1

I have an experimental condition (dummy-coded) as an categorical predictor, and one continuous predictor variable (treatment frequency). The dependent variable (Y) is consumer satisfaction enter image description here This is an SPSS output,

My Q: is how to interpret the coefficients and Exp(B) of the categorical Variable and continuous predictor.

For Categorical variable: Is it correct to say " Condition 0 is .34 more effective than Condition 1 in enhancing consumer satisfaction. Or, can I say if the treatment frequency is fixed, Condition 0 compared to Condition 1 will increase the odds of satisfaction by 0.40." ?

For continuous predictor: Is it correct to say " When 1 unit increases in frequency, the satisfaction increase by .15 when the condition is 1. Or, if the condition is fixed to 1, treatment frequency will increase the odds of satisfaction by 0.17." ?

My last question is how to interpret the significant interaction term. Could you help me interpret the interaction term using the coefficient and also Exp(B)?

If you can recommend an published article that describe results from generalized linear model using interaction term, that'd be appreciated. Thank you.

StatK
  • 11
  • 1
  • How is Consumer Satisfaction coded? Is it just binary? This looks sort of like the output I'd expect from a glm with a binomial link function (i.e., a logistic regression). – russellpierce Sep 18 '14 at 18:15
  • the Y was measured weekly as binary "yes" or "no" for 30 weeks. Each person gets the total number of weeks saying YES divided by the total Study Weeks s/he participated. – StatK Sep 18 '14 at 18:17
  • So the Syntax is : GENLIN WeeksYES OF StudyWeeks BY COND (ORDER=ASCENDING) WITH FREQUENCY /MODEL COND FREQUENCY COND*FREQUENCY INTERCEPT=YES DISTRIBUTION=BINOMIAL LINK=LOGIT /CRITERIA METHOD=FISHER(1) SCALE=1 COVB=MODEL MAXITERATIONS=100 MAXSTEPHALVING=5 PCONVERGE=1E-006(ABSOLUTE) SINGULAR=1E-012 ANALYSISTYPE=3(WALD) CILEVEL=95 CITYPE=WALD LIKELIHOOD=FULL /MISSING CLASSMISSING=EXCLUDE /PRINT CPS DESCRIPTIVES MODELINFO FIT SUMMARY SOLUTION (EXPONENTIATED). – StatK Sep 18 '14 at 18:20
  • see: http://maartenbuis.nl/publications/interactions.html – Maarten Buis Sep 18 '14 at 18:45
  • How to interpret an interaction & how to interpret logistic regression are very common questions here. Try reading some of the threads that are already on the site ([here](http://stats.stackexchange.com/q/34636/7290) are 2 [more](http://stats.stackexchange.com/q/65268/7290)). If you still have a question afterwards, come back here & edit your Q to state what you've learned & what you still need to know. – gung - Reinstate Monica Sep 18 '14 at 18:46
  • rpierce, gung, Maarten Buis, Thank you SO MUCH for all your suggestions!! – StatK Sep 18 '14 at 18:52

0 Answers0