Ameliorating culturally based extreme response tendencies to attitude items

Area of Science:

• Educational Measurement
• Psychometrics
• Cross-Cultural Psychology

Background:

• Cross-cultural studies using Likert scale attitude items often face challenges with response biases.
• Extreme response tendency (ERT) can inflate or deflate scale scores, impacting validity.
• Understanding and addressing ERT is crucial for accurate international comparisons.

Purpose of the Study:

• To investigate the differential manifestation of extreme response tendency (ERT) across culturally distinct countries.
• To explore methods for ameliorating ERT in Likert scale attitude items.
• To assess the impact of ERT on attitude scale outcomes and achievement correlations.

Main Methods:

• Utilized Rasch item response models with data from the PISA 2006 field trial.
• Examined a single attitude scale across eight culturally distinct countries.
• Compared dichotomous item variants with Likert variants and incorporated country-specific ERT into models.

Main Results:

• ERT was exhibited differentially across countries when answering Likert-type attitude items.
• Hierarchical facet models showed significantly better data fit when including an interaction between country and item difficulty.
• Likert variants were preferred over dichotomous variants when considering test information.

Conclusions:

• Cross-cultural reporting of attitudes measured with Likert items requires careful consideration of country-specific response tendencies.
• Accounting for the interaction between country and item parameters improves model fit.
• The study highlights the importance of psychometric adjustments for valid cross-cultural attitude assessment.