Should we constrain the thresholds in factor analysis of rating scale responses?

Sočan, Gregor (avtor)

In factor analysis of ordinal variables, category thresholds determine the values of a hypothetical latent response at which a transition to a higher response option occurs. In general, category thresholds are estimated as free parameters. In the Rasch measurement theory framework, the rating-scale model has been proposed that prescribes equal sets of distances between category thresholds for all items. As a factor analytic parallel, I propose a model with a constrained threshold structure (FACTS). The application of the model is illustrated with a real data example. A simulation study showed that the thresholds are estimated more accurately with the FACTS model than with the standard unconstrained model for different sample sizes, test lengths, and number of response categories. In addition, the likelihood ratio test generally showed good power in compar-ing the two models. Because the FACTS model performs well and provides a meaningful interpretation of category thresholds, it may be used routinely in factor analysis of categorical item responses.

• Jezik: Angleški jezik
• Ključne besede: rating scales, threshold, categorical data, factor analysis, simulation
• Vrsta gradiva: Članek v reviji
• Tipologija: 1.01 - Izvirni znanstveni članek
• Organizacija: FF - Filozofska fakulteta
• Status publikacije: Objavljeno
• Različica publikacije: Objavljena publikacija
• Leto izida: 2024
• Št. strani: Str. 159-170
• Številčenje: Vol. 31, no. 2
• PID: 20.500.12556/RUL-165614
• UDK: 159.9.072
• ISSN pri članku: 1972-6325
• DOI: 10.4473/TPM31.2.2
• COBISS.SI-ID: 201807363
• Datum objave v RUL: 10.12.2024

Gradivo je del revije

• Naslov: TPM Testing, psychometrics, methodology in applied psychology
• Skrajšan naslov: TPM Test., psychom., methodol. appl. psychol.
• Založnik: CISES
• ISSN: 1972-6325
• COBISS.SI-ID: 119631363

Licence

• Licenca: CC BY-NC-ND 4.0, Creative Commons Priznanje avtorstva-Nekomercialno-Brez predelav 4.0 Mednarodna
• Povezava: http://creativecommons.org/licenses/by-nc-nd/4.0/deed.sl
• Opis: Najbolj omejujoča licenca Creative Commons. Uporabniki lahko prenesejo in delijo delo v nekomercialne namene in ga ne smejo uporabiti za nobene druge namene.

Sekundarni jezik

• Jezik: Slovenski jezik
• Ključne besede: ocenjevalne lestvice, prag, kategorialni podatki, faktorska analiza, simulacija

Projekti

• Financer: ARIS - Javna agencija za znanstvenoraziskovalno in inovacijsko dejavnost Republike Slovenije
• Številka projekta: P5-0062
• Naslov: Uporabna razvojna psihologija