categorical variables

[Libi]

I have been searching around the forum and the web but still could not find a step by step details on :
How can I use categorical variables in SmartPLS 3?

My research is on UTAUT model and I use Gender and Age as moderators.
I understood that I do not have to use MGA and I can define it as categorical variables, the question is what step should I follow to do so?

Many thank for your help,

BR,

Libi.

[jmbecker]

You can use categorical variables as dummy variables in your PLS model as you would also do in a normal regression analysis.
If you have only two categories (i.e., gender) you simply use a variable with 0=male and 1=female (or the other way round).
If you have several categories (age groups) you need the number of categories -1 as dummies. The omitted category is then the reference group. However, if you have age as a continuous variable it would be even better to include it as a continuous moderator.

[Libi]

Thank you Dr. Jan.
I did enter my data as a dummy variables into Excel ( age groups are entered in 3 columns Q1a, Q1b.. 0,1,0 if second option was selected and so on...)
The question is : when I upload the data into SmartPLS, how do I indicate it as categorical Variable?
as it is a moderator and no LV, no indicators should be included (correct ?!)
Do I actually upload it same as the LV with indicators and than press on moderator effect?

I just really try to understand the steps on SmartPLS after the data was entered in excel.

[Libi]

UTAUT_mode_ figure 1 Venkatesh et al. (2003).jpg (233.13 KiB) Viewed 7347 times

[jmbecker]

There is no way to indicate that a variable is categorical in SmartPLS.
First, you need to include each dummy as a single indicator construct in your model.
Second, if you have several categories in your moderator and thus several dummies you need to include a moderating effect for each dummy (category).

[Lena LUO]

I have a dummy variable (i.e., gender) as indepentent variable, do I use it as a single indicator construct in my model?
if I can't use multigroup Analysis, will it influence the bootstrapping results?

Many thank for you,

Yumei