A comparison of procedures for handling mobility and missing level-2 identifiers in two-level data
by Lindsey J. Wolff Smith; S. Natasha Beretvas
International Journal of Quantitative Research in Education (IJQRE), Vol. 2, No. 2, 2014

Abstract: This simulation study compared three ad hoc procedures for handling mobility of level-1 units across level-2 units and handling missing level-2 (here, school) identifiers for some level-1 units (students). The HLM-Delete procedure used the conventional hierarchical linear model (HLM) to analyse a dataset from which mobile and missing-identifier students had been removed. The MMREM-Delete procedure involved removing only students who were missing level-2 identifiers and estimating a multiple membership random effects model (MMREM). The third procedure (MMREM-Unique) involved no deletion and entailed creation of a pseudo-identifier for each missing level-2 identifier. All three procedures omitted a predictor of mobility from the estimating model. To provide a baseline comparison, a correctly specified MMREM was also estimated. Relative parameter bias was calculated for each parameter estimated. Across conditions, each procedure had some level of substantial bias with HLM-Delete performing worst. Results and implications for applied researchers are discussed.

Online publication date: Sat, 30-Aug-2014

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Quantitative Research in Education (IJQRE):
Login with your Inderscience username and password:

    Username:        Password:         

Forgotten your password?


Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com