Item discrimination moderates detection differently by index: NCDIF benefits from high a-parameters only at large DIF magnitude¶

Claim · All claims
Evidence · 1 study · causal r2 · q2
Currency · newest study 1997 · not yet checked against its sources · not reviewed by a person · overdue for a recheck since 2000: tool and model accuracy changes quickly
Evidence quality · No study behind this claim is a peer-reviewed article with a DOI: it rests on an ERIC document (a report, conference paper or thesis). It is kept because a design page cites it; treat it as a lead to check, not an established finding.

Subclaims¶

q2 i? At .25 DIF magnitude, high-discriminating items had better NCDIF detection rates, but this trend did not appear at .10 magnitude; SIBTEST and Lord's chi-square detection rose with discrimination in all conditions. → Flowers 1997

Evidence¶

Flowers 1997¶

Flowers, C. P., Oshima, T. C., & Raju, N. S. (1997). The Relationship between Polytomous DFIT and Other Polytomous DIF Procedures. Paper presented at the NCME Annual Meeting, Chicago. https://eric.ed.gov/?id=ED410300

q2 · i? · causal · r2

Per-item detection rates from Table 5 for studied items with a-parameters of .55, .75, 1.00, and 1.36. The moderation by DIF magnitude is a scope limit the article states explicitly.

"When the magnitude of DIF was .25, high discriminating items had a better detection rate for NCDIF. This trend was not noted in the .10 magnitude condition. Both SIBTEST and Lord's chi-square had higher detection rates as the item discrimination increased in all conditions."

Discussion¶

Related Claims¶

• DIF detection rates for all indices are higher with larger samples, equivalent distributions, fewer DIF items, greater DIF magnitude, and larger a-parameters — related
• The polytomous DFIT framework shows Type I error rates close to nominal alpha except when the number of DIF items and DIF magnitude are highest — related
• Lord's chi-square identified more DIF items than Raju's unsigned area in the without-purification condition — related
• The chi-square significance test for NC-DIF is overly sensitive at large sample sizes, requiring an empirical cutoff value — related
• Type of DIF affected detection: nonuniform DIF items with higher a-parameters were not detected whereas lower a-parameter items were — related
• Type I error rates of all three polytomous DIF indices increase as the number of DIF items and DIF magnitude increase — related
• Test length, focal group distribution, number of DIF items, and direction of DIF had little effect on detection rates — related