When data are missing because a participant was absent or chose not to answer, we take our best guess concerning what they would have responded based on their other responses. This is called imputation. Typically, a previous response to the same question is simply carried forward. If no previous response exists (e.g., if participants took the second survey but not the first), then their response is carried backwards. For example, if they chose “Agree” on an item the first time they took the survey, we assume they would “Agree” again the next time. This type of imputation stabilizes the data. It ensures that any changes you see on reports are the result of real changes in participants’ responses, not merely an artifact of who happened to respond to a given survey.
