IRT-based repeated-measurement linking with re-randomized groups and concurrent links¶
Research Method · All research methods
Evidence · 3 claims (3 for) · 2 studies (1 design, 1 theoretical),q2· 0 of 2 report an effect size · 3 claims rest on one study
Description¶
The article recommends an IRT framework for its repeated measurement linking design because IRT "easily allows for incomplete designs and for equality constraints between item parameters." The author notes that when new randomized groups are formed at each occasion and all links are put to work in concert, a classical test theory framework may be less suited. This makes the design practical for longitudinal growth measurement.
Accounts¶
- Use IRT rather than classical test theory when implementing repeated measurement linking designs with re-randomized groups and multiple concurrent links: The article recommends an IRT framework for its repeated measurement linking design because IRT "easily allows for incomplete designs and for equality constraints between item parameters." The author notes that when new randomized groups are formed at each occasion and all links are put to work in concert, a classical test theory framework may be less suited. This makes the design practical for longitudinal growth measurement. (Rijmen (2010))
- When test design changes may introduce multidimensionality, prefer separate calibration with as few equating transformations as possible over concurrent calibration: The article recommends considering separate calibration with minimal equating transformations when accumulated tests share overlapping items but blueprint or item-type changes may violate IRT assumptions. It notes that "concurrent calibration is more accurate when the data fit the IRT model" but is "less robust to violations of the IRT assumptions due to multidimensionality," and its results showed the separate-calibration transformations (CH and SM) agreed closely while concurrent calibration diverged. The study operationalized this by transforming multiple tests simultaneously (SM) to minimize equating steps. (Jungnam Kim et al. (2017))
- Repeated measurement linking design combining equivalent groups with common items over time: A longitudinal data collection design in which randomly equivalent groups receive parallel forms at each occasion, with item blocks rotated across occasions so that each occasion shares common items with the previous and next occasion. As the article summarizes, "The common items were common across measurement occasions but never presented to the same individual twice." This is possible because the design is incomplete at each measurement occasion, and it allows multiple links whose assumptions can partly be tested. (Rijmen (2010))
Claims¶
- Meaningful longitudinal linking depends on the construct not changing across measurement occasions [+W]
- Concurrent calibration yielded consistently lower and less fluctuating scale score means than separate calibration with chain or simultaneous equating, especially for older tests [+M]
- Item parameter differences between chain and simultaneous equating were small (at most 12 items, 7%, flagged by RMSD), while comparisons involving concurrent calibration flagged up to 11% of items [+M]
Related Research Methods¶
-
Key Sources¶
- Rijmen, F. (2010). Measuring Multidimensional Latent Growth. ETS Research Report RR-10-24. https://www.ets.org/research/contact.html
- Jungnam Kim, Furong Gao, & Dong-In Kim. (2017). Robust IRT Scaling: Considerations in constructing item bank from tests across years. Paper presented at the annual meeting of the National Council on Measurement in Education, San Antonio, TX.
Page metadata
| Field | Value |
|---|---|
type |
research-method |
id |
irt-based-repeated-measurement-linking-with-re-randomized-groups-and-concurrent-links |
title |
IRT-based repeated-measurement linking with re-randomized groups and concurrent links |
description |
The article recommends an IRT framework for its repeated measurement linking design because IRT "easily allows for incomplete designs and for equality constraints between item parameters." The author notes that when new randomized groups are formed at each occasion and all links are put to work in c |
status |
review |
moved_from |
element/repeated-measurement-linking-design-equivalent-groups-common-items, principle/irt-preferred-for-repeated-measurement-linking |
generated |
by process:settle-candidates on 2026-10-09 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — rijmen-2010 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.