Relative risk ratio

[Adeola Oyewale]

how to interpret relative risk ratio in multinomial logistic regression. Also, what is the difference in using odd ratio or RRR to interpret in multinomial logistic regression

[Marcos Almeida]

The multinomial logistic regression presents the ratio of RRs (hence RRR), whereas the logistic regression presents the ratio of odds (hence OR). The interpretation is very similar, i.e, for a one-unit change in a given predictor, the RRR of one level of Yvar (compared to the base level and adjusted for the other predictors) will change by # .

[Adeola Oyewale]

thanks for the information, pls I want to ask how can one intrepret rrr value that is bringing out large values. e.g 945971.8, 1283833.

[Leonardo Guizzetti]

thanks for the information, pls I want to ask how can one intrepret rrr value that is bringing out large values. e.g 945971.8, 1283833.

Those values are non-sense and should not be interpreted. It means that Stata cannot make a reliable estimate of that effect. You might consider reducing your model or reconsidering the model specification.

[Marcos Almeida]

The question in #3 in different from the question in #1.

[Rupa Jose]

If you wanted to interpret a RRR as a percent and you have a RRR = 2.0 OR greater what is the most appropriate way of stating things?
For example say my outcome is "order take out" vs. base group of "cook from home" and I see that the RRR for "urban" (0=rural) is 3.00...

Would a correct interpretation be: Individuals living in urban communities are 200% more likely than those living in rural communities to order take as opposed to cook from home? Or does correct phrasing require it in terms of "risk" even if it reads a bit odd?