|
Activity Number:
|
307
- Challenges and Advances in Psychological and Behavioral Data Analysis
|
|
Type:
|
Topic-Contributed
|
|
Date/Time:
|
Wednesday, August 11, 2021 : 3:30 PM to 5:20 PM
|
|
Sponsor:
|
Mental Health Statistics Section
|
|
Abstract #317039
|
|
|
Title:
|
Reporting Proficiency Levels for Examinees with Missing Data on Large-Scale Educational Assessments
|
|
Author(s):
|
Sandip Sinharay*
|
|
Companies:
|
Educational Testing Service
|
|
Keywords:
|
Multiple imputation;
Reliability;
Regression
|
|
Abstract:
|
Takers of educational tests often receive profi ciency levels instead of scores. For example, profi ciency levels are reported for the AP and United States Medical Licensing examinations. Technical difficulties and other unforeseen events often lead to missing item scores on these tests. The reporting of pro ficiency levels to the examinees with missing scores requires estimation of the performance of the examinees on the missing part and essentially involves imputation of missing data. In this paper, six approaches from missing data analysis are brought to bear on the problem of reporting of profi ciency levels to the examinees with missing scores. Data from four large-scale educational tests are used to compare the performances of the approaches to the approach that is operationally used for reporting profi ciency levels for these tests. A multiple imputation approach based on chained equations is shown to lead to the most accurate reporting of profi ciency levels for data that were missing at random or completely at random, while a model-based approach performed the best for data that are missing not at random.
|
Authors who are presenting talks have a * after their name.
