Regression with missing Ys: An improved strategy for analyzing multiply imputed data

By Paul T. von Hippel. 2007

1,403 citations recorded by the source corpus.

When fitting a generalized linear model—such as linear regression, logistic regression, or hierarchical linear modeling—analysts often wonder how to handle missing values of the dependent variable Y. If missing values have been filled in using multiple imputation, the usual advice is to use the imputed Y values in analysis. We show, however, that using imputed Ys can add needless noise to the estimates. Better estimates can usually be obtained using a modified strategy that we call multiple imputation, then deletion (MID). Under MID, all cases are used for imputation but, following imputation, cases with imputed Y values are excluded from the analysis. When there is something wrong with the imputed Y values, MID protects the estimates from the problematic imputations. And when the imputed Y values are acceptable, MID usually offers somewhat more efficient estimates than an ordinary MI strategy.

Subfields: Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/d8de61eb7bf6288c7f819cc8eb52ad8778c0f661

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistics and Probability

Citations in the corpus

The Impact of Dual Enrollment on College Degree Attainment (2013, background): “I include the dependent variable in the imputation approach but later remove imputed values of the dependent variable because these values add little to regression estimates and may induce noise in estimates (von Hippel, 2007).”
Eviction’s Fallout: Housing, Hardship, and Health (2015, methodology): “We include both treatment and outcome measures in the imputation equation but in our analyses do not use imputed outcomes (von Hippel 2007).”
Pooling ANOVA Results From Multiply Imputed Datasets (2016, methodology): “Second, likelihood-based methods may be considered, which oﬀer some advantages over LD, for example, to include auxiliary variables or to condition on possible causes of missing data (see Enders, 2010; Little & Rubin, 2002; von Hippel, 2007).”
Rebutting Existing Misconceptions About Multiple Imputation as a Method for Handling Missing Data (2019, background): “Furthermore, we have found references (e.g., Von Hippel, 2008) that do not express a misconception but that could easily be misinterpreted by researchers, causing misconceptions to form or persist.”
What Improves with Increased Missing Data Imputations? (2008, background): “…along with the corresponding variability in the critical value of the reference distribution, generates the imputation variability O h m and O p m for independent runs of m stochastic imputations. von Hippel (2007) provided an expression demonstrating that the variance of O T m decreases with m .”
The Immigrant Advantage in Adolescent Educational Expectations (2016, methodology): “We did not impute immigrant generation (a key independent variable) and we utilized the dependent variable to impute missing data in our covariates, but did not use imputed values in our dependent variable for our final analysis (Von Hippel 2007).”
Are “Failing” Schools Really Failing? Using Seasonal Comparison to Evaluate School Effectiveness (2008, methodology): “Excluding imputations of the dependent variable is a strategy known as multiple imputation, then deletion (MID), which increases efficiency and reduces biases resulting from misspecification of the imputation model (von Hippel 2007).”
Participation in learning and wellbeing among older adults (2011, methodology): “Whilst it was essential to include the dependent variable in the imputation step in order to obtain good estimates, it has been shown that including imputed values of the dependent variable in the subsequent analysis adds unnecessary noise to the regression model estimates (von Hippel 2007 ), and s”