudithS — Thu, 08 Jun 2023 16:36:24 GMT

I performed an ordinal logistical fit of my data then selected “Save > Save Expected Value” option from the red triangle menu, which adds a column of values in my data table labeled “Ord Expected”. I believe the added values are marginal predictions, that is estimated Ys given my Xs. First, is this true? If so, second, how can I get the confidence bounds about the expected Ys?

Re: Ordinal Logistic Fit/Marginal Prediction/CI

Victor_G — Mon, 13 Feb 2023 17:12:29 GMT

Hi @JudithS,

Welcome in the Community !

You'll find an answer about what is "Expected Value" here : Options for Ordinal Fits (jmp.com). It's a linear combination of the fitted probabilities of the class : Expected value = prob(1)x1 + prob(2)x2 + prob(3)x3 + ...

I have reproduced the calculation on an exemple dataset "Salt in Popcorn" with the formula for Expected Value in the column "Calculation Expected Value" :

:"Prob[1]"n * 1 + :"Prob[2]"n * 2 + :"Prob[3]"n * 3 + :"Prob[4]"n * 4 + :"Prob[5]"n * 5

I don't know if it's possible to have confidence intervals for the Y's, but you can save probability formulas for each class, as well as quantiles, and look for confidence intervals for the model's parameters.

I hope this first answer will help you,

topic Ordinal Logistic Fit/Marginal Prediction/CI in Discussions

Ordinal Logistic Fit/Marginal Prediction/CI

Re: Ordinal Logistic Fit/Marginal Prediction/CI