# Thread: Author's interpretation of Odds Ratio and Confidence intervals

1. ## Author's interpretation of Odds Ratio and Confidence intervals

Hi all,

Having some difficulty understanding the results of this paper I am including in a review.
It compares the effects of 3 interventions aimed at preventing HIV on couple's relationships. A multivariate logistic regression is used and effect estimates are given (which we are told are odds ratios despite using the beta sign for them in the text which is rather confusing).
When comparing a couple's intervention to a women's only (standard) intervention (CHC vs. WHC) they get an odds ratio of 0.92 for women's control/influence in relationships (Table 3). In the results section they describe the couples intervention as having IMPROVED women's control (top of page 522). In my understanding of odds ratios, an odds ratio of 0.92 would mean that the intervention DECREASED the odds by 1.87 times. I don't understand how they came to this conclusion.

Moreover, the confidence interval crosses 1 (0.02-1.83) which to me means it is NOT statistically significant at 0.05. But I have since seen elsewhere that significance is based on whether it crosses zero, as opposed to 1. Which is it?!

The link to the article is here: https://www.dovepress.com/relationsh...t-article-IJWH

Really appreciate any help at all!

2. ## Re: Author's interpretation of Odds Ratio and Confidence intervals

I have since found an answer to both of these questions. That figure was not an odds ratio.

 Tweet