Gender comparison of a sample -> alternating signs for items

PLS is broadly applied in modern business research. This forum is the right place for discussions on the use of PLS in the fields of Marketing, Strategic Management, Information Technology etc.
Post Reply
schabalaba
PLS Junior User
Posts: 1
Joined: Wed Sep 13, 2017 1:28 pm
Real name and title: Thorben Volters, B. Sc.

Gender comparison of a sample -> alternating signs for items

Post by schabalaba » Wed Sep 13, 2017 1:52 pm

Hello everyone,

I stumbled upon an interesting result in my analysis. I've got a sample of n= 140. 96 of which are male while the other 44 are female.
The study aims to identify the effect of privacyx concerns on the intention to use of mobile payment systems.
full.png
full sample
full.png (56.88 KiB) Viewed 796 times
Now, looking at the item loadings of the full sample ("full.png") and comparing them to the ones of the female-only sample ("female.png"), one can see that the item loadings don't just slightly differ, they even have alternating signs.
The results seem absolutely odd.
female.png
female subsamlpe
female.png (87.09 KiB) Viewed 796 times
Another example: the path of "Perceived Trust" -> "Privacy Concerns" for the female-only sample. The results suggest that if a woman believes a mobile payment service provider (such as Apple Pay) to be more trustworthy, her privacy concerns also increase.
Furthermore, the female sample suggests that higher privacy concerns increas the intention to use a mobile payment service...

I'd really appreciate your help on this since I ran out of ideas.

Thank you!
Thorben
Attachments
male.png
male subsample
male.png (52.83 KiB) Viewed 796 times

User avatar
cringle
SmartPLS Developer
Posts: 796
Joined: Tue Sep 20, 2005 9:13 am
Real name and title: Prof. Dr. Christian M. Ringle
Location: Hamburg (Germany)
Contact:

Re: Gender comparison of a sample -> alternating signs for i

Post by cringle » Wed Sep 20, 2017 5:54 pm

Is privacy concern a second-order construct using the indicators of the lower order constructs?

Then the problem is the coding of the lower order constructs with positive and negative signs. You should always use the same orientation of indicators (e.g., rescale the negative ones).

An alternative: SmartPLS uses intial values of +1. Alternatively, use a +1 for those outer relationships where you expect a positive sign and -1 where you expect a negative sign. It involves some careful "typing" but may be the solution to your problem. Here are some additional info: https://www.smartpls.com/documentation/ ... cification

Best regards
Christian

Post Reply