Thank you very much Dr. James. If we have a model with higher order constructs, should we conduct the Importance Performance Map Analysis on the whole model or on the higher order construct only? Thanks for your clarification!
Hi James, thx for your excellent Videos, as ever. One question to IPMA: how do I Interpret negative total effects in the IPMA considering the Abalo Et al 2007 Quadrant? Do I need to produce reverse scaled items first in order to have them all positive in the IPMA?
Negative effects just imply an inverse relationship. If the total effect is negative, that can be due to one negative part of an indirect effect. If a negative effect was not expected, then check the scale and wording of items to make sure they are aligned with the construct. For example, if measuring job satisfaction, the items should measure job satisfaction, not job dissatisfaction. Also, the scale itself should go from left to right, low to high, e.g., strongly disagree to strongly agree, so that as the scale value increases, so does the trait value of the construct being measured.
Thx for the answer. I understand this well. But there are structural models, where you have an expected negative total effect and and expected positiv total effect at the same time in one structural model, e.g. impact of Burnout on Intention to leave (positiv total effect) and Training on Intention to leave (negativ total effect). Then the IPMA goes from e.g. - 0.6 to + 0.6 on the X-axis (Importance-Axis). I think - for the management advices to draw from this - one has just to interpret the negativ strong effects mirror-inverted, right? (high-negativ = strong importance)
@@jojohahe I'm not sure I understand. Values on the left side of the x axis should be considered "less important". Here is a useful reference with further references: www.smartpls.com/documentation/algorithms-and-techniques/ipma/
I want to confirm; is it the same process for a model with a repeated indicator approach (in the second stage). Because Smart-PLS says, "Ranges The IPMA rescales the data to provide performance scores on a scale from 0 to 100. For the correct rescaling, the original scales of data are essential information. Here, the user can check and correct the possible ranges of the manifest variables. For example a 7-point Likert sale must have a minimum value of 1 and a maximum value of 7 in the IPMA settings. The preconfigured ranges are based on the actual minimum and maximum values found in the dataset. This must not equal the theoretically possible ranges and should therefore be adjusted. Again, the correct ranges are necessary to calculate correct performance values for the importance-performance map." and at this stage, it is the latent factors' scores I am using Smart-PLS 3.3.9
Thank you Prof. James Gaskin for another excellent video!
Thank you very much Dr. James.
If we have a model with higher order constructs, should we conduct the Importance Performance Map Analysis on the whole model or on the higher order construct only?
Thanks for your clarification!
It's intended for any dependent variable.
Keep up the great work. I enjoyed your SEM Boot Camp at BYU.
These videos are great. I'd like to see a beginning to end analysis in smart PLS. 😃
Thank you for clarifying the IPMA.
Thank you Prof. James Gaskin for the videos on Smartpls 4. Are there any video which explains how to do gaussian copula approach in smartpls 4?
No. It is still out for debate, so I have waited to make a video about it.
Hi James, thx for your excellent Videos, as ever. One question to IPMA: how do I Interpret negative total effects in the IPMA considering the Abalo Et al 2007 Quadrant? Do I need to produce reverse scaled items first in order to have them all positive in the IPMA?
Negative effects just imply an inverse relationship. If the total effect is negative, that can be due to one negative part of an indirect effect. If a negative effect was not expected, then check the scale and wording of items to make sure they are aligned with the construct. For example, if measuring job satisfaction, the items should measure job satisfaction, not job dissatisfaction. Also, the scale itself should go from left to right, low to high, e.g., strongly disagree to strongly agree, so that as the scale value increases, so does the trait value of the construct being measured.
Thx for the answer. I understand this well. But there are structural models, where you have an expected negative total effect and and expected positiv total effect at the same time in one structural model, e.g. impact of Burnout on Intention to leave (positiv total effect) and Training on Intention to leave (negativ total effect). Then the IPMA goes from e.g. - 0.6 to + 0.6 on the X-axis (Importance-Axis). I think - for the management advices to draw from this - one has just to interpret the negativ strong effects mirror-inverted, right? (high-negativ = strong importance)
@@jojohahe I'm not sure I understand. Values on the left side of the x axis should be considered "less important". Here is a useful reference with further references: www.smartpls.com/documentation/algorithms-and-techniques/ipma/
Nice explaination
I want to confirm; is it the same process for a model with a repeated indicator approach (in the second stage). Because Smart-PLS says,
"Ranges
The IPMA rescales the data to provide performance scores on a scale from 0 to 100. For the correct rescaling, the original scales of data are essential information. Here, the user can check and correct the possible ranges of the manifest variables. For example a 7-point Likert sale must have a minimum value of 1 and a maximum value of 7 in the IPMA settings.
The preconfigured ranges are based on the actual minimum and maximum values found in the dataset. This must not equal the theoretically possible ranges and should therefore be adjusted.
Again, the correct ranges are necessary to calculate correct performance values for the importance-performance map."
and at this stage, it is the latent factors' scores
I am using Smart-PLS 3.3.9
Good question... I'm not sure I'm confident about an answer. I would suggest searching for it here: forum.smartpls.com
@@Gaskination Thank you, James. I will try to find it there and thank you for all your informative videos.
How to do multigroup analysis in smart pls4?
Here you go: ua-cam.com/video/L2LdAgMKmBo/v-deo.html
@@Gaskination Thanks a lot sir