Exploring how estimates differ when using student attendance versus student enrollment as the basis for your sampling weights

• In Event: Roundtable Session (Wednesday II)
  In Refereed Round-Table Session: The importance of sampling weights and practical considerations for their use in research

Proposal

One of the most commonly used approaches to weighting is to weight sample groups toward their known proportions in the population. However, what happens when there is some doubt about the “known” proportion? For example, education research that involves students often relies on enrollment data provided by schools for determining weights, even when these enrollment data may be inaccurate or inconsistent with what enumerators see when they visit the school. School-to-School International explored this issue by examining Early Grade Reading Assessment data from several countries and comparing mean-score results derived from weights that used student enrollment data with those derived from weights that used student attendance data on the day of the assessment. Findings from this exploration provide practical lessons related to the importance of weighting and the data used to determine weights.

Author

• Peter Cooper, School-to-School International