Dear all,
I have two experimental groups (failure severity:high vs. low and received word of mouth: positive vs negative) as well as a variable called brand image. I want to examine interaction effect of received word of mouth on the relationship between failure severity and brand image. So, I ran univariate analysis. My results indicated a significant interaction effect. However, due to especial type of my hypothesis, I need to conduct simple slope analysis as this paper has done. I searched a lot but I found only simple slope analysis using regression not for univariate analysis. I just copy a part of the article for more clarity. How could I calculate simple slope coefficient as this paper did? I will be appreciate if give me your answer.
Thanks
"The univariate results indicate significant interaction effect of failure
severity and brand reputation on satisfaction (F = 5.84, p b 0.05). The
simple slope analysis for less severe failure group showed that high
reputation group (M= 3.18) reported higher levels of customer satisfaction
than less reputation group (M = 2.31) (simple slope = 0.26,
p b 0.01). Cell means of high severe failure did not reveal difference in
customer satisfaction for the less reputation group (M = 2.26) and
high reputation group (M= 2.15), (simple slope= −0.03, p = 0.72).
This provides support for H5. The interaction effect is presented in Fig. 2.
For behavioral intentions, the interaction effect of failure severity
and brand reputation was marginal (F = 2.80, p = 0.09). Simple
slope analysis revealed significant difference between less severe-high
reputed group (M = 2.55) and less severe-less reputed group (M =
3.71), simple slope = 0.41, p b 0.01. Similarly for high failure severity
group, respondents evaluated greater behavioral intentions in high
reputed group (M= 2.47) than less reputed group (M= 1.85), simple
slope = 0.25, p b 0.01. This provides marginal support for H6. Fig. 3
plots the interaction effect on behavioral intentions."
http://www.sciencedirect.com/science/article/pii/S0148296314002616
when the moderating variable is non metric (yes/No; high/low) then it is advised to conduct multi-group analysis. Please check Gaskin's videos on U tube for the same.
Hi Jeevan,
Thanks a lot for your response. Please consider that my independent variable is nonmetric either. As I've found, you mean if I were SEM, I would perform a multi-group analysis. Am I right?
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA