Regularisation and reliability assessment of data in survey analysis

Surveys are considered a powerful tool in business strategy planning and decision making and in many cases provide essential data for managers in different circumstances. However, while it is evident that each respondent presents a difference in evaluating and rating questions, their psychological differences are rarely assessed and data is generally considered of the same value and reliability. This paper proposes the use of predefined questions to evaluate the differences of value in the answers provided by the respondents. The information is then employed to regularise the data accordingly and reduce the distance of the provided answers to that of the respondent's actual thoughts. The respondents' understanding and effort towards the survey are also evaluated in the form of assessing the reliability of the provided answers, which offer beneficial information for the analyst. The proposed method is implemented and demonstrated through a real case study. An increase in data quality and reliability was achieved, as well as aiding information in future analysis processes.