Hello!
I ran a logistic regression with a categorical IV (control, low experimental, high experimental) and a dichotomous DV (no/yes).
RQ: Were participants in the low or high experimental condition more likely to select "yes" and request further information on the topic than those in the control?
Attached is the output in PDF for and I am having difficulty interpreting it.
Black 0 shows that a model without variables would predict that most participants would select no—which is not a surprise because there were more participants who selected no.
Then, the Omnibus Tests of Model Coefficients table shows that once the variables are entered, they are a better predictor compared to the null model (p = .009). The pseudo-R2 is very, very small, though.
The classification table under Block 1 then shows the same ratio of observed to predicted values as Block 0. Doesn't this mean that the addition of the IV did not help predict the no vs yes DV?
Finally, in the Block 1 variables in the equation table, the row labeled condition (2), the high experimental condition, is statistically significant (Wald = 9.19, p =.002, Exp(B) = 1.72).
How do I interpret these findings? Did those in the high experimental condition have higher odds of selecting "yes" compared to those in the control condition?
Thank you so much!
What do you think is confusing?there is an effect, which appears to be different from zero, but there is a lot of uncertainty in predictions.
Steve s.
thank you for your help, that does make sense.Kelvyn Jones the part that was confusing was whether the addition of the variable was a meaningful predictor compared to a model without the variable. The variable was statistically significant but the classification table was the same with or without the variable. However, I believe I understand it now.
I am grateful for all input—thank you.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS