Revisiting response errors with a phone survey in India

Date

2020-06-25

Authors

Srivastav, Nikhil

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

This analysis shows that non-sampling biases like consistency and saliency effects are real in the Indian population. Consequently, researchers should carefully consider them while designing a survey. Data from a phone survey of social attitudes in India which uses Qualtrics to randomly assign respondents to different question orders suggests that a significantly higher proportion of respondents support a law prohibiting marriage between high and low caste individuals when the question is preceded by two more personal questions on whether they would oppose or not oppose a family member marrying a low caste person or a person from a different religion. Similarly, within the two personal questions, we observe that a significantly higher proportion report that they would oppose a relative marrying a low caste person when the question appears after they have been asked if they would oppose or not oppose a family member marrying a person of a different religion.

Department

Description

LCSH Subject Headings

Citation