Wow! This is really spectacular. Thanks so much for sharing this. And it's Mac compatible. Do you know if it runs multilevel models for couples/groups?🙂
I haven't tried this myself but I would think that it should work via lavaan (or other R) syntax that you can use within the SEM module in JASP. Best, Christian Geiser
@@QuantFish Thank you. I think I'm not describing the multilevel modeling approach correctly - I'll be using the regression based approach (I think it's called) using dyadic data rather than the SEM approach due to small sample size. I have Stata so I can use that. It's also really exciting that JASP has a beta version of the PROCESS Macro by Hayes. Thank you again.
Yes, it does! In that case, under Advanced --> Estimator choose DWLS (a proper estimation method for binary and ordinal data) and under Advanced --> Missing data handling, choose Pairwise. Best, Christian Geiser
@@QuantFish Thank you! I'm curious if you might know the answer to this- I tested a dataset in R vs JASP where my item response matrix was not complete (e.g., 5-point likert scale where response 5 [always] had 0 responses in a subgroup). In R, this does not run successfully when treated ordinally and necessitated collapsing my data. in JASP, it ran the model. Do you have any insight about why that might function? The syntax in JASP doesn't look too different...
You remain a treasure my friend!
Great! Is the SEM covariance-based? Thanks!
Yes, it's based on the lavaan package in R.
Best, Christian Geiser
Wow! This is really spectacular. Thanks so much for sharing this. And it's Mac compatible. Do you know if it runs multilevel models for couples/groups?🙂
I haven't tried this myself but I would think that it should work via lavaan (or other R) syntax that you can use within the SEM module in JASP.
Best, Christian Geiser
@@QuantFish Thank you. I think I'm not describing the multilevel modeling approach correctly - I'll be using the regression based approach (I think it's called) using dyadic data rather than the SEM approach due to small sample size. I have Stata so I can use that. It's also really exciting that JASP has a beta version of the PROCESS Macro by Hayes. Thank you again.
Super interesting! For the MGCFA, does it support MGCFAs for ordinal data (e.g., something with a 5-point likert-type scale)?
Yes, it does! In that case, under Advanced --> Estimator choose DWLS (a proper estimation method for binary and ordinal data) and under Advanced --> Missing data handling, choose Pairwise.
Best, Christian Geiser
@@QuantFish Thank you! I'm curious if you might know the answer to this- I tested a dataset in R vs JASP where my item response matrix was not complete (e.g., 5-point likert scale where response 5 [always] had 0 responses in a subgroup). In R, this does not run successfully when treated ordinally and necessitated collapsing my data. in JASP, it ran the model. Do you have any insight about why that might function? The syntax in JASP doesn't look too different...
@@wildmanstinger I'm not sure. This would require a deeper look at your data, syntax, etc.
Best, Christian Geiser
👏🏻