Hi, all
if the dependent LV is simply measured by an ordinal scale, (such as 1=stong, 2=weak, 3= non), is it possible to test it in SmartPLS, and how? Thank you!
Yiqing
If LV is measured by ordinal scale
-
Diogenes
- PLS Super-Expert
- Posts: 899
- Joined: Sat Oct 15, 2005 5:13 pm
- Real name and title:
- Location: São Paulo - BRAZIL
- Contact:
Hi,
PLS-PM treats the data as interval numbers, if 1 < 2 < 3 in your case, and it is interpretable, ok.
Just another idea.
1) Run a confirmatory factor analysis:
1.1 connect all the exogenous LV
1.2 Run the PLS algorithm with the “factor” weighting scheme.
1.3 Assess the validity and reliability
1.4 Copy the LV scores to an Spreadsheet (Excel or SPSS, for instance)
1.5 Copy the “dependent LV” column to this new spreadsheet.
2) Run ANOVA (or Kruskal-Wallis test), to know if the mean score are different between the levels of the dependent variable.
Some kind of logistic regression could be used to.
3) You could use standardized scores (mean = 0; std error = 1) or the unstandardized LV scores, if the items are in the same scale.
Bido
PLS-PM treats the data as interval numbers, if 1 < 2 < 3 in your case, and it is interpretable, ok.
Just another idea.
1) Run a confirmatory factor analysis:
1.1 connect all the exogenous LV
1.2 Run the PLS algorithm with the “factor” weighting scheme.
1.3 Assess the validity and reliability
1.4 Copy the LV scores to an Spreadsheet (Excel or SPSS, for instance)
1.5 Copy the “dependent LV” column to this new spreadsheet.
2) Run ANOVA (or Kruskal-Wallis test), to know if the mean score are different between the levels of the dependent variable.
Some kind of logistic regression could be used to.
3) You could use standardized scores (mean = 0; std error = 1) or the unstandardized LV scores, if the items are in the same scale.
Bido