exp(B) interpretation multinomial logistic regression - multiplicative vs. additive

[natalia malancu]

hi guys!

this is a doubt I've been having for some time now. the way i've been taught to interpret the RRR resulting from mlogit is multiplicative: x times as high RR/ x times more likely to be in category A than in category B. however, often time if x is a small figure- say a factor of 1.04- people interpret it as a 4% higher RR of being in category A than in B. moreover, quite of number of tutorials out there put forth this interpretation. is that correct?

cheers,
natalia

[Maarten Buis]

It is correct: if you multiply a number by 1.04, than it increases by 4%.You can move from one to the other using (RRR-1)*100%.

I suspect that part of the confusion is the distinction between a percentage change and a percentage point change. If we start with a baseline value of 1% and change by 1 percentage point, then the result will be 1 + 1 = 2%. If we change the baseline value by 1%, the result will be 1 * 1.01 = 1.01%. To be clear, the interpretation is correct if we talk about a percentage change not percentage point change. (The latter would not make sense anyhow as relative risks aren't expressed in terms of percentages.)

[natalia malancu]

Thank you, Maarten. That simplifies a lot my phrasing in the future.