I apply the following design principles for each survey I design:

1. Positively-worded items: it is better - although not problem free - to use only positively worded questions. Mixing positive and negative or using negative only questions increases threat to the validity and reliability of the data, so I don't mix the two. Also, I don't use double negatives since they are confusing for respondents.

2. Use of a midpoint: depending on the context for the survey, I may or may not use a midpoint. It is generally advisable to avoid the midpoint for respondents that are more familiar with the subject matter. A range of research and scholarly recommendations exist, including using an “N/A” or “don’t know” option in addition or in place of a midpoint, which aligns with my preferred survey design method.

3. Use ascending order: I use ascending order (i.e strongly disagree, disagree, neutral, agree, strongly agree) to reduce selection bias. If the survey design tool allows, I also present the responses vertically to help with mobile compatibility.

4. Ceiling effects: I try to ensure that the survey is sensitive enough to capture the distribution among the highest performers by piloting surveys and including more than 3 responses on a Likert scale.

References

Chyung, S.Y., Barkin, J.R., Shamsy, J.A. (2018). “Evidence-based Survey Design: The Use of Negatively Worded Items in Surveys.” Performance Improvement, Volume 47, Number 3.

Chyung, S.Y., Roberts, K., Swaonson, L., Hankinson, A. (2017). “Evidence-based Survey Design: The Use of a Midpoint on the Likert Scale.” Performance Improvement, Volume 56, Number 10.

Chyung, S.Y., Kennedy, M., Campbell, I. (2018). “Evidence-Based Survey Design: The Use of Ascending or Descending Order of Likert-Type Responses.” Performance Improvement, Volume 57, Number 9.

Chyung, S.Y., Hutchinson, D., Shamsy, J.A. (2018). “Evidence-based Survey Design: Ceiling Effects Associated with Response Scales.” Performance Improvement, Volume 59, Number 6.