Parallel Optimal Calibration of Mixed-Format Items for Achievement Tests

Area of Science:

• Educational Measurement
• Psychometrics
• Test Theory

Background:

• Large-scale achievement tests require regular item calibration for operational use.
• Existing optimal assignment methods primarily address sequential examinee arrivals.
• Simultaneous or parallel test administration is common in calibration settings.

Purpose of the Study:

• To develop an optimal calibration design for parallel test setups.
• To investigate the efficiency gains of the proposed method.
• To demonstrate the method's applicability in real-world calibration scenarios.

Main Methods:

• Developed an optimal calibration design tailored for parallel test administration.
• The method handles mixed-format items and accounts for varying expected response times.
• Applied the method to calibrate items for Swedish national mathematics tests.

Main Results:

• The proposed method significantly enhances calibration efficiency.
• Demonstrated successful implementation in a real-world case study.
• The design is effective for mixed-format tests with diverse response time expectations.

Conclusions:

• The new optimal calibration design is efficient and practical for parallel test setups.
• This method offers substantial improvements for calibrating items in large-scale assessments.
• The approach is suitable for mixed-format tests, addressing practical challenges in educational measurement.