How to calculate a percentage with ambiguous numbers?

What do you think constitutes “enjoying” the class? Just 5s? 4s and 5s? 3s, 4s, and 5s? I would not interpret N/A as enjoying the class, but since they still took the survey they should be counted in your whole.

Once you set your criterion, count the number of students who selected those values (just 5s, 4s and 5s, etc) and divide that by the total number of students who took the survey.

You put the n/a in there for a reason, I would assume. If the n/a means “not applicable” then it really is a zero, I would think. If the student answer n/a then it means the question didn’t apply to them so they are not part of the equation and you can eliminate them. If it means “no answer” then I think can also be then count them as a zero, or don’t give it as an option on the next survey or replace it with a scale of 0–5. Not a mathematical answer, just my answer.

I don’t think the option of choosing N/A for this question should even exist…either you enjoyed the class or you didn’t. Unless of course the student never attended…lol!

I edited the above…

I would use only the number of people who gave a numeric answer in figuring the percentage and just note how many people chose the N/A option for each particular question. Use the percentages from each individual question to calculate the total percentage just noting again the number of N/A responses.

I would think the percentage on each individual question is more important than the total percentage as each individual question presumably deals with different aspects of the class. The individual information might tell you where you need to improve and where you are doing well. However, as someone who has given and taken such evaluations, the way a person scores you is also important. If someone gives all 4’s and only one 5, I would think that person was really happy with that particular aspect of your class. I usually do not give really low or really high scores unless someone was particularly bad or exceptionally good.

Maybe I am not understanding, but wouldn’t you just change your denominator? Say 20 kids all circled numbers – you calculate the average score by adding all 20 of the numbers then dividing by 20. But if one kid circles N/A, you add the 19 numbers and divide by 19.

You have to account for them separately because they are literally off the scale. per excellent analysis from @optimisticpessimist. But I’d say don’t call anything percentages unless it’s survey population.
Give for example: 60% of the participants chose 5 on a scale of 1 to 5, 3% selected N/A.

Ref “then the average evaluation is 90%” whoa ! ... ninety percent what? There is already the completely arbitrary but useful set of ( N/A, 1, 2, 3, 4, 5 ) values into which the evaluation participants are asked to project their experience. Unless your questions explain and define to the participant that percentage mapping from the numbers that is a possibly misleading abstraction of the data you got from them. And unless you explicitly informed them also about N/A value equal zero that is not a safe assumption IMHO.
Unless your questions explain and define to the participant that percentage mapping from the numbers that is a possibly misleading abstraction of the data you got from them. And unless you explicitly informed them also about N/A value equal zero that is not a safe assumption IMHO.
If you predefined a linear scale with N/A at the bottom then….um,er, never mind !

@dabbler I totally agree with you second paragraph (until the last sentence where you lost me.) It explains what I meant when I said, “If someone gives all 4’s and only one 5, I would think that person was really happy with that particular aspect of your class. ” When I used to look at these type of evaluations, I would scan each person’s answers and if they gave a 3 for almost every question I knew a deviation from the normal score is what I needed to pay attention to.

You can say, “40% of the students reporting their satisfaction said they enjoyed the class.”

You can also represent the data as a pie chart and either include “n/a” as one of the slices or add a footnote saying “x out of y students declined to provide a rating.”

I agree with @Mariah.

I also agree with @6rant6 answer.

But, there is an argument for counting all 20. Seems you could just give all the stats. Blank % answered 1, blank % answered 2, then 3, then 4, 5, and NA.

I think you just shouldn’t calculate the n/a answers into the overall percentage. For example, if you were calculating averages you would just disregard it if you didn’t have data. I would consider a n/a answer a lack of data.

Yeah, ignore the n/a answers.

Unless you have a good reason to believe n/a means something else (e.g. they hated the class but are too nice to say so).

I would compute the score in two ways. One way is to ignore the N/As and use @Mariah‘s method. The other way is to treat an N/A as being indifferent, which would be half way between 1 and 5, or a 3.