This indicator is coming from the summation of answers to relevant survey questions.
(i.e Urogenital = Q23(Survey question result ) +Q25 + Q27 + Q28 + Q30 + Q32 + Q 35 + Q37
example a respondent "Rachel" has Urogenetial value 42 because she answered 3 for Q23 and 5 for Q25 ... )
There are some indicators that have a lower standardized regression weight than 0.5.
However, for my research, I want to keep the variable in my Structured Equation Model(SEM).
What kind of option I can have to keep this variable more reasonably(?), and more rationally?
I have thought about some options.
1. JUST keep it by providing some theoretical background ( not statistical reason).
In this case, what kind of theoretical background I should find?
2. By removing the sub-question (i.e Q23) in the Urogenenital variable, I put the Urogenital_2(excl Q23) variable instead of the Urogenital.
( assume Urogenital_2 has better-standardized regression weight).
However, I'm not comfortable with this option because this survey tool is used in a different direction than intended.
3. If you know any solution for this case, please let me know.
Thank you,
Bests,
Jiwon
Hulland (1999) argued that 0.4 should be the acceptable loading value where Henseler et al.(2009) suggested that manifest variable with loading values between 0.4 and 0.7 should be reviewed before elimination.
Note : if your AVE and CR are not above the cut-off value then you need to delete those items which are less than 0.7.
Hulland J. Use of Partial Least Squares (PLS) in strategic management research: a review of four recent studies. Strategic Management Journal. 1999;20(2):195–204.
Henseler J, Ringle CM, Sinkovics RR. The use of partial least squares path modeling in international marketing. Advances in International Marketing. 2009;20:277–319.
In addition standardized regression weight greater than 1 is problematic. it is called " Heywood case " and you need to fix it !
Hi Jiwon,
when I see the indicators, I have strong doubts that a factor model is appropriate. What is the fit of your model. Please consider that parameter estimates (like factor loadings) are only reasonable when the model structure is correct. Hence, your low loadings may stem from the problem of a misspecification which should motivate you hopefully to pay attention to that.
Best,
Holger
Alireza shabani shojaei
Your answers have been of great help to me. Can I ask you one more question? Can the error term of the observed variable within the same potential variable be linked according to the principle to increase the model fit in the confirmatory factor analysis?
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression