ABSTRACT
More than 30 years ago, Rubin (1976) demonstrated the negative impact of missing data in a database on the validity of results. Results from large-scale education assessments can therefore be invalid if missing data are not treated correctly. Some authors, such as Fichman and Cum-mings (2003) and Enders (2004), mention that in spite of the potential bias caused by missing data, very few scientific papers identify the methods by the researchers to compensate for those missing data. This is even more relevant in the context of studies in education.
