SPSS – Multiple Regression; Coefficients not significant?
Dear all,
As part of my bachelor thesis I ask:
How do I interpret the regression if
- Corrected R^2 = .482
- ANOVA significant with .014
- But: All but one coefficient are not significant (already tested for multicollinearity, tolerance and VIF value are OK)
This is a rare disease, and the sample includes 50 patients. The criterion is a questionnaire score, and predictors are descriptive and disease-specific characteristics.
Thank you!
(1 votes)
Loading...
The model can predict the dependent variable (where a model with numerous “descriptive and disease-specific properties” should be somewhat overloaded at n=50). In none of the individual predictors shows a statistically significant relationship, which can easily happen in many predictors and few cases. There is not much to interpret, especially since the exact question for the analysis is not given. There are two different levels.
Thank you. That is, fewer predictors, and two regression analyses would help? (Of course, a coefficient was significant, the remainder was not)
Hypothesis: The questionnaire score is predicted by some descriptive and disease-specific (spinal muscle atrophy) properties. I set up subhypotheses, e.g. “will be predicted by age”, “will not be predicted by gender”
Descriptive: Gender (Dummy), age, highest educational degree (ordinal, probably it would be better to take the metric variable “number of years of education”)
disease-specific properties: severity of the disease, number of genes, motor functions, pain and interpersonal interactions
Pardon that one was statistically significant, I had read. Separate analyses make no sense. One has to ask oneself precisely what one wants to incorporate so many predictors into his model in scientific and/or practical terms. Less is sometimes more.