2
$\begingroup$

Hello can anyone kindly guide me about the correct way to interpret adjusted odds ratios.

I am working on a project (in SPSS) on women's autonomy and domestic violence. When I take the (simple) odds ratios then women's non-participation in health decisions, mobility and household decisions is significantly related to experiencing domestic violence. (women's age, education level and wealth quintile are also significantly related to domestic violence)

Yet, when I control for women's age, education level and wealth quintile, only the non participation in household decisions is now significantly related to experiencing domestic violence. How do I interpret the change in results since the other two types of decision are now no longer significantly related to experiencing domestic violence?

Improve this question
$\endgroup$
1
  • $\begingroup$ This is a FAQ (or a combination of FAQs). For the interpretation of odds ratios, see here; for how p-values change when you add variables (ie become non-sig), see eg here; for more on the underlying issue, see here & here. $\endgroup$
    – gung - Reinstate Monica
    Mar 8 '15 at 14:56
2
$\begingroup$

First, don't emphasize the difference between significant and not-significant. Look at effect sizes.

Second, the simple odds ratio is (like it says) a ratio of odds. So, the odds of a woman participating in health decisions is significantly different for women who experience domestic violence and those who do not.

The adjusted odds ratio is the same thing, only it holds the other variables (age, education, wealth quintile) constant.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Thank you for the answer. In terms of effect sizes, the (simple) odds ratios for experiencing domestic violence are 1.192, 1.269 and 1.277 for non-participation in the three types of decision making. Are these moderate or weak effects? $\endgroup$
    – NH2015
    Mar 8 '15 at 14:10
  • $\begingroup$ Whether they are weak or strong depends on the field and its standards. What have other people found for similar questions? $\endgroup$
    – Peter Flom
    Mar 8 '15 at 14:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.