Dealing with Attrition and Missing Data in Longitudinal Studies: A Critique

Abstract

Longitudinal studies in teenage pregnancy, like any other topical issue, have been greatly affected by the problem of participant drop out and missing data. Although existing evidence indicate that data from longitudinal studies can provide useful insights regarding individual behaviours; the quality of the data and representativeness of the findings can adversely be compromised by attrition and missing data. Using literature review, this article criticises two papers on teenage pregnancy that have utilised longitudinal data by examining the effectiveness of the measures taken to correct for the attrition and missing data. The article concludes that procedural strategies used to eliminate or reduce attrition and missing data before and during data collection are more effective than employing statistical strategies to deal with the effects afterwards. If statistical procedures have to be used, it is important to first make distinctions of the missing data mechanisms since this has a bearing on whether certain strategies of handling missing data – such as list-wise deletion, pairwise deletion, mean imputation or multiple imputation –will result in biases or not.