Hello,
I conducted a model with 3 exogenous formative constructs and one endogenous construct with two variables. After checking for Multicolineratity, I also checked Indicator weights (significance) and loadings (weights and significance) and removed 13 from 29 indicators, 10 items are significant (recommendation according to Hair 2017). From these 13 formative indicators 5 (4),6 (4 sign) and 2 (2 sign.) per construct are left. Among the 5 and 6 indicator constructs there is always one negative item.
Since all VIF values are below the value of 2 I interpret this according to Cenfettili "The negative weight can then be interpreted as follows: when xxx, yyy and tangibles are otherwise equal, increased amounts of NEGATIVE INDICATOR will reduce the degree of CONSTRUCT".
However, to my question: Can i interpret the NEGATIVE (significant) INDICATOR in combination with a NEGATIVE (significant) PATH WEIGHT, e.g. that an increased amount of NEGATIVE INDICATOR together with the NEGATIVE PATHCOEFFICIENT leads to A POSITIVE EFFEKT / an increase of my dependent Variables (an Investment-sum and a number of measures)? And are 2 items for a reflective goal construct are okay? Do i have to do the multiple checks for my goal construct, too?
I would really appreciate your help!
Thank you!
HANSDAMPF3000
Neg. indicator weight + neg. path coeff -> pos. effect? ;)
-
- PLS User
- Posts: 15
- Joined: Sat Feb 04, 2017 3:34 pm
- Real name and title: Robert Boomhoff
-
- SmartPLS Developer
- Posts: 1284
- Joined: Tue Mar 28, 2006 11:09 am
- Real name and title: Dr. Jan-Michael Becker
Re: Neg. indicator weight + neg. path coeff -> pos. effect?
Generally, the interpretation of negative weight * negative path coefficient --> positive effect is correct. However, I would also check the loading of the indicator with the construct (i.e., the bivariate correlation). Is that also negative? Does it make sense that this indicator has a negative effect on the construct? Sometimes already subtle collinearity patterns (with low VIFs) can cause strange sign changes in regressions. I would always carefull evaluate if the effect is plausible.
Two indicators for a final construct are possible. Whether this is a good meaurement of your construct, however, is always a case-by-case decision.
Two indicators for a final construct are possible. Whether this is a good meaurement of your construct, however, is always a case-by-case decision.
Dr. Jan-Michael Becker, BI Norwegian Business School, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
-
- PLS User
- Posts: 15
- Joined: Sat Feb 04, 2017 3:34 pm
- Real name and title: Robert Boomhoff
Re: Neg. indicator weight + neg. path coeff -> pos. effect?
Hello and thank you, Dr. Becker!
Asked in an easier was: should both values be negative for a negative Interpretation? And what should be done if only one of Botz values is negative?
My VIFs are btw all below 2.
If this value is not okay and since the investment sum can be influenced by other measures than those i added up to get value 1, would it be okay to form 2 goal constructs with one dependent variable each whereas the 3 exogenous predictors are at the same time connected with endogenous construct a) with value 1 and endogenous construct b) with value 2?
I really appreciate your help!
Kind regards,
HansDampf3000
Unfortunately i don't have the data by my side at the moment, but if the weights and the loadings are negative, does this mean i can interpret it the way I did (a negative effect on the construct)? or what does a co-occurance of a negative (positive) weight and a positive (negative) loading mean? how should one interpret such a case? or should i delete the respective item?jmbecker wrote:Generally, the interpretation of negative weight * negative path coefficient --> positive effect is correct. However, I would also check the loading of the indicator with the construct (i.e., the bivariate correlation). Is that also negative? Does it make sense that this indicator has a negative effect on the construct? Sometimes already subtle collinearity patterns (with low VIFs) can cause strange sign changes in regressions. I would always carefull evaluate if the effect is plausible.
Asked in an easier was: should both values be negative for a negative Interpretation? And what should be done if only one of Botz values is negative?
My VIFs are btw all below 2.
My final construct consists out of a number of measures (value 1) and an investment sum (value 2) (are differnt scales okay within one construct?). since both stand for the extent of measures a person delivered I characterized this construct as reflective. Unfortunately the cronbachs alpha is with 0.5 quite low - or is this an acceptable value if there are only 2 items? (all other values for reflective constructs are okay)jmbecker wrote:Two indicators for a final construct are possible. Whether this is a good meaurement of your construct, however, is always a case-by-case decision.
If this value is not okay and since the investment sum can be influenced by other measures than those i added up to get value 1, would it be okay to form 2 goal constructs with one dependent variable each whereas the 3 exogenous predictors are at the same time connected with endogenous construct a) with value 1 and endogenous construct b) with value 2?
I really appreciate your help!
Kind regards,
HansDampf3000
-
- PLS User
- Posts: 15
- Joined: Sat Feb 04, 2017 3:34 pm
- Real name and title: Robert Boomhoff
Re: Neg. indicator weight + neg. path coeff -> pos. effect?
In the attachment is a picture of my oughter weights (left) and loadings (right) + the respective VIFs (all below 2,1)
At some items the loadings as well as the weights are negative, sometimes only the loadings or the weights (???)
How can this be interpret? Do all 5 items have a negative effect on the constructs? or only those where the loadings as well as the weights are negative?
Thank you for your help!
At some items the loadings as well as the weights are negative, sometimes only the loadings or the weights (???)
How can this be interpret? Do all 5 items have a negative effect on the constructs? or only those where the loadings as well as the weights are negative?
Thank you for your help!
Last edited by HansDampf3000 on Fri Jul 21, 2017 5:28 am, edited 1 time in total.
-
- PLS User
- Posts: 15
- Joined: Sat Feb 04, 2017 3:34 pm
- Real name and title: Robert Boomhoff
Re: Neg. indicator weight + neg. path coeff -> pos. effect?
Is there nobody who can help me and tell me how to interpret negative weights (and/or) negative loadings?
- cringle
- SmartPLS Developer
- Posts: 818
- Joined: Tue Sep 20, 2005 9:13 am
- Real name and title: Prof. Dr. Christian M. Ringle
- Location: Hamburg (Germany)
- Contact:
Re: Neg. indicator weight + neg. path coeff -> pos. effect?
Please check the correlation matrix of indicators. Are they all positive? If not, it may be a data problem or you need to rescale/recode the indicators.
Best
CR
Best
CR
Prof. Dr. Christian M. Ringle, Hamburg University of Technology (TUHH), SmartPLS
- Literature on PLS-SEM: https://www.smartpls.com/documentation
- Google Scholar: https://scholar.google.de/citations?use ... AAAJ&hl=de
- Literature on PLS-SEM: https://www.smartpls.com/documentation
- Google Scholar: https://scholar.google.de/citations?use ... AAAJ&hl=de