Polytomous item response theory parameter recovery: An investigation of non-normal distributions and small sample size

• Author / Creator

  Bahry, Louise M

• Item Response Theory (IRT) has been extensively used in educational research with large sample sizes and normally distributed traits. However, there are cases in which distributions are not normal, and research has shown that the estimation of parameters becomes problematic with non-normal data. This study investigates the effects of skewness on parameter estimation using the Graded Response Model (GRM) and MULTILOG. Three distribution types (extreme and moderate skewness and a baseline condition (i.e. normal) and seven sample sizes (from n = 100 to n = 3,000) were investigated using simulations. In keeping with previous findings, the extremely skewed distribution condition resulted in the poorest estimates regardless of sample size. In general, the accuracy of parameter estimation increased as sample size increased. For the normally distributed conditions, results suggest a minimum sample size of 750 for accurate estimation. Implications of these findings are discussed.

• Subjects / Keywords

  • Item Response Theory
  • Parameter Recovery
  • Graded Response Model

• Graduation date

  Spring 2012

• Type of Item

  Thesis

• Degree

  Master of Education

• DOI

  https://doi.org/10.7939/R36H8X

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.