This Article Abstract Full Text Full Text (PDF) An erratum has been published Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Enders, C. K. Search for Related Content PubMed PubMed Citation Articles by Enders, C. K. Related Collections Statistical Corner

A Primer on the Use of Modern Missing-Data Methods in Psychosomatic Medicine Research

From Arizona State University, Tempe, Arizona.

View larger version (23K): [in a new window]   Figure 1. Path diagram representation of the hypothetical regression model. The top panel of the Figure represents the standard multiple regression model with two predictors. The bottom panel of the figure represents the same regression model but includes four auxiliary variables (qol2, qol3, age, ed). Note that the auxiliary variables are correlated with (a) each other, (b) the substantive predictors (qol1 and tx), and (c) the residual term.

View larger version (35K): [in a new window]   Figure 2. Path diagram representation of the latent growth curve model. Note that the auxiliary variables are correlated with (a) each other, (b) the substantive predictor (tx), and (c) the residual terms of each repeated measure. Note also that the residual covariance between the growth factors was omitted from the figure but was estimated in the analysis.

View larger version (16K): [in a new window]   Figure 3. Estimated growth trajectories from the example analyses. The top panel compares the growth trajectories for DML and LW, whereas the bottom panel gives the estimated growth curves for DML and LOCF. The DML growth curves were virtually identical to those of the complete data, whereas both LW and LOCF produced distorted estimates of the trajectories.