Dear all,
I just want to know what is the overall measurement of goodness of fit in
PLS and how to measures the significance of the coefficient?
Thank you
Best regards
iin
Goodness of Fit
-
Diogenes
- PLS Super-Expert
- Posts: 899
- Joined: Sat Oct 15, 2005 5:13 pm
- Real name and title:
- Location: São Paulo - BRAZIL
- Contact:
Hi,
The GoF is a geometric mean of structural model (R2) and the measurement model (AVE), see:
TENENHAUS, M.; ESPOSITO VINZI, V.; CHATELIN, Y.-M.; LAURO, C. PLS path modeling. Computational Statistics & Data Analysis, v. 48, n. 1, p. 159-205, 2005.
https://studies2.hec.fr/jahia/webdav/si ... S_PM_5.pdf
To measure the significance you must use the bootstrap procedure.
Sign changes = construct level or individual changes
Cases = sample size (rows in the dataset)
Samples = number of resamples, usually 1000 or more.
Use the search function for more details about bootstrap.
Best regards,
Bido
The GoF is a geometric mean of structural model (R2) and the measurement model (AVE), see:
TENENHAUS, M.; ESPOSITO VINZI, V.; CHATELIN, Y.-M.; LAURO, C. PLS path modeling. Computational Statistics & Data Analysis, v. 48, n. 1, p. 159-205, 2005.
https://studies2.hec.fr/jahia/webdav/si ... S_PM_5.pdf
To measure the significance you must use the bootstrap procedure.
Sign changes = construct level or individual changes
Cases = sample size (rows in the dataset)
Samples = number of resamples, usually 1000 or more.
Use the search function for more details about bootstrap.
Best regards,
Bido
-
indrawati
- PLS User
- Posts: 20
- Joined: Tue Nov 30, 2010 3:56 am
- Real name and title:
- Location: Indonesia
Multicolinearity and non normal distribution
Dear Prof. Bido,
Thank you for the answers. I have other questions.
Why in PLS there is no distributional requirements of data?
And, there is No Constraint on Multicollinearity?
Thank you very much.
Best regards
iin
Thank you for the answers. I have other questions.
Why in PLS there is no distributional requirements of data?
And, there is No Constraint on Multicollinearity?
Thank you very much.
Best regards
iin
-
Diogenes
- PLS Super-Expert
- Posts: 899
- Joined: Sat Oct 15, 2005 5:13 pm
- Real name and title:
- Location: São Paulo - BRAZIL
- Contact:
Hi,
1) The significance is estimated fro bootstrap procedure, for this reason, the normality is not a assumption.
2) The multicollinearity is a problem, in the same sense as multiple regression.
It could happen in the formative measurement model or in the structural model, resulting in some nonsignificant path or even paths with contrary signal from the correlations.
Best regards,
Bido
1) The significance is estimated fro bootstrap procedure, for this reason, the normality is not a assumption.
2) The multicollinearity is a problem, in the same sense as multiple regression.
It could happen in the formative measurement model or in the structural model, resulting in some nonsignificant path or even paths with contrary signal from the correlations.
Best regards,
Bido
-
indrawati
- PLS User
- Posts: 20
- Joined: Tue Nov 30, 2010 3:56 am
- Real name and title:
- Location: Indonesia
Heteroscedaticity and autocorrelation
Dear Prof. Bido,
Thanks for your kindness answering my questions. Now I have another questions.
Should we consider heteroscedasticity and autocorrelation in PLS?
Thank you.
Best regards
iin
Thanks for your kindness answering my questions. Now I have another questions.
Should we consider heteroscedasticity and autocorrelation in PLS?
Thank you.
Best regards
iin
-
indrawati
- PLS User
- Posts: 20
- Joined: Tue Nov 30, 2010 3:56 am
- Real name and title:
- Location: Indonesia
Cases in Bootstrap
Dear Prof. Bido and Prof. Ghozali or Any body
I just want to make sure about cases in bootstrap.
Prof. Bido wrote previously as follow:
[To measure the significance you must use the bootstrap procedure.
Sign changes = construct level or individual changes
Cases = sample size (rows in the dataset)
Samples = number of resamples, usually 1000 or more. ]
Prof. Imam Ghozali (in his book SEM model alternatif PLS, 2006:63) wrote that cases = 100 by default.
As you know different cases influence the t-value result.
Could you please verify this?
Thank you
Best regards
indrawati[/b]
I just want to make sure about cases in bootstrap.
Prof. Bido wrote previously as follow:
[To measure the significance you must use the bootstrap procedure.
Sign changes = construct level or individual changes
Cases = sample size (rows in the dataset)
Samples = number of resamples, usually 1000 or more. ]
Prof. Imam Ghozali (in his book SEM model alternatif PLS, 2006:63) wrote that cases = 100 by default.
As you know different cases influence the t-value result.
Could you please verify this?
Thank you
Best regards
indrawati[/b]