Search found 1072 matches

by jmbecker
Mon Jan 20, 2020 8:54 am
Forum: FAQ (Methodology)
Topic: Low R square value
Replies: 1
Views: 24

Re: Low R square value

Low R² means that you are not able to explain much variance in your dependent variable. That can have several reasons. For example, your DV is very noisy (i.e., has a lot of variance). This may be because of error variance or because of many things contribute to the DV and your predictor is only one...
by jmbecker
Mon Jan 20, 2020 8:45 am
Forum: FAQ (Methodology)
Topic: SmartPLS basics
Replies: 1
Views: 56

Re: SmartPLS basics

You may want to read the Primer on PLS-SEM to get a basic understanding of the method and evaluation:
https://www.smartpls.com/documentation/ ... s-sem-book
by jmbecker
Mon Jan 20, 2020 8:43 am
Forum: Requested Features
Topic: Moderation Analysis
Replies: 1
Views: 417

Re: Moderation Analysis

You have to calculate that yourself using the data provided under Path Coefficients > Samples
by jmbecker
Sat Jan 18, 2020 3:45 pm
Forum: SmartPLS Small Talk Corner
Topic: Interpretation of Factor Scores
Replies: 1
Views: 75

Re: Interpretation of Factor Scores

The interpretation of standardized coefficients is: when you increase the IV by one standard deviation, the DV increases by 0.53 standard deviations. It is a dimension-less measure, similar to a correlation (but not necessarily bound between -1 and 1). If all you indicators are on the same scale you...
by jmbecker
Sat Jan 18, 2020 3:34 pm
Forum: FAQ (Methodology)
Topic: Index of moderated mediation
Replies: 1
Views: 66

Re: Index of moderated mediation

The index of moderated mediation is given by p2*p5. You can find this as the specific indirect effect of Y1*M -> Y2 -> Y3. This is exactly p2*p5. You can assess its significance using bootstrapping.
by jmbecker
Sat Jan 18, 2020 3:31 pm
Forum: FAQ (Methodology)
Topic: Poor effect size, predictive relevance with good R square
Replies: 1
Views: 10

Re: Poor effect size, predictive relevance with good R square

Maybe there is simply no relationship between your variables in your context? Or you have the wrong predictors?
by jmbecker
Sat Jan 18, 2020 3:29 pm
Forum: FAQ (Methodology)
Topic: P-values of reliability and AVE
Replies: 2
Views: 18

Re: P-values of reliability and AVE

The p-values are based on a true null, which might not be very useful. It is only the standard option in bootstrapping. However, 1) I would not use typical p-value / t-value test here because the distribution of the reliabilities is unknown, hence I would use bootstrap confidence intervals. 2) confi...
by jmbecker
Wed Dec 11, 2019 5:32 pm
Forum: FAQ (Methodology)
Topic: IMPA for model involving second order constructs
Replies: 2
Views: 234

Re: IMPA for model involving second order constructs

You need to run the IPMA also for the first-stage model and then take the unstandardized latent variables scores as indicators into the next stage. This is something that I would generally recommend, as you will get the same standardized results, but have some additional flexibility. For example, it...
by jmbecker
Wed Dec 11, 2019 5:26 pm
Forum: SmartPLS Small Talk Corner
Topic: Validating second order formative construct
Replies: 10
Views: 1357

Re: Validating second order formative construct

Sorry for the late response, I was also confused in the first moment, but you are using the example data of the corporate reputation model that is built-in to the SmartPLS software. However, the paper uses a replication of the model on a different dataset. I remember that this might have been some n...
by jmbecker
Wed Dec 11, 2019 2:36 pm
Forum: Application of PLS
Topic: EFA and PLS SEM
Replies: 1
Views: 233

Re: EFA and PLS SEM

PLS SEM is more like a CFA and thus you do not do EFA on the same dataset. If you want (or need) to do an EFA before the CFA, you should ideally have an additional dataset for this.
by jmbecker
Wed Dec 11, 2019 2:34 pm
Forum: FAQ (Methodology)
Topic: Controlled variables in SmartPLS
Replies: 1
Views: 85

Re: Controlled variables in SmartPLS

If the categories are not ordered and equidistant you may use a dummy coding scheme to model the ordinal variables.
by jmbecker
Wed Dec 11, 2019 2:31 pm
Forum: FAQ (Methodology)
Topic: Factor Score
Replies: 1
Views: 73

Re: Factor Score

All data is routinely standardized in the normal PLS algorithm. Hence, latent variable scores are also standardized. If you want unstandardized LV scores you may use the IPMA, which provides unstandardized LV scores.
by jmbecker
Wed Dec 11, 2019 2:29 pm
Forum: SmartPLS Small Talk Corner
Topic: Prediction oriented segmentation POS (help me)
Replies: 1
Views: 67

Re: Prediction oriented segmentation POS (help me)

You may read the following literature to understand POS segmentation: Hair, J. F., Sarstedt, M., Ringle, C. M., & Gudergan, S. P. 2018. Advanced Issues in Partial Least Squares Structural Equation Modeling (PLS-SEM). Thousand Oaks, CA: Sage. Becker, J.-M., A. Rai, C. M. Ringle, and F. Völckner (2013...
by jmbecker
Wed Dec 04, 2019 10:49 am
Forum: FAQ (Methodology)
Topic: Variable rated between 1 to 7
Replies: 1
Views: 84

Re: Variable rated between 1 to 7

Usually, researchers treat those 7-point Likert scales as quasi-metric and thus continuous. That would suggest using an interaction effect approach. In contrast, dichotomization using high and low values, for example, using a median split might lose information and is generally seen less favorably. ...
by jmbecker
Wed Dec 04, 2019 10:45 am
Forum: FAQ (Methodology)
Topic: Reflective Formative factor
Replies: 3
Views: 229

Re: Reflective Formative factor

If the construct has its own indicators, it is not a higher-order construct. A higher-order construct is generally describes as a construct where not some direct observed indicators are available for measurement, but some other latent variables are the actual measures (the lower-order constructs). T...