forum.smartpls.com

In general, the sample size will not be larger and therefore you will not have more information to estimate the model. Thus, sample size will still be an issue if you go for dummy variables. However, these dummy moderator models are slightly more efficient than multi-group models, because they restr...

If you want to assess the effect of Generations on D, you will need to put the variables as dummies in the model. You will need to use 3 single-item dummy variables, and the left out category will be the reference category (i.e., you compare the effect of the other generations against this reference...

1) If you don't have a variable that can act as a global single-item (representing the concept in whole) in your dataset, then you simply cannot do redundancy analysis. There is no way of creating such an item out of data. 2) The same evaluation and quality criteria that are also relevant for primar...

This should come from your theory and construct conceptualization. You will need to ask yourself the typical questions for distinguishing between reflective and formative measures, for example: Are your dimensions (the lower-order constructs) reflections of the higher-order construct or are they for...

SmartPLS 4 is now always using the two-stage approach because it has been found to be equal or better than the other approaches in simulation studies.

Please also find the information about moderation on our resource pages: https://www.smartpls.com/documentation/ ... moderation

Research had shown that using sign change methods produces incorrect results in some situations and is rarely helpful. Therefore, SmartPLS has long abandoned all sign change options (since 2018). Thus, it is basically always using "no sign change".

Doing an ANOVA on the latent variable scores would be possible, but it would not be in line with the non-parametric nature of PLS and probably not giving you the best result. However, you get information about mean differences from the permutation MGA invariance testing: Step 3a (mean). This is basi...

You cannot analyze a categorical variable directly. As PLS is regression-based you will need to dummy code or something similar. If you dummy code you will always have one (dummy) variable less than categories and the omitted category becomes the reference category. If you would include a dummy for ...

First, try not to bootstrap but to run the basic CB-SEM analysis. Do you get results? Not all moderator models are empirically identified in CB-SEM and it is not unlikely that estimation errors can occur (e.g., non convergence).

Probably, because your sample size is too small and there is not enough variance in the data.

We have, for now, disabled the inclusion of both quadratic and interactions terms because it is not directly clear what the correct model would be. There is also little guidance in the literature about such types of models. When you have an interaction, you have a conditional effect: The effect of X...

If you have the same variables in the regression for the outcome (dependent/endogenous) construct, you will get the same VIF values. Only if you have different predictors/regressors you will get different VIF.

VIF are overall very low, so this shows no problems with collinearity.

First, you seem to have a second-order construct. Engagement does not have any indicators directly attached. Thus, there is no reliability measures reported because they are usually based on the loadings of the indicators. As second-order construct is a theoretical lens, it is impossible for the sof...

The binary variable approach is one possibility, but as you say it has the downside of only estimating changes relative to the reference group. In such a situation, the "direct" simple effect is the effect in the reference group and the interaction effects are the changes in the effect whe...

There is unfortunately no easy answer to this question. We know from the literature that models that fit well, do not necessarily need to predict well. In fact, models that fit well can often predict more poorly than models that do not fit so well. This is the problem of overfitting, where the model...