How to calculate questionnaire scores when some items require reverse scoring🔗

If you would like to reverse the scoring for some of a participant's questions, we recommend doing so at the stage when subscores are calculated, rather than using a reversed answer scale for individual questions. This is because using a reversed answer scale for some questions can be a source of confusion or can increase the risk of errors when interpreting the data later.

For example, in a set of questions asking participants if they agree or disagree, if a score of "3" indicates that a participant strongly agrees, but you've included some questions with a reverse answer scale (where "3" means they strongly disagreed), then when you are later looking at the responses in your dataset, if you haven't specifically labeled the question as "reversed," then you might misinterpret the reverse-answer questions and think that a "3" meant that the participant strongly agreed with whatever the question was asking. This makes it harder to quickly review your CSV data file, and comes with a higher risk of your data being misinterpreted by other researchers (or even by yourself when you come back to it some time after administering it).

Another example would be if you reverse coded the "I often feel calm" question because it's part of a neuroticism scale. Then, when you see "I often feel calm" in your data set and a participant has a high score, you're likely to assume this means they are a calm person, forgetting that you reverse coded it! Instead, code it the normal way (not reversed coded) and just subtract it rather than adding it to your neuroticism subscale.

Good reverse scoring example🔗

Here's a good example of how to use a reverse-coded question in a summary variable:

>> agreementScale = [["Totally agree", 3], ["Agree", 2], ["Somewhat agree", 1], ["Neither agree nor disagree", 0], ["Somewhat disagree", -1], ["Disagree", -2], ["Totally disagree", -3]]

*maintain: Please indicate the extent to which you agree or disagree with the following question.

*question: I am satisfied with my life.
	*answers: agreementScale
	*save: satisfactionScore

*question: I am dissatisfied with my life.
	*answers: agreementScale
	*save: dissatisfactionScore

>> satisfactionScoreTotal = satisfactionScore - dissatisfactionScore

In the example above, the hypothetical score, satisfactionScoreTotal, is created using the positive value of one scale and the negative value of the other scale. This allows the satisfactionScoreTotal to reflect satisfaction, without changing the interpretability of participants' responses to the question, "I am dissatisfied with my life," (since a seven in response to that question still represents agreement and will therefore still be easy to interpret in your CSV data file).

Bad reverse scoring example🔗

Here's a bad example of how to use reverse-scoring for some questions, making the CSV difficult to read:

>> agreementScale = [["Totally agree", 3], ["Agree", 2], ["Somewhat agree", 1], ["Neither agree nor disagree", 0], ["Somewhat disagree", -1], ["Disagree", -2], ["Totally disagree", -3]]

>> reverseScale = [["Totally agree", -3], ["Agree", -2], ["Somewhat agree", -1], ["Neither agree nor disagree", 0], ["Somewhat disagree", 1], ["Disagree", 2], ["Totally disagree", 3]]

*maintain: Please indicate the extent to which you agree or disagree with the following question.

*question: I am satisfied with my life.
	*answers: agreementScale
	*save: satisfactionScore

*question: I am dissatisfied with my life.
	*answers: reverseScale
	*save: dissatisfactionScore

>> satisfactionScoreTotal = satisfactionScore + dissatisfactionScore

In the example above, the hypothetical score, satisfactionScoreTotal, is created by adding the satisfactionScore and the dissatsifactionScore. Although there's nothing inherently wrong with doing this per se, if the person strongly disagreed with the statement, the data CSV will now contain a "3" as their dissatisfactionScore. If someone were to read that score out of context (e.g., later on, if they review the CSV without the program code in front of them), they may conclude that that respondent had strongly agreed that they were dissatisfied with their life. A slightly better option would have been to save the answer as REVERSEDdissatisfactionScore so that its meaning was more readily apparent; but this is still less intuitive to interpret (and it will likely take longer to process the double negative inherent in saying that someone disagreed that they were dissatisfied with their life) than the results of the good example above.


Next: