Contextual Responses to Affirmative and/or Reversed-Worded Items.

Area of Science:

• Psychometrics
• Social Sciences

Background:

• Response biases, such as yea-saying, can influence survey results.
• Item response theory (IRT) models are used to analyze survey data.
• Differential item functioning (DIF) can occur due to various factors, including item presentation.

Purpose of the Study:

• To systematically investigate how affirmative and polar-opposite items, presented jointly or separately, affect yea-saying tendencies.
• To re-analyze existing data using item response models to understand yea-saying effects.
• To examine how yea-saying tendencies impact the dimensionality of trait measurement.

Main Methods:

• Utilized item response models to estimate yea-saying tendencies.
• Re-analyzed data from Zhang et al. (2016).
• Compared yea-saying tendencies for affirmative and polar-opposite items under different presentation conditions (joint vs. separate).

Main Results:

• Yea-saying tendencies are stronger for affirmative items compared to polar-opposite items.
• Presenting polar-opposite items with affirmative items reduced yea-saying tendencies for the polar-opposite items.
• Standard IRT models ignoring these yea-saying effects can lead to inaccurate trait dimensionality (e.g., a two-dimensional representation).

Conclusions:

• Item presentation context significantly influences yea-saying tendencies.
• Contextual information within item scales acts as a determinant of differential item functioning.
• Accurate measurement requires accounting for yea-saying effects influenced by item format and context.