Not offering don't know options in brand image surveys contaminates data

The aims of this study were to (1) understand the extent to which offering or not

offering a Don’t know option has the potential of contaminating survey data, and

(2) investigate the interaction between offering a Don’t know option and the

verbalisation of scale points.

Results from an experimental study with 196 online panel members confirm that

substantial proportions of empirical data sets can be contaminated if Don’t know

options are not offered to respondents who do not feel competent to assess an

object under study due to lack of familiarity. The maximum extent of data

contamination could not be determined because only one product category was

examined. But the contamination for the less known fast food restaurant under

study amounted to almost 20% of the data. Furthermore results show that using

the typical Likert scale verbalisation of the middle point (“neither agree not

disagree”) is often misinterpreted as a Don’t know option by respondents, thus

increasing the risk of data contamination that cannot be corrected

retrospectively.

Practical recommendations for market researchers are derived from these results.