I interpret these data to mean that hybrid courses have a 34% increase in the odds a student will be successful (DV) when taking a ZTC course, above that which traditional courses have. Students taking a traditionally delivered course that is ZTC have a 77% increase in the odds that they will be successful.

Would that mean then that students taking hybrid ztc courses are 2.03 times more likely to be successful than students taking traditional ztc courses (ztc)?

I understand you have to interpret the interaction effects in relation to the main effect... without the graph I would have interpreted this differently... so I am stumped. Please help me know if my interpretation is on track.

I added a 2nd model which is causing me to be so confused. The data are similar but the graph looks different than the first model I talked about here. Now I am just baffled at how to interpret. See attached both models.

Hello Lisa,

Unfortunately, I can't quite deduce what it is you're trying to accomplish, and why you have explored two apparently distinctive models. (And, what is ztc?)

Would you kindly explain what your specific aim (the particular research question/s you'd like to address), as well as what each of the variable names represent (including the outcome)? (And the motivation for trying "Test 2"?)

If you do, and show the full output, I suspect that you'd be able to get more focused interpretive advice.

Good luck with your work.

Hi David,

The research question is does ZTC have a greater effect on student success rates for different groups? ZTC being the intervention. The different models test for either different course types: hybrid, online, or traditional and then FT or PT status of students. Not sure what you were suggesting in regards to the full output...I thought I included everything.

Thanks for any help you can provide.

Hello again Lisa,

It's not clear why Test 1 result list omits ORs/p-values for so many variables, and ditto for Test 2...that's one of the reasons for asking for clarification.

Looking at "Test 2" output,

1. Having ztc significantly improves odds of success;

2. Online course significantly improves odds of success (vs. traditional);

3. Hybrid course significantly reduces odds of success (vs. traditional);

4. ztc combined with online significantly reduces odds of success (more than you would expect from the related "main effects" directions, _and_ relative to traditional);

5. ztc combined with hybrid significantly increases odds of success (more than you would expect from the related main effects directions, _and_ relative to what happens with traditional classes).

If you look at the slopes of the Test 2 output plot, you'll see that ztc appears to have the sharpest increase in likelihood of success for the hybrid condition (but that the hybrid condition is still lower in odds of success than either of the other two methods). That is completely consistent with the significance test outcomes.

As well, ztc has less influence on success likelihood for online class than for traditional class. Again, consistent with the test results.

Good luck with your summary and presentation.

In addition to David Morse 's points, interpreting interactions in logistic regressions is often aided by plots the predicted probabilities with uncertainty intervals.

Daniel Wright my attachment actually has plots shown.

I believe I have this figured out now.... after adding the coefficients and exponentiating it, I get a new odds ratio of 9.06. I take this to mean that Students who are FT and take a ZTC course (the intervention/var of interest) are 9 times more likely to be successful in their course, compared to students who are PT and take a non-ZTC course.

Wouldn't the net effect of just taking a ZTC course and being a PT student be the 1.9 odds (i.e., (B0 + B1)exp)?

If this is all correct, then what is it that people actually submit when they publish? Is it the new calculated odds or is it the computer generated output with Bs, ORs, and pvalues? I do not want to confuse my readers by discussing the calculated odds and reporting what is generated in tables.

Appreciate your help with this. See attachment with output data and calculations. David Morse

Hello again Lisa,

Last question first:

Minimally, it's customary to furnish a table of all IVs in the model, their estimated regression coefficient, OR, and p-value for each, as well as a test of the overall model (usually as a chi-square with k df, where k is the number of IVs).

First question (referring to most recent post, showing "Test 1" results,

If ZTC is coded 0/1 (for no/yes)

If Full-time status is coded 0/1 (for part-time/full-time)

Then the first interpretation is: (a) ztc confers an advantage (as regards the estimated odds of success in course) vs. non-ztc; (b) being full-time student confers an advantage (for same DV); and (c) because there is an interaction, those who are both ztc _and_ full-time status are predicted to have even higher odds of success than predicted by the main effects alone.

Here's how to use the values to generate a "what-if" comparison, based on the model as shown:

For cases that are ztc and full-time (and each variable coded 0/1):

log-odds of success = 0.353 + 0.306 + 1.135 + 0.195 = 2.204

odds of success = exp(2.204) = 9.06

For cases that are neither ztc nor full-time:

log-odds of success = 0.353

odds of success = exp(0.353) = 1.42

OR of the two groups: 9.06 / 1.42 = 6.4

Good luck wrapping this one up.