Basic SmartPLS Model Assessment - Measurement Model - Reliability and Validity

[@davidtai3485]

Dr. Latif, you have done an excellent job in teaching us PLS-SEM. Your teaching is better than Hair, Ringle, and Sarstedt. In this video, you shared you'd talk about deleting indictors with low-loading. I searched in other videos, but didn't find them. I have 2 contracts with low-loading and I need to figure out a way to see what I need to do, Your kind help is much appreciated.

[@researchwithfawad]

Dear Mr. Tai, I am glad you liked the videos. May i know how low the loadings are. If the are low but removing them doesn't improve CR and AVE, you should keep tjem. Otherwise delete the items. Plus also the see if deleting the items doesn't significantly affects tye operationalization.

[@davidtai3485]

@@researchwithfawad Dr. Latif, I studied the low loading survey questions and decided to keep everything above .50 based on theory. For this reason, I only deleted 2, which is less than 10%. Suddenly the model made sense now. I used two existing questionnaires to obtain data. One produced negative loadings, but I realized that negative ones were actually positive because Strongly Disagree (1) actually equals to Strongly Agree (5) in another survey. I think need to convert the negative loadings to be positive. What method do you usually use to do this kind of negative to positive conversion?

[@researchwithfawad]

Hi. if you have reverse items, you will need to reverse the coding/response, like 5 to 1, 4 to 2, 4 to 2 and 1 to 5. You can email me on [e-mail] if you have any further queries.

[@gpdang]

It's best video on pls .hats off dil se shukriya 🙏🙏🙏

[@researchwithfawad]

Thanks. I am glad you liked it.

[@abebenegeri1823]

It is a very nice and helpful video. Many thanks, Doctor Latif!!!

[@researchwithfawad]

Thanks. I am glad you liked it

[@educationtestingservice5171]

i need other videos of CSR & OP videos . please share links here. Excellent video is explained

[@researchwithfawad]

Here is the complete playlist on using SMART-PLS kzbin.info/aero/PLb7vm6tsQ3KvTOFnl9QfW5YnXtz4kKsyK

[@ghanimsaqib6101]

Very Informative.

[@researchwithfawad]

Thanks. I am glad you liked it

[@researchwithfawad]

Research Scholars and Academics are welcome to suggest the next Research Webinar. Please select from the options and also circulate in your research circle. The duration will be One Day (Day: To be Announced, Timing: 10:00 - 5:00 Pakistan Standard Time) Proposed Fee: Rs. 2000 (Pakistani Scholars) $20 (International Scholars) Please Make Your Choice forms.gle/wYN1JWgcRdGATJE19 Also Kindly circulate in your Research Circle For Queries, Please email me on [e-mail]

[@muhammadusmanali669]

MaShaa Allah Great Sir ❤️

[@researchwithfawad]

Thank you

[@Alhamzah_F_Abbas]

Thank you for this video, if we have dimensions for a specific variable, is it better to run first-order and second-order constructs?

[@researchwithfawad]

Thanks. I am glad you liked it. Actually, if you are assessing the inter-relationship at lower level, then you do not need higher order analysis. But if hypotheses are proposed at higher level, then you will need to have higher order analysis

[@Alhamzah_F_Abbas]

@@researchwithfawad Thank you for your clarification

[@talhamansoor7108]

Thank you so much dear Sir❤️

[@researchwithfawad]

Pleasure. I am glad you liked it.

[@thechenmin]

Hello, Dr. Latif, in this video, you have assessed the AVE which has a convergent validity problem, but I didn't find the video that you solve this problem. I have also this problem, originally I have only 5 items of one construct, two items have loading >0.77, an other three items loaded only around 0.55. And this construct after I deleted two indicators with Lowes loading, then reached only to AVE = 0.457 and CR =0.711. I would like to know, how could I in this situation to prove there's no convergent validity problem under meine constructs. Your kind help is much appreciated. Thanks so much in advance. regards, Min

[@researchwithfawad]

Thanks for your comment Here is a video that may help kzbin.info/www/bejne/jKfYYoakqctpnMU Also, if your CR is over .70 and AVE is between 0.40 and 0.50, you can say Convergent Validity is established. Fornell, C., & Larcker, D. F. (1981). Structural equation models with unobservable variables and measurement error: Algebra and statistics.

[@anakBAPAK1]

Nice Dr..ok if we have good reliability and validity..CR, CA, AVE..all good and green but we still have item with outer loading..below 0.3..do we still need to remove it? thanks Dr

[@researchwithfawad]

Delete items that have loading less than 0.40

[@Shravan1231000]

Sir in case of pilot study, how to calculate relaibility? Is cronbach's alpha values using SPSS are enough? Or we have to run the analysis as shown in this video? In my thesis I am using PLS SEM and got stuck in pilot study.

[@researchwithfawad]

For pilot study, Reliability analysis using SPSS is fine.

