Relevance & Research Question: Open-ended questions are often used to gather short numeric information in self-administered web questionnaires. Respondents are encouraged to enter numbers, quantities or frequencies into input fields, most often without any computerized formatting constraints predominantly in order to prevent item nonresponse. However, the absence of any formatting restrictions encourages large variances in answers deviating from the desired format, including value ranges, estimations, alphanumeric supplements, or even different measuring units which affect data quality negatively, and increase the efforts for data cleansing and preparation. Thus, concise and clear formatting instructions are needed to guide respondents providing answers in the desired format. Considering the fact that instructions are likely to be ignored the question arises how different modes of verbal instructions and visual cues can be applied to improve the impact of formatting instructions, and finally to enhance data quality.
Methods & Data: In a between-subjects field experiment conducted among university freshman students in an opt-in panel (N=670), we tested different visual modes of formatting instructions in open-ended numeric questions: (1) conventional instruction in a static manner, (2) dynamic instruction in a tooltip appearing when the mouse cursor hovers over the input field, and (3) symbolic instruction in terms of pre-defined default values in the input field indicating the desired response format. The effectiveness of each instruction mode was determined by the proportion of formally correct answers.
Results: Findings indicated that the implementation of dynamic formatting elements in terms of tooltips or default values had no positive effect on an improvement of response quality compared to conventional static formatting instructions. Even a combination of tooltips and pre-filled symbols could not achieve a significant increase in correctly formatted answers compared to the sole presentation of a fixed instruction.
Added Value: The results indicated that static formatting instructions should not be replaced hastily without examining the effect of dynamic elements sufficiently. However, initial findings suggested the potential of dynamic formatting instructions in enhancing the positive effect of conventional instructions.