The Effects of Sample Selection on Item Parameter Estimation

• In Event: Roundtable Session 26
  In Roundtable Session: Rasch Measurement SIG Roundtable Session

Abstract

Research has shown that item parameter estimation may be affected by different sampling conditions. However, few research studies have investigated scenarios in K-12 assessments, where convenience samples, which is unpredictable and may result in extremely skewed score distributions, are often used for analyses. In these cases, sampling may have to be done within a larger convenience sample to obtain a representative sample.
This simulation study investigates the item parameter estimations with two kinds of samples: skewed convenience samples and representative samples that are drawn from the convenience sample. The purpose of the study is to determine if the convenience sample is skewed, can the estimation with a representative sample drawn from the skewed sample produce more accurate results.

Authors

• Lixiong Gu, ETS

• Venessa F. Lall, ETS

• Maxwell D. Wise, ETS