IPMA (negative performance)

Frequently asked questions about PLS path modeling.
Post Reply
tinker18
PLS Junior User
Posts: 8
Joined: Thu Jul 26, 2018 1:33 pm
Real name and title: tinker

IPMA (negative performance)

Post by tinker18 » Tue Jul 31, 2018 3:53 am

Hello everyone,

The results of my IPMA showed normal figures for both importance and performance between 0-100, and I was satisfied with the outcome. Then, I realized the scales for min and max were not 1 - 7. It showed decimal numbers; negative min (-1.238) and positive max (1.185). So I changed the scale to 1 - 7. Now this is where my problem comes in. The performance indicator showed all of the 7 constructs were -16.666! I am just wondering where did I go wrong and how to rectify this issue. Just to add, my model is a second higher order model (reflective-formative).

Thank you.

jmbecker
SmartPLS Developer
Posts: 879
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: IPMA (negative performance)

Post by jmbecker » Tue Jul 31, 2018 8:07 am

You should use 1-7 if the data is in that scale and not standardized. If you import standardized data (and your numbers suggest something like this) you need to set the standardized min and max value.
The software automatically identifies the empirical min and max. Of course it cannot know the implied scale range. The identified empirical min and max are is fine as long as your data includes the full range of values (i.e., at least one 1 and one 7 for every variable). However, if your data does not include the full range (i.e., a variable has only values from 3 to 6 even though the scale was from 1-7) then you need to re-specify the values the min and max for that variable in the software.
Dr. Jan-Michael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de

tinker18
PLS Junior User
Posts: 8
Joined: Thu Jul 26, 2018 1:33 pm
Real name and title: tinker

Re: IPMA (negative performance)

Post by tinker18 » Tue Jul 31, 2018 8:59 am

jmbecker wrote:
Tue Jul 31, 2018 8:07 am
The identified empirical min and max are is fine as long as your data includes the full range of values (i.e., at least one 1 and one 7 for every variable). However, if your data does not include the full range (i.e., a variable has only values from 3 to 6 even though the scale was from 1-7) then you need to re-specify the values the min and max for that variable in the software.
Thanks Dr Becker,

Since my model is a HOC, is there any other ways to run the IPMA or is it the same as running an LOC model? I calculated the IPMA using the two stage approach in my model. When I tried to re-specify the values of min and max 1-7 for some variables (as mentioned in the quote above), I get a decent number for the HOC, whereas the LOC still showing -16.666. I am curious to know how IPMA is being implemented in HOC models.

Thank you.
Last edited by tinker18 on Wed Aug 01, 2018 1:55 am, edited 1 time in total.

jmbecker
SmartPLS Developer
Posts: 879
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: IPMA (negative performance)

Post by jmbecker » Tue Jul 31, 2018 12:02 pm

As you are using the two-stage approach you import the standardized LOC variables into your dataset. Hence, you have the problem and cannot specify the 1-7 range. If all your variables for the LOC cover the full range from 1 to 7 you can just trust the empirical min and max for the standardized LOC latent variable scores and not respecifiy anything.
You could also try to use the unstandardized LOC scores from the IPMA in the first stage as indicators of your HOC in the second stage.
Dr. Jan-Michael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de

tinker18
PLS Junior User
Posts: 8
Joined: Thu Jul 26, 2018 1:33 pm
Real name and title: tinker

Re: IPMA (negative performance)

Post by tinker18 » Tue Jul 31, 2018 3:09 pm

Awesome! Thanks, Dr Becker.

Post Reply