loadings value

This forum is closed, and read-only.
lopez
PLS Junior User
Posts: 1
Joined: Mon Oct 29, 2007 9:45 am
Real name and title:

loadings value

Post by lopez »

I would like to know which is the minimum acceptable value of the "loadings" (0.6 or 0.7).

Thank you
User avatar
Diogenes
PLS Super-Expert
Posts: 899
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes »

Hi Miguel,
It is recommended 0,7 because 0,7^2 = 0,49
49% (almost 50%) of the variance of the indicator is explained by its LV.

Remenbering that AVE also should be > 50% (all loading = 0,7; the AVE will be 49%).

In practical cases we have some indicators lower than 0,7, but others bigger than 0,7, with AVE > 0,5;
in this case is common to keep the indicator even when its loading < 0,7, mainly in cases where we have just 3 or 4 indicators.

Best regards.
Bido
NitinJain
PLS Junior User
Posts: 5
Joined: Mon Jan 21, 2008 9:45 pm
Real name and title:

Can Outer Loadings be negative for a reflective indicator

Post by NitinJain »

Hi,

Can Outer Loadings be negative for a reflective indicator.
In my case 3 out 6 indicators for a reflective dimension have a negative outler loading.

Thanks,
Nitin
User avatar
Diogenes
PLS Super-Expert
Posts: 899
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes »

Hi Nitin,
It could be a problem because this has a influence in the signal of the structural paths (could change the signal).
I suggest that:
1) Compare these results with the simple correlations between these indicators (the signals should be the same)
2) You could invert the scale of some indicators to have all of them with positive correlations.
3) After that, the results in SmartPLS will be more interpretable.
Best regards.
Bido
naggi
PLS Junior User
Posts: 5
Joined: Thu Oct 23, 2008 10:39 pm
Real name and title:

Re: loadings value

Post by naggi »

lopez wrote:I would like to know which is the minimum acceptable value of the "loadings" (0.6 or 0.7).

Thank you
And these results should be taken from the "Cross loading" under "Quality criteria" section in the the regular (not Bootstrap) results? Just to be sure!

Thank you,

Pankaj
User avatar
Diogenes
PLS Super-Expert
Posts: 899
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes »

Hi Nitin,
the simple correlations between the indicators could be assessed in Excel or SPSS.
The "outer loadings" or "cross loadings" will show the same results that you could see in the model.
Best regards,
Bido
chinasun
PLS User
Posts: 14
Joined: Thu Mar 20, 2008 1:41 pm
Real name and title:

Post by chinasun »

Diogenes wrote:Hi Nitin,
It could be a problem because this has a influence in the signal of the structural paths (could change the signal).
I suggest that:
2) You could invert the scale of some indicators to have all of them with positive correlations.

Bido
professor
How to invert the scale of indicators that have negative sign in PLS?

Thanks in advance!
User avatar
Diogenes
PLS Super-Expert
Posts: 899
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes »

Hi Sun,

with 5 point Likert scale, to have 1 --> 5; 2 --> 4...
Just do "New value = 6 - Old value"

with 7 point Likert scale, to have 1 --> 7; 2 --> 6...
Just do "New value = 8 - Old value"

With numeric value (measured variable) wwe could multiply the values by (-1).

Best regards,
Bido
chinasun
PLS User
Posts: 14
Joined: Thu Mar 20, 2008 1:41 pm
Real name and title:

Post by chinasun »

Thank you , professor
Could you tell me the principal of inverting the numerical values,
or provide some papers about it ?

I am looking forward. Thanks in advanced!
sterdelias
PLS Junior User
Posts: 1
Joined: Sun Apr 12, 2009 6:22 am
Real name and title:

loadings

Post by sterdelias »

is it possible for loadings to exceed 1 ?
User avatar
Diogenes
PLS Super-Expert
Posts: 899
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes »

Hi,

1) Some references about reverse coded item:
Nunnally, J. C. & Bernstein I. H. (1994). Psychometric Theory. 3rd Ed. New York: McGraw-Hill, Inc. (see p.326).
Netemeyer, R. G., Bearden, W. O., & Sharma, S. (2003). Scaling Procedures: Issues and Applications. Thousand Oaks: Sage Publications. (see p.99).
DeVellis, R. F. (2003). Scale Development: theory and applications. 2nd Ed. Thousand Oaks: Sage Publications. (see pp.91-92).
Spector, P. E. (1992). Summated Rating Scale Construction: an introduction. Newbury Park: Sage Publications. (see p.22).

2) Outer loading usually is standardized (like correlations between LV and its reflective indicators), for this reason, it is not expected that its value be greater than 1.
If you are using formative indicators, the outer weights could have values greater than 1 because the multicollinearity between them.

Best regards,

Bido
User avatar
Diogenes
PLS Super-Expert
Posts: 899
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes »

Hi,

A classical reference about this:

LIKERT, Rensis. 1932. A Technique for the Measurement of Attitudes. Archives of Psychology, n.140, pp.1-50.
That was republished in Spanish in 1976:
WAINERMAN. C. H. (compil.). 1976. Escalas de medición em ciências sociales. Buenos Aires: Ediciones Nueva Visión. pp.199-260.

He recommends the inversion of the scale when the item-total correlation is negative (p252).

Best regards,

Bido
Melvin
PLS Junior User
Posts: 3
Joined: Thu Jan 21, 2010 8:03 am
Real name and title:

Post by Melvin »

Diogenes wrote:Hi Miguel,
It is recommended 0,7 because 0,7^2 = 0,49
49% (almost 50%) of the variance of the indicator is explained by its LV.

Remenbering that AVE also should be > 50% (all loading = 0,7; the AVE will be 49%).

In practical cases we have some indicators lower than 0,7, but others bigger than 0,7, with AVE > 0,5;
in this case is common to keep the indicator even when its loading < 0,7, mainly in cases where we have just 3 or 4 indicators.

Best regards.
Bido
Hi, Professor Diogenes Bido,

Since, i'm not good at statistic. now i know the standard and i wanna know about others like AVE and T-value's Standard. And from where that i can get the reference about support such loading is acceptable. can professor give me some reference paper. thank you

Best regards
Melvin
iris_afandiphd
PLS Senior User
Posts: 28
Joined: Thu Jun 24, 2010 5:38 am
Real name and title:

Post by iris_afandiphd »

Hi Prof. Bido,

What happen if the outer loadings for INDICATORS are below 0.7?, what was the problem actually?, it is caused by most of responses across constructs are 1,2,3 , which group as Strongly Disagree etc..

Thanks
haslindar
PLS Junior User
Posts: 8
Joined: Sat Jan 30, 2010 3:40 am
Real name and title:

Post by haslindar »

Hi Prof Bido,

What happen if one of the variable's AVE is <0.5 while others are >0.5.

Thanks.
Locked