CODEWITHSUNDEEP
statisticsdata-analysisspssamos
sadini maheshika
sadini maheshika

Reputation: 1

AMOS The model is probably unidentified. In order to achieve identifiability, it will probably be necessary to impose 1 additional constraint

enter image description hereI am trying to make a SEM for my dataset using AMOS and I do get below error . Does anyone know why do i get this error .I am very new to this software

I wanted to build a structural equational model for my dataset . But I am not able to generate the results using AMOS

Upvotes: 0

Views: 1339

Answers (1)

hanna-without-h
hanna-without-h

Reputation: 189

Could you maybe share a picture of the model itself, i.e. the one you built?

Some basic potential error sources that come to mind are:

  • You did not fix the path coefficient of the error term (of the latent DV and the manifest items) to 1. If you don't do that, these parameters will need to be estimated as well and could thus lead to model-underidentification. In lavaan this is usually done automatically, but in AMOS it is possible to forget this step.
  • Your measurement models are locally underidentified. Maybe you used too few indicators per factor (you would need at least 3).
  • Your model is nonrecursive, meaning you have a loop or bidirectional effects in your model.
  • You have formative factors (arrows point from the items to the factor) instead of reflective factors (arrows point from the factor to the items).

There are instances where nonrecursive models, or models with formative factors can work but this can be very tricky. Rex B. Kline's chapter in Hancock & Mueller's "Structural Equation Modeling: A Second Course (2nd ed.)" discusses potential solutions in these cases.

Upvotes: 0

Related Questions