Both small-study effects approaches and selection models allow extensions to correct for the potential presence of bias. For example, one can estimate the extrapolated effect size for a study with theoretically infinite sample size that is immune from small-study bias [23]. For selection models, one may impute missing non-significant studies in meta-analyses to estimate corrected effect sizes [16]. Another popular approach is the trim-and-fill method, and variants thereof, for imputing missing studies [24]. Other methods try to explicitly model the impact of multiple biases on the data [25, 26]. While these methods are a focus of active research and some are even frequently used, their performance is largely unknown. They may generate a false sense of security that we can remedy distorted/selectively reported data. More solid approaches for correction of bias would require access to raw data, protocols, analyses codes, and registries of unpublished information, but this is usually not available.
