latent variable score

Frequently Asked Questions about the SmartPLS software. Please refer to this section first, if you have problems with the software.
needpls
PLS Expert User
Posts: 119
Joined: Mon Jun 11, 2012 4:41 pm
Real name and title:

latent variable score

Post by needpls » Tue Jul 24, 2012 11:33 pm

hii

1)plz say me how to compute latent variable score with hand,i want to compare with result of pls.

2)lisrel and pls compute latent variable score in diffrent ways or with same formula????



best regards

User avatar
Diogenes
PLS Super-Expert
Posts: 905
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes » Wed Jul 25, 2012 12:13 am

Hi,

LV = w1 * MV1 + w2 * MV2 + ... + wn * MVn

Where MV are the manifested variable (indicators) after standardization [(Xi – Mean) / Standard Dev]
and wn are the outer weights.

After compute LV you should to standardize the LV scores to have the “standardized LV socres”, with mean = 1 and s.d. = 1

If you want to compute the LV scores unstandardized follow: viewtopic.php?t=812


Best regards,

Bido

needpls
PLS Expert User
Posts: 119
Joined: Mon Jun 11, 2012 4:41 pm
Real name and title:

Post by needpls » Wed Jul 25, 2012 8:46 am

hiiii

thx so muchhhhh


but i had 2 question plzzz say my answre about question 2.


latent variable score in lisrel and pls have different way??

User avatar
Diogenes
PLS Super-Expert
Posts: 905
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes » Wed Jul 25, 2012 4:03 pm

Hi,
PLS-PM is consistent and consistente at large = as sample size and the number of indicators by LV increase --> the results of PLS-PM tend to be equal the results of LISREL.

LISREL is covariance based = the relations are computed from LV to the indicators, the last thing to be computed are the scores of LV.
PLS-PM is variance based = the relations are computed from indicators to the LV, the first thing to be computed are the scores of LV.

For these reasons, the correlation between scores computed by LISREL and scores computed by SmartPLS should be high, but they will not be the same.

For example (p.192):
Tenenhaus, M., Esposito Vinzi, V., Chatelin, Y.-M., & Lauro, C. (2005). PLS path modeling. Computational Statistics & Data Analysis, 48(1), 159-205. doi:10.1016/j.csda.2004.03.005

Best regards,

Bido

needpls
PLS Expert User
Posts: 119
Joined: Mon Jun 11, 2012 4:41 pm
Real name and title:

Post by needpls » Wed Jul 25, 2012 4:33 pm

thx so much,

so measure ment error that we have in lisrel in pls is too???

needpls
PLS Expert User
Posts: 119
Joined: Mon Jun 11, 2012 4:41 pm
Real name and title:

Post by needpls » Wed Jul 25, 2012 4:38 pm

sorry i can not find that article can u upload ??

thx

User avatar
Diogenes
PLS Super-Expert
Posts: 905
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes » Thu Aug 02, 2012 1:46 pm


needpls
PLS Expert User
Posts: 119
Joined: Mon Jun 11, 2012 4:41 pm
Real name and title:

Post by needpls » Sat Aug 04, 2012 9:24 am

thx

belowzero
PLS User
Posts: 11
Joined: Thu Apr 18, 2013 2:53 am
Real name and title:

Post by belowzero » Wed May 01, 2013 12:48 am

Diogenes wrote:Hi,

LV = w1 * MV1 + w2 * MV2 + ... + wn * MVn

Where MV are the manifested variable (indicators) after standardization [(Xi – Mean) / Standard Dev]
and wn are the outer weights.

After compute LV you should to standardize the LV scores to have the “standardized LV socres”, with mean = 1 and s.d. = 1

If you want to compute the LV scores unstandardized follow: viewtopic.php?t=812


Best regards,

Bido
hi bido
for reflective model,how to compute LV score?

User avatar
Diogenes
PLS Super-Expert
Posts: 905
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes » Wed May 01, 2013 11:20 am

Hi,
in the same way, use the "outer weights".

SmartPLS already computes the LV scores, see::
Report / PLS / Calculation results / Latent variable scores /

Best regards,

Bido

belowzero
PLS User
Posts: 11
Joined: Thu Apr 18, 2013 2:53 am
Real name and title:

Post by belowzero » Wed May 01, 2013 11:29 am

Diogenes wrote:Hi,
in the same way, use the "outer weights".

SmartPLS already computes the LV scores, see::
Report / PLS / Calculation results / Latent variable scores /

Best regards,

Bido
but i remember what you said before,the lv scores(original) SMARTPLS computed are not correct? i need compute them by hand?

thanks!

belowzero
PLS User
Posts: 11
Joined: Thu Apr 18, 2013 2:53 am
Real name and title:

Post by belowzero » Wed May 01, 2013 11:49 am

in addition,outer loadings and outer weights are standardized of the report?

User avatar
Hengkov
PLS Super-Expert
Posts: 1619
Joined: Sun Apr 24, 2011 10:13 am
Real name and title: Hengky Latan
Location: AMQ, Indonesia
Contact:

Post by Hengkov » Wed May 01, 2013 11:45 pm

Hi,

Yes, choose "mean 0 var 1" for standardized and LVs to standardized (original not standardized).

Regards,
Hengky

belowzero
PLS User
Posts: 11
Joined: Thu Apr 18, 2013 2:53 am
Real name and title:

Post by belowzero » Thu May 02, 2013 12:42 am

Hengkov wrote:Hi,

Yes, choose "mean 0 var 1" for standardized and LVs to standardized (original not standardized).

Regards,
Hengky

i see,thank you!

behnammousavi
PLS Junior User
Posts: 1
Joined: Mon Nov 06, 2017 9:11 am
Real name and title: behnam mousavi

Re: latent variable score

Post by behnammousavi » Mon Nov 06, 2017 9:28 am

hi
i'm new in PLS and i have a model that i wanna run with pls3.
in my model i have a latent variable that must calculate from 4 other latent variable. in lisrel it could compute latent variable from other latent variable without any indicator but in pls when i want to do such thing it warn me that your model is invalid. is there any way so i can calculate this latent variable is pls?

Post Reply