An Examination of Result Pooling for Reporting t Test Findings From Multiple Imputation

• In Event: AERA Poster Session 4
  In Poster Session: Poster Session (San Diego)

Abstract

Unprecedented opportunities have been created by Multiple imputation (MI) for missing data treatment. In practice, While some of the imputed data suggest the use of Welch/Satterthwaite t test after rejecting a null hypothesis from Levene’s test, others may support application of Student’s t test under a condition of “equal variances assumed”. Software packages offer the result pooling in each side. This study demonstrates that the needs of incorporating mixed result pooling to reflect heterogeneity of imputed data from MI. The result partitions can avoid blind interpretation of SAS or SPSS printout and report biased conclusions in educational research.

Author

• Jianjun Wang, California State University - Bakersfield