2011
DOI: 10.1080/10705511.2011.607721
|View full text |Cite
|
Sign up to set email alerts
|

Missing Data Imputation versus Full Information Maximum Likelihood with Second-Level Dependencies

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
123
0
1

Year Published

2015
2015
2023
2023

Publication Types

Select...
9

Relationship

0
9

Authors

Journals

citations
Cited by 219 publications
(124 citation statements)
references
References 15 publications
0
123
0
1
Order By: Relevance
“…MMSE scores in the retained sample were stable over the course of study (mean latent change = −0.02, p = 0.66; individual differences in change = 0.02, p = 0.15) and remained high (see Table 1). Thus, the missing longitudinal data were treated as missing at random and were handled via full information maximum likelihood (FIML), a method that utilizes all available data to optimize estimation without imputation (Muthén et al 1987; Larsen 2011). …”
Section: 0 Materials and Methodsmentioning
confidence: 99%
“…MMSE scores in the retained sample were stable over the course of study (mean latent change = −0.02, p = 0.66; individual differences in change = 0.02, p = 0.15) and remained high (see Table 1). Thus, the missing longitudinal data were treated as missing at random and were handled via full information maximum likelihood (FIML), a method that utilizes all available data to optimize estimation without imputation (Muthén et al 1987; Larsen 2011). …”
Section: 0 Materials and Methodsmentioning
confidence: 99%
“…We used full information maximum likelihood estimation (FIML) to handle missing data because this procedure produces robust parameter estimates using all of the information available in the data (Muthén & Muthén, 1998. When performing multilevel analyses, the FIML procedure has shown to perform equivalently or even better with regard to producing unbiased estimates for missing data than multiple imputation procedures (Larsen, 2011). …”
Section: Missing Datamentioning
confidence: 99%
“…FIML relies on data missing at random (MAR), and research has found that this method resulted in unbiased parameter estimates, even in some cases that violated this assumption (e.g. Enders & Bandalos, 2001;Graham, 2009;Larsen, 2011 Table 2 shows the results that allow us to evaluate hypotheses 1 to 4 concerning direct transfer and sex-role learning. We start with the simplest model, in which parents' characteristics are constrained to be equal across groups (model 1).…”
Section: Analysesmentioning
confidence: 99%