To parcel or not too parcel

[Arjan van Dam]

It is recommended to parcel when scales are unidimensional and items (Likert-type) are used in SEM. I tried this, it resulted in very different estimates and p-values than when the items were used as indicators. Has anyone a clue? 

[Mikko Rönkkö]

Hi,

Parceling can mask misspecification. If the model is correctly specified and your sample is sufficiently large, the estimates between the LVs should be very close regardless of whether you parcel or not.

Without additional details of your model and data, it is difficult to say much more.

I recommend that you take a look at these if you have not already

Little, T. D., Rhemtulla, M., Gibson, K., & Schoemann, A. M. (2013). Why the items versus parcels controversy needn’t be one. Psychological Methods, 18(3), 285–300. http://doi.org/10.1037/a0033266

Marsh, H. W., Lüdtke, O., Nagengast, B., Morin, A. J. S., & Von Davier, M. (2013). Why Item Parcels Are (Almost) Never Appropriate: Two Wrongs Do Not Make a Right—Camouflaging Misspecification With Item Parcels in CFA Models. Psychological Methods. http://doi.org/10.1037/a0032773

Mikko

On 27 Jul 2015, at 17:52 , Arjan van Dam <arjan...@googlemail.com> wrote:

It is recommended to parcel when scales are unidimensional and items (Likert-type) are used in SEM. I tried this, it resulted in very different estimates and p-values than when the items were used as indicators. Has anyone a clue? 

--
You received this message because you are subscribed to the Google Groups "lavaan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lavaan+un...@googlegroups.com.
To post to this group, send email to lav...@googlegroups.com.
Visit this group at http://groups.google.com/group/lavaan.
For more options, visit https://groups.google.com/d/optout.

[Arjan van Dam]

Hi Mikko, 

Thanks, article of Marsh was unknown to me, I am going to read it. I hope it will shine some light. 

Keep you posted!

Greetings, 

Arjan 

Op maandag 27 juli 2015 16:55:57 UTC+2 schreef Mikko Rönkkö:

[Arjan van Dam]

Hi Mikko, 

The article of Marsh et al. focuses on unidimensionality, but the scales I use are well validated and unidimensional. I found a notion in Principles and Practices of Structural Equation Modeling of Rex Kline (2011) on Heywood cases: "A combination of small sample sizes (e.g., N < 100) and only two indicators per factor". Although my sample is bigger (N=275) I made a mistake by creating two parcels per indicator, this might result in the strange estimates. Don't exactly how it works, but I suspect that this is the case.  

Greetings, 

Arjan 

Op maandag 27 juli 2015 17:51:34 UTC+2 schreef Arjan van Dam: