2
$\begingroup$

I have data from a questionnaire study structured like so:

  • Age - Ordinal (18-24, 25-34, 35-44, 45-54, 55+)

  • Gender - Nominal (Male, Female)

  • AnxietyType - Nominal (Self-diagnosed, Professionally diagnosed)

  • AnxietyYears - Scale

  • ChronicPain - Nominal (No, Yes)

  • Response - Ordinal (Strongly Agree, Agree, Neutral, Disagree, Strongly disagree)

I am using SPSS to run an ordinal logistic regression with 'response' as my dependent variable and the other 5 as my independent variables.

When putting the data into SPSS I have coded it as follows:

  • Age - (18-24, 0) (25-34, 1) (35-44, 2) (45-54, 3) (55+, 4)

  • Gender - (Male, 0) (Female, 1)

  • AnxietyType - (Self-diagnosed, 0) (Professionally diagnosed, 1)

  • AnxietyYears - Scale

  • ChronicPain - (No, 0) (Yes, 1)

  • Response - (Strongly Agree, 1) (Agree, 2) (Neutral, 3) (Disagree, 4) (Strongly disagree, 5)

When I run the regression, this is my output with a significant result highlighted in yellow.

enter image description here

From what I've read and understood about interpreting the results of an ordinal logistic regression, this is saying that:

"The odds ratio of being in a higher category of the dependent variable for males versus females is 2.244" which is saying that males are more likely to agree more strongly than females.

However, when I create a graph looking at the split of responses between males and females it shows that females are actually more likely to agree more strongly than males. enter image description here

I would be grateful if anyone could help me to understand what I'm doing wrong - either in my modelling or my interpretation.

$\endgroup$

1 Answer 1

4
$\begingroup$

I believe you are just interpreting the response variable in the opposite direction than it should be.

""The odds ratio of being in a higher category of the dependent variable for males versus females is 2.244" which is saying that males are more likely to agree more strongly than females. "

It should be interpreted that males are actually more likely to disagree more strongly compared to females.

Because your response variable is coded from 1-5 with strongly agree at the low end and strongly disagree at the high end, an increased likelihood of having a higher category corresponds to an increased likelihood of disagreeing. Therefore your plot shows the correct output.

You can reverse the coding of the response variable to get the opposite interpretation (although your estimates will change).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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