My dependent variable is performance of safety task by Health and safety officer ( where NO= 0 is non performance and Yes= 1= performance making variable binary )
and the independent variable is sector the health and safety officer is employed in (this is measured using nine categories of responses numbered from 1-9) each number represent sector for instance 1= agricultural, 2 contraction. 3= manufacturing, 4= communication,5= education, 6= mining,7= wholesaling,8= public administration and 9= transportation.I am using SPSS
One of the Odds Ratio is 786489514775 and another is 2889553982.064.....What is the cause of this extremely large OR? and are they plausible?....how can i correct them?
Thank you
Hi Michael
As Your independent variable is categorical, I guess, You have dummy coded it into 8(?) single variables and calculated an OR for each one of them.
How many cases does Your data set contain? How does the data look, when You compare the outcomes for each group? At first glance I wouls suspect, there might be some groups, that have a very low cell count, e.g. almost only 0 or 1 as an outcome.
Best regards
Thank you Michael Beck ......my data set has 106 cases,its true some some groups have low cell count of 0 and others 1......how can i remedy this? do i reduce the number of categories for the independent variable especially those with small cell values ?....please check attached output extruct
Hi Michael
You indeed have very low cell counts and even empty cells, that's the reason for Your weird ORs. Your sample is rather small for an analysis with an independent variable with 10 levels. Like You suggested, You can collapse the levels into only two or three groups (depending on Your research question on theoretical assumptions of course).
Best regards
- Badges
- Science topic
- Similar topics
- Medicine
- Epi
- Age Groups
- Children