Fsquare, t-test, fit model scores

[sale71111]

Hello

I have a simple questions that may be somewhat silly, but because I am new in this area I have difficulty in understanding:

1-what is the difference between the values of Rsquare and the value of Fsquare? does it related to t-test scores where all my path t-test above the 1.96? see the attached pic
my model passed the test(Rsquare 0.48-cronpachs alpha-AVE-composite reliability) all values great
2-what is rho_A and what the threshold value?
3-pleas see the fit model attached pic, what these values means?

iam using smartpls 3
thanks in advance

[asasi.kamal]

Please search in YouTube. There are plenty useful videos that explain these concepts visually.

[cringle]

@1: Please take a look at the PLS-SEM book --> https://www.smartpls.com/documentation/ ... on-pls-sem R² is the amount of explained variance and f² represents the effect size.

@2: It's a construct reliability measuere; use .7 as threshold.

@3: You usually do not need these results. But take a look here: https://www.smartpls.com/documentation/fit

[Alexandra]

Hello,
How to interpret an effect size (f squared) =1.000 ?
In assessing change of R squared in structural model
Thanks
Alexandra

[cringle]

Here, you have a substantial / strong effect. For example, when R²_included is 0.55 and your R²_excluded is 0.1, you f² effect size is 1. Hence, the exluded construct has a strong R² value contribution.

Best
Christian

[bsmithx]

Related question here:
Looking at two paths coming from a single LV (V1 => V2, V1 => V3) but getting f2 values greater than 1. How is this possible? Is it problematic?

[cringle]

Why should it be problematic? f² = (R²_included - R²_excluded) / (1 - R²_included). The result does not have an upper limit of one; values larger than 1 are possible; just take 0.6 instead of 0.55 in the above example. A f² value above 0.35 is considered to be substantial.

Best
CR

[bsmithx]

alright. seems logical. just trying to avoid / address any red flags in advance.
thank you.

[Levimich]

Hi,
I have a basic question about the T-statistics
I added my model a control variable and the path coefficient turned out to be negative.
you would expect that the T-statistics will also be negative but it is actually positive. why does it happen and how the bootstrapping relate to that because statistically, it doesn't make sense.

I hope to solve this

Michal

[cringle]

the t values are absolute values. but you should use the 95% bca bootstrap confidence intervals from the SmartPLS 3 report for significance testing.