10th International
Conference on
Teaching Statistics
8 – 13 July 2018
Kyoto, Japan
Contributed paper list

Contributed Paper C220


An Application of the Mixture Rasch Model: Classifying Pre-service Teachers on the Self-Efficacy to Teach Statistics


Leigh Harrell-Williams (USA)
Jennifer Lovett (USA)
M. Alejandra Sorto (USA)
Rebecca Pierce (USA)
Lawrence Lesser (USA)
Teri Murphy (USA)


Leigh Harrell-Williams (United States)


Rasch mixture modeling identifies latent class membership and assesses differential item functioning. From homogeneities in item response patterns on the Level A subscale of the Self-Efficacy to Teach Statistics - Middle Grades (SETS-MS) instrument, this methodology identified four distinct latent classes for n= 588 pre-service middle and secondary mathematics teachers (PSTs). One class had consistently high efficacy; one had consistently low efficacy. Item location parameters indicated that the four items related to creating/using graphical displays contributed to differences as large as 1.5 logits between the middle two classes. Results somewhat contradict previous findings that PSTs, on average, are more confident about teaching graphical displays than other topics. Hence, evaluation of statistics teaching self-efficacy should possibly include examination of item-level responses to specific content areas.