means and standard deviations of the constructs

[Cheryl Hsu]

Dear all,

Can we get the means and standard deviations of the constructs?

In PLS3, the default value setting is standardized. The value of average is 0 and standard deviation is 1.

Is it possible to change this setting? Can we get non-standardized values from PLS3?

Thanks in advance!

Cheryl

[cringle]

The PLS algorithm uses only standardized data (mean = 0). Run the IMPA to get unstandardized LV scores. The performance is their mean value.

Run Bootstrapping to get standard errors which equals the standard deviation.

Best regards
Christian

[Everist]

After bootstraping in the updated version of PLS3 I do not get standard errors anymore.
Can it be that the updated version of PLS3 changed the name of the column from “Standard Error (STERR)” to “Standard Deviation (STDEV)” but the values are the same?

Thank you in advance!

[cringle]

Thanks. In bootstrapping, the standard error and standard deviation are equivalent. In the old version, we had two columns with the same values. Now, we only report the results only once and label is standard deviation which equals the standard error in bootstrapping.

Best
Christian

[miueno]

Could you kindly let me know why they are identical in bootstrapping? Thank you.

[jmbecker]

In bootstrapping you calculate the model, for example, 1000 times (the number of subsamples). The bootstrap estimate of the standard error of a parameter is the standard deviation of the bootstrap distribution that you get from those 1000 estimates. Thus, the standard deviation of the 1000 parameter estimates is used to estimate the standard error. Therefore, standard deviation and standard error are equivalent in this case.

See for example: Chernick 2007; Efron & Tibshirani 1993

[miueno]

Great. Thank you!