Hi everybody,
Maybe the question I am asking is stupid and I apologize in advance. I hope someone can help me.
If we have a latent endogenous variable Y explained by three latent exogenous variables X1, X2 and X3. In the table "Overview", the AVE of Y is the Percentage of variance explained by the three variables X1+X2+X3.
How to know the percentage of variance of Y explained by X1?
I have a proposal how to answer this question. I hope you will correct it if it's wrong.
If I take the values in the table "Latent variables correlations" and I square them, I will obtain the percentage of variance of the endogenous variable explained by each exogenous variable. Is it correct?
Thanks a lot for your help.
Best.
Kaouther
AVE and Variance
-
Diogenes
- PLS Super-Expert
- Posts: 899
- Joined: Sat Oct 15, 2005 5:13 pm
- Real name and title:
- Location: São Paulo - BRAZIL
- Contact:
Hi Kaouther,
AVE is the average % of the variance of all the indicators of the same LV that is explained by this LV.
R2 is the % of the variance of the endogenous LV (Y) that is explained by all exogenous LV (x1, x2, x3).
R2 = path11 * r11 + path 21 * r21 + path31 * r31
Then the % of variance of the LV (Y) that is explained by x1 will be = path11 * r11
Best regards,
Bido
AVE is the average % of the variance of all the indicators of the same LV that is explained by this LV.
R2 is the % of the variance of the endogenous LV (Y) that is explained by all exogenous LV (x1, x2, x3).
R2 = path11 * r11 + path 21 * r21 + path31 * r31
Then the % of variance of the LV (Y) that is explained by x1 will be = path11 * r11
Best regards,
Bido
-
JELASSI
- PLS User
- Posts: 14
- Joined: Fri May 21, 2010 8:23 am
- Real name and title:
- Location: Paris (FRANCE)
Dear Professor,
I followed your advice and I computed the % of variance of the LV (Y) that is explained by x1 and x2 (path 11 * r11 and path 21 * r21). X1 explained 31% of variance of Y and X2 explained 22% of variance of Y.
But the path31 of my 3rd variable is negative.
Can I apply the same formula to compute the % of variance of the LV (Y) that is explained by x3?
How could I interpret the result?
Thank you very much
Best regards.
Kaouther
I followed your advice and I computed the % of variance of the LV (Y) that is explained by x1 and x2 (path 11 * r11 and path 21 * r21). X1 explained 31% of variance of Y and X2 explained 22% of variance of Y.
But the path31 of my 3rd variable is negative.
Can I apply the same formula to compute the % of variance of the LV (Y) that is explained by x3?
How could I interpret the result?
Thank you very much
Best regards.
Kaouther
-
Diogenes
- PLS Super-Expert
- Posts: 899
- Joined: Sat Oct 15, 2005 5:13 pm
- Real name and title:
- Location: São Paulo - BRAZIL
- Contact:
Hi Kaouther,
Yes, use the same way.
If the path31 is negative, probably the r31 is negative too, then ( – * – = + ), this parcel of explained variance will be added to the other two (x1 and x2).
Interpretations with a numeric example:
a) Path31 = – .23 (when x3 increases by 1 standard deviation, y will decreases .23 standard deviation.
b) Path31 * r31 = +.09 (x3 explain 9% of the variance of y)
c) Path31 * r31 = – .09 (you should not interpret, because there is a suppression problem).
If the path31 is negative and r31 is positive, you have a suppression problem (multicollinearity between x1, x2, x3), even in this case, compute the R2 in the same way.
Best regards,
Bido
Yes, use the same way.
If the path31 is negative, probably the r31 is negative too, then ( – * – = + ), this parcel of explained variance will be added to the other two (x1 and x2).
Interpretations with a numeric example:
a) Path31 = – .23 (when x3 increases by 1 standard deviation, y will decreases .23 standard deviation.
b) Path31 * r31 = +.09 (x3 explain 9% of the variance of y)
c) Path31 * r31 = – .09 (you should not interpret, because there is a suppression problem).
If the path31 is negative and r31 is positive, you have a suppression problem (multicollinearity between x1, x2, x3), even in this case, compute the R2 in the same way.
Best regards,
Bido