By: Andrew Grenville – Chief Research Officer
It drives me bonkers when researchers, trying to understand people’s attitudes, use a scale that allows people to say they “neither agree nor disagree”. This renders the statement nigh upon useless. The problem is a variation on the conundrum of comparing apples and oranges.
When we ask someone how much they agree or disagree with a statement, we are asking them where, on a continuum, they stand. The continuum can be sliced anyway; a four point scale or one hundred point scale—it hardly matters—so long as we are looking at where on the spectrum respondent lives. When we throw in a “neither agree nor disagree” option, we step off the continuum and tumble into a completely different kettle of fish. And we can’t compare apples and fish.
The problem that results is serious. If we want to use the attitudes to create a segmentation, or for modelling drivers of choice, or just about anything that isn’t purely descriptive, we are assuming linearity—that everyone’s answers are somewhere along a continuum. But when you have an answer that is off the continuum, you’ve gutted that assumption.
If you were making apple sauce, would you throw in a fish? I doubt it. It shouldn’t happen with agree/disagree statements either.
“Neither agree nor disagree” is also a vacuously vague response. What does it tell you? Nothing, except that the respondent chose not to stay on the continuum.
There are statements that some people might not be able to agree or disagree with. “I have stopped setting fire to myself” is, for example, the kind of statement that not everyone can agree or disagree with.
If you think that some of the statements you are using will be hard for some people to agree or disagree with, then you are much better off adding options like “don’t know/unsure” or “rather not answer”. Then, at least you know what you are measuring.
What do you think? Pick a side, any side. But just don’t say you neither agree nor disagree.
Makes one wonder where the poll takers head was at when they made such a statement. Either you agree or you do not agree. Of course this also means that you are taking a stand when you agree or do not agree. It sounds like some kind of a wishy washy liberal plot.
If the question is either/or, of course you may neither agree or disagree because you may not know enough about the subject or you simply may not care. If you ask, “Uranus is larger than Neptune, agree or disagree?”, you may get a lot of “Neither”.
If the question is as a continuum, The neither/ nor would be right in the middle, indicating the responder is again in a don’t care, don’t know condition, which may also be statistically interesting.
As a researcher it can be advantageous in many situations to know if someone chooses “not to stay on the continuum” and even to intentionally give them that choice. If I’m trying to measure people’s feelings along a line, but some people choose to effectively say “wait a minute, I don’t even agree with the premise here” getting those people OUT of the measurement can be to my benefit and improve accuracy of the remaining that do buy into the survey.
Lots of questions… maybe even most questions have built in assumptions. The occasional intelligent survey recipient may buy into those assumptions and mark their spot on the gradient scale, or choose not to buy into those assumptions. The survey taker can benefit from allowing those to “opt-out” of the “perceived as biased” question and effectively say… “No, there is a 3rd option here that you’re not giving me and I neither agree nor disagree!”
Let’s say I ask the question “Are Cats better companions than dogs?” Some may feel that both animals are terrible companions and want to opt out. As a survey writer, in many situations we may want this type of person to do so and not force them to make a choice between animals, neither of which they appreciate and throw their weight towards.
You do make some good points though, and after thinking about it I could go either way on this. In other words… I neither agree nor disagree. -Mark
I have on occasion been enticed to participate in questionnaires. One of my main reason for *not* participating in them more often is that my assesment of the situation being measured does not fit within the constraints of the question asked of me. In such a case, “neither agree or disagree” is my best friend.
The comments here are all valid. However, I agree with Andrew that if the respondent has no basis for opinion then give them an option to say “unsure”.
Greg