A contribution on the nature and treatment of missing data in large market surveys

ABSTRACT

Nonresponse (or missing data) is often encountered in large-scale surveys. To enable the behavioural analysis of these data sets, statistical treatments are commonly applied to complete or remove these data. However, the correctness of such procedures critically depends on the nature of the underlying missingness generation process. Clearly, the efficacy of applying either case deletion or imputation procedures rests on the unknown missingness generation mechanism. The contribution of this article is twofold. The study is the first to propose a simple sequential method to attempt to identify the form of missingness. Second, the effectiveness of the tests is assessed by generating (experimentally) nine missing data sets by imposed missing completely at random, missing at random and not missing at random processes, with data removed.

KEYWORDS:

• Treatment of missing data
• survey sampling
• tests for missingness
• consumer preferences
• United States

JEL CLASSIFICATION:

• C18
• C81
• D12
• L96
• O51

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Park and Davis (1993) extended Little’s test (1988b) to the categorical variable case. Chen and Little (1999) generalised Little’s (1988b) idea to propose a Walt-type test. Qu and Song (2002) address some of the limitations of Chen and Little’s (1999) approach, while Potthoff et al. (2006) focus on testing for MAR processes.

² About 183 responses are omitted. About 180 responses have no mobile phone, and three questionnaires are incomplete.

³ All household members (over age 13) participate in the survey.

⁴ The ANOVA results are available from authors upon request.