Claims

The ICC and LDF methods identify substantially different sets of items as exhibiting DIF on a state science performance assessment

Claim · All claims
Evidence · 1 study · associational r2 · q2
Currency · newest study 1998 · not yet checked against its sources · not reviewed by a person · overdue for a recheck since 2001: tool and model accuracy changes quickly

Subclaims

Reading the evidence codes

q is the design tier, i the impact, n the sample, and the kind with its r the rigour judged within that kind: what each code means.

This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.

q2 i? On a state-level performance assessment, only items 3, 4 and 5 were flagged by both methods; four items were unique to LDF and four unique to ICC. → Holweger 1998

Evidence

Holweger 1998

Holweger, Nancy; Weston, Timothy. (1998). Differential Item Functioning: An Applied Comparison of the Item Characteristic Curve Method with the Logistic Discriminant Function Method. https://eric.ed.gov/?id=ED422362

q2 · i? · associational · r2

Applied comparison on a state-level performance-based assessment with approximately 16,000 examinees (8,539 females, 8,029 males after removing zero scores). The authors report "there is no uniformity across techniques" in the seven-item lists each method produced, with only partial overlap.

"Items 3, 4 and 5 are identified by both techniques as having substantial DIF. However, items 12, 29, 13, and 24 show DIF using the LDF technique but not the ICC technique. Items 18, 23, 25 and 26 show DIF using the ICC technique but not the LDF technique."

Studies bearing on this claim

How far does a statistical item-bias (DIF) flag identify a biased item, rather than the method, sample or form that produced it? This table gathers the studies behind this claim and the 6 narrower claims that bear on it: 5 distinct studies (3 associational, 2 causal), q1–q2, 0 of 5 with an effect size coded. One row per study, codes as each entry records them; no effect is averaged and no verdict is drawn. Built by scripts/synthesize_claim.py from eval/synthesis/; a row that looks wrong is fixed on the claim it comes from.

Study Design Effect as recorded Bears on this claim through
Holweger 1998 q1 / q2 · associational · r2 i? this claim; The assessment violates IRT assumptions of unidimensionality and local independence, which (qualifies)
Srdjan Verbić & Boris Tomić 2008 q2 · associational · r1 i? DIF was found for four items but could not be anticipated from differential response times (qualifies)
Soysal 2021 q2 · associational · r2 i? Item purification made both DIF methods more sensitive, flagging more items (supports); Some items positioned close together across booklets were also flagged as DIF, possibly du (supports)
Flowers 1996 q2 · causal · r2 i? The chi-square significance test for NC-DIF is overly sensitive at large sample sizes, req (supports)
Flowers 1997 q2 · causal · r2 i? Type I error rates of all three polytomous DIF indices increase as the number of DIF items (supports)

Discussion

Page metadata
Field Value
type claim
id icc-and-ldf-methods-identify-different-dif-items
title The ICC and LDF methods identify substantially different sets of items as exhibiting DIF on a state science performance assessment
description The ICC and LDF methods identify substantially different sets of items as exhibiting DIF on a state science performance assessment
status review
evidence_strength moderate
newest_evidence 1998
recheck_by 2001
generated by process:wiki-ingest on 2026-09-26
trust unverified — no one has confirmed this page's content
sources 1 — holweger-1998

Frontmatter is the page's machine-readable record — see the Schema & Guide.