just as findings from primary intervention studies can be biased by missing data, missing data in
meta-analyses can bias estimates of intervention effect sizes and of relationships between moderator
variables and effect sizes. For example, information needed to calculate effect sizes for some or all of
the outcome variables and/or information needed to code moderator variables may not be reported,
and may not be obtainable. However, some methods, including multiple imputation methods, have
been developed to handle missing data in meta-analyses (Pigott, 2001, 2009).