Multinomial Logistic Regression- finding the probability of my response variable happening

[applecross]

Hello, I am a political science university student so am not very familiar with R, or many of the technical terms, so apologise if the formatting or anything here is weird. Basically, I have made a MLR model that looks about right (I.e., coefficients and p values), with the correctly recoded explanatory variables (for example, having voted for a specific party before as a dummy variables etc). Here is my issue: One of my questions on a problem sheet is asking for the probability(and thus the odds) of my response variable happening in the event of specific cases in my explanatory variables (I.e., if a person polled rates the EU a 2 out of 10, a politician 7 out of 10, and having voted for a party before). All of these explanatory variables the question is asking for are included in my model. How can I answer the question? Can this be done through code, or do I need to manually calculate it? If the latter, how do I do this? Thank you all for the help, I'm new to this so it is much appreciated.

[technocrat]

See the FAQ: How to do a minimal reproducible example reprex for beginners and also the Homework FAQ.

As a polisci major, you should install the {pscl} package and take a look at its vignette.

I have a brief introduction to the fundamentals of logistic regression. One of the keys is that the odds ratio and probability are fundamentally different. Probability is lower bound by 0, and upper bound by 1. An event can not have a probability of 1.5. Odds ratios are lower bound by 0, but not upper bound.

This S/O post is also helpful.

[phil_hummel]

You need to remember that the logit transform is the logarithm of an odds ratio. So when you do the multiplication of the variables times the coefficients you still need to undo the logit transform to get back to the odds ratio.