My technology adoption model after running tests shows significant and insignificant relationships. Some of the variables moderate the insignificant relationships. Should I keep them or drop them?
Is it ok if some variables moderate insignificant relationships in the technology adoption model?
Think you might want to recheck your literature review & provide appropriate discussion / justification. E.g. if the literature you'd reviewed indicating previous research have significant moderating effect & if you replicate the similar research but the moderating effect is not significant, then perhaps you need to find out why with supporting data / artifacts to discuss / justify your empirical finding.
Possible factors that might influence insignificant moderating effect include:
The wrong moderator is being selected - the rationale to select a variable as moderator should be based on literature review.
Larger sample size is needed when there are many moderators - a researcher needs to collect larger sample size because when many moderators are being introduced in a study, the interaction effect between IVs and moderators will create more new predictors that can influence DV. Some statistical analyses method used dependent on no. of predictors x certain constant to determine the no. of sample required. With small sample size, most of the moderating effect will not be significant.
Pl use stepwise regression. This will clearly identify the significant contributing factors and insignificant factor. Therefore your problems may be resolved. pl see.
Ultimately, you need to have a theoretical justification for the inclusion of any variables in a model. If you have the model y= b1x1 + b2x2 + b3x3 + b4x1x2, and if b1 is not significant but b4 is significant, then it is still important to include x1 in the model because b4 is the effect due to the moderation (i.e. where b1 is not significantly different than zero), and if the moderation effect was not included then you would not be revealing how x1 varies by differences in x2. Depending on the variables of interest this can be an important finding.