Longitudinal Survey in United Kingdom: Assessing Construct, Method - - PowerPoint PPT Presentation

An Integrated Approach to Bias in a Longitudinal Survey in United Kingdom:

Assessing Construct, Method and Item Bias in the General Health Questionnaire (GHQ-12).

Isabel Benítez, Byron Adams & Jia He

Outline contents

 Background: Bias in cross-cultural studies  Aim of the Project  Design of the study  Results  Conclusions & discussion

Background

A general common aim is to compare different cultural and linguistic groups Valid comparisons require evidence of the equivalence level reached The equivalence depends on the presence of bias Cross- cultural studies Equivalence Bias

Bias concept and evaluation

BIAS is defined as any nuisance factor in score comparisons (Van de Vijver & Tanzer, 2004) or as the presence of differences in a measurement instrument whose meaning is not exactly the same across cultures (Poortinga, 1989).

Three levels Construct Method Item

Van de Vijver & Tanzer, (2004)

“Gaps” in the bias interpretation

• Analyses are conducted

separately for different sources of bias

Lack of integrated results

• Knowledge about the

evolution of bias

• Identification of systematic

elements behind the presence of bias

Lack of results from longitudinal studies

Aim of the study

Discover systematic and stable sources of bias in the General Health Questionnaire administered in a longitudinal study

Design of the study

Participants

• People responding to GHQ-12 across the

four waves of Understanding Society

• Division: Natives, First Generation &

Second Generation of immigrants

Instruments

• GHQ-12: Frequent tool used in the

clinical and research context for assessing non psychotic psychiatrics diseases in communitarian contexts

Design of the study

Data cleaning and participants selection Analyses of contruct, method and item bias across waves Results integration

Construct Measurement Invariance (MCFA) Method Extremity response style (MANOVA) Item Differential Item Functioning

Results: Contruct bias

Results: Method bias

Waves 1-2

Natives First Generation Second Generation

BIAS!!!!

First Generation Immigrants used extremity significantly more than the other groups

Results: Method bias

Waves 3-4

Natives First Generation Second Generation

COMPARABLE LEVEL OF EXTREMITY

Results: Item bias

Natives-First Generation Global Total Items W1 W2 W3 W4 Total 1 X X 2 4 2 X 1 1 3 4 X 1 4 5 1 6 X X X 3 6 7 1 8 2 9 2 10 X 1 3 11 12 2 T

• tal

4 2 1 1 8

Integrated results

• Items asking about the ability

to develop cognitive tasks

• Item 6 includes the only

negation The highest amount of bias (Items 1, 4 & 6)

Item 1. Have you been able to concentrate on what you are doing? Item 4. Did you feel capable of making decisions about things? Item 6. Did you had the feeling you couldn’t overcome your difficulties?

Integrated results

• Items asking self-perceptions

about personal values

The lowest amount of bias (Items 3 & 11)

Item 3. Did you feel you are playing a useful part in things? Item 11. Have you been thinking of yourself as a worthless person?

Conclusions

Bias Time Decreases across time Related to Negations, but not type

• f wording

Areas assessed

Thank you all for your attention

Acknowledgments The research leading to these results has received support under the European Commission’s 7th Framework Programme (FP7/2013-2017) under grant agreement n°312691, InGRID – Inclusive Growth Research Infrastructure Diffusion. Funding This work was supported by a visiting grant from the Integrating expertise in inclusive growth (InGRID). Grants references: c17-13, c17-16 and c17-17.