Hi,
I am a new user of SmartPLS and I am currently busy with writing my master's thesis using PLS method. I read many of the posts in this forum but still could not clarify two issues. They look quite easy but I do not want to make any mistakes on these. I would be extremely grateful if someone could help.
1- One of my indicators is reversed when asking the respondents. Therefore, when I import my data, it correlates negatively to the construct it measures. How can I fix this? Do I need to reverse the numbers manually in my data sheet or is there a reverse option for indicators in Smart PLS that I can use?
2- I understood that when you would like to examine a moderation effect of a categorical variable ( or make a group comparison) you have to split the data set and apply different analysis to both sets of data. So does that mean if we use "create moderation effect" option with a categorical construct, it would be technically wrong and give misleading results?
I look forward to hear from you on these issues. Thanks in advance for your help.
Kind regards, Handan
reversed measurement of indicator
- Diogenes
- PLS Super-Expert
- Posts: 899
- Joined: Sat Oct 15, 2005 5:13 pm
- Real name and title:
- Location: São Paulo - BRAZIL
- Contact:
Hi,
1 – You must reverse the indicator before the importation to SmartPLS.
If they are in a Likert format:
5 categories --> New value = 6 – old value
7 categories --> New value = 8 – old value
If they are in a scale format:
New = Old * ( – 1)
2 – Yes, it will be wrong, for instance, if we have the variable “color” where: 1 = green, 2= blue, 3 = red. The numbers 1, 2, 3 have no quantitative meaning.
If the categorical variable is a dummy (0, 1) you could use the “create moderation approach”, but the better way is multigroup analysis.
Best regards,
Bido
1 – You must reverse the indicator before the importation to SmartPLS.
If they are in a Likert format:
5 categories --> New value = 6 – old value
7 categories --> New value = 8 – old value
If they are in a scale format:
New = Old * ( – 1)
2 – Yes, it will be wrong, for instance, if we have the variable “color” where: 1 = green, 2= blue, 3 = red. The numbers 1, 2, 3 have no quantitative meaning.
If the categorical variable is a dummy (0, 1) you could use the “create moderation approach”, but the better way is multigroup analysis.
Best regards,
Bido