Relevance & Research Question: Our paper focusses on how to evaluate and increase data quality in online surveys. We will briefly recapp our past experiences (fraudsters, questionnaire design & incentives). This paper will include new empiric research on how the mood of respondents and how the incentives influence data quality.

Part A: Respondent Mood: We will a series of experiments in which we can evaluate the influence of the respondent mood on data quality as well as vice-versa the influence of the survey on the respondent mood.

Part B: Conditional & Variable Incentives: We are currently already experimenting with variable incentives - either chosen by the respondent themselves on a pre-defined scale or higher incentives paid if exceptionally good quality interviews are submitted (with a feedback loops).

Methods and Data: 

Part A: Empiric research with n= 1000 nat rep interviews conducted online and required booster samples for low (bad) and high (good) mood.

Part B: Empiric survey with a minimum of n= 1000 nat rep interviews and any required booster samples.

Results: Our results show how the mood of respondent influences their behaviour in surveys and what effect this has on data quality. We also evaluate how the incentive can be used to motivate respondents to supply an increased reliability and data consistency. We will also be reviewing the effects of variable / conditional incentives and longitudinal learning effects by respondents.

Added Value: Contrary to stringent data cleaning and dedupe technology it has alsways been our firm belief that good respondent behavior can be nurtured. Our series of surveys proves clearly how good behaviour can be promoted and offers practical guidelines on how to evaluate and increase data quality with cost effective and pragmatic recommendations.