[@Shravan1231000]

@@researchwithfawad thank you for the prompt reply sir

[@adankhaan6857]

Can we use the items with the loadings till.4 ..as i saw somewhere that .4 to .7 and above can be reported

[@researchwithfawad]

Yes, if they can help improve the reliability and validity over the required threshold.

[@TheSilentSid]

In your video CSR had low AVE. You said it's because CSR has four dimensions. Kindly guide how to solve construct reliability and validity issues of such variables in smart pls. Secondly, can we report reliability and validity statistics drawn from smart pls while perform mediation using sobel macro for spss in same research. Jazak Allah!

[@researchwithfawad]

To improve Validity statistics check for outer loadings. To improve Discriminant validity check cross loadings. Why use sobel test, it is better to use mediation analysis using pls. A more detailed video that can help kzbin.info/www/bejne/n3SkkIqpoteEY6M

[@MsSaddal]

sir do we have to reverse code our negative items before feeding file in smart pls software too?? as we do in spss

[@researchwithfawad]

Yes, reverse items shall be re-coded.

[@muhammadishtiaq1858]

Assalamo Alaikum. I am PhD scholar and recently started research work using smart pls. Can i run SEM with two latent variable and one Dependent variable measured through financial ratios. Plz guide me.

[@researchwithfawad]

For financial data you will need to check for assumptions. SmartPLS may not support those yet.

[@suryapokhrel2243]

Are not we suppose to use consistent PLS Algorithm for reflective model?

[@researchwithfawad]

According to experts we can use PLS for reflective measures.

[@marleyalvory3044]

Dear Dr. Latif could u demonstrate EFA and CFA through smartpls? Thank uuu

[@researchwithfawad]

Hi. Thanks for watching. Smartpls is not for EFA. You will have to use SPSS. As for CFA, you can refer* to measurement model assessment as CFA. Although in SmartPLS it is CCA: Composite Analysis.

[@marleyalvory3044]

@@researchwithfawad thank you sooo much for clarifying Dr. Latif. 😄👍🏻 Your explanation has helped me to clear my mind.

[@khalidashraf2690]

Asalam O Alikum, sir what if my path coefficients are low but p values are 0.000 and other things remain perfect. Will this be an issue?

[@researchwithfawad]

No that is not an issue, It is possible to have lower regression weight and your p-value still significant.

[@MsSaddal]

i have problem in discriminant validity as my few values are not less than .85. how i can resolve this?

[@researchwithfawad]

If it is less than .90, you are good to use them. If not, check if the items are cross loading. If so, remove cross-loaded items.

[@muhammadatharabbas5776]

What to do if your constructs are on dichotomous scale???

[@researchwithfawad]

Please watch kzbin.info/www/bejne/j3-Temt6rpx0os0

[@kristinegwen8280]

How do i count measurement model?

[@researchwithfawad]

I am sorry i do not understand? What do you mean by Count.

[@aliashfaq2017]

Dear Dr Latif, can you please guide me how to justify my composite reliability rho_a greater than 0.9 ? I have read your paper about "entrepreneurial leadership"in which your composite reliability values are following: 0.914 0.916 0.922 O.93 Your reply will be great help for me because I am working on my Master's thesis. Just because of your videos I have learned alot. You had done a remarkable contribution in knowledge sharing.

[@aliashfaq2017]

If someone else know's about the answer to my question please do let me know.!

[@researchwithfawad]

Values over 0.90 are fine. The values over 0.95 are much of a concern. Composite Reliability is a liberal measure and values tend to be high. You may refer to the book as well. Hair Jr, J. F., Hult, G. T. M., Ringle, C. M., Sarstedt, M., Danks, N. P., & Ray, S. (2021). Partial least squares structural equation modeling (PLS-SEM) using R: A workbook.

[@aliashfaq2017]

@@researchwithfawad Dear Sir, Thankyou so much for the response. But in this book it is stated on page # 77 that: ” Values above 0.90 (and defnitely above 0.95) are problematic, since they indicate that the indicators are redundant, thereby reducing construct validity (Diamantopoulos, Sarstedt, Fuchs, Wilczynski, & Kaiser, 2012).

[@researchwithfawad]

You are not way over 0.90, for just marginal. You are good to go.

[@TheSilentSid]

In your video CSR had low AVE. You said it's because CSR has four dimensions. Kindly guide how to solve construct reliability and validity issues of such variables in smart pls. Secondly, can we report reliability and validity statistics drawn from smart pls while perform mediation using sobel macro for spss in same research. Jazak Allah!

[@researchwithfawad]

To improve Validity statistics check for outer loadings. To improve Discriminant validity check cross loadings. Why use sobel test, it is better to use mediation analysis using pls. A more detailed video that can help kzbin.info/www/bejne/n3SkkIqpoteEY6M