Is there a difference between how people respond to a 10 (1-10) or an 11 (0-11) point scale? Are differences seen just at the bottom end of the scale, or does it impact the whole scale? This question on the e-group sparked an interesting debate which is summarised below:
A – 11pt scale (0-10) provides a true ‘out-of-ten’ score, whereas 1-10 is an ‘out-of 9+1’ score i.e. the lowest average score possible is 1.0 (or in percentage terms 10% i.e. under 10% isn’t possible).
In reality, respondents probably won’t answer much differently, depending on topic, although 0-10 allows for a middle (neutral) value of 5 to be selected whereas the mid-pt of 1-10 is 5.5, not able to be selected by respondents – so they’re forced to be either slightly lower or slightly higher than absolute neutrality, if they answer with a 5 or 6 (in reality the average is likely to be between 6.0-8.0 if it’s something like importance ratings).
It’s important to keep in mind the scale when interpreting results e.g. on a 1-10 scale an average of 5.4 is nearer to the bottom than the top, whereas it’s slightly positive on a 0-10 scale. I prefer using 0-10 as it allows for the mid-pt and it’s also easier to understand in people’s heads (as we’re all brought up on percentages and 5.4 equates to 54% of 0-10).
A – I’m not aware of any systematic difference across the scale points. My own experience is that it mainly affects the bottom end of the scale. Those who use 10-point or 11-point scales tend to use 1-10 or 0-10 so it really only matters whether your lowest point on your scale best lends itself to a 1 or a 0.
As, most commonly, these types of scales were developed for probability/intention/prediction, the best version I’m aware of – with both numeric and semantic labelling – is the Juster Scale:
This scale actually turns it into a non-linear scale with ‘0’ and ‘10’ a long way apart from ‘1’ and ‘9’. Some would argue that the 0 score should be labelled “no chance” not the “no chance or almost no chance” that Foresight uses.
A – The main point of difference is if you want a mid-point that is “neutral” which means you need an odd number. Here is a link from waypointgroup.org titled “why a 0-10 scale is your best option.
11 point scale is better for calculating and reporting mean scores. An 11 point scale (0 to 10) produces a mean score between 0 and 10. With a 10 point (1 to 10) scale you can never get a mean score below 1, so the calculated mean score would be skewed if everyone scores at the bottom end of the scale.
A – I would argue that they’re both too much. I don’t recall seeing too much real differentiation between the points on the scale, especially when I’m actually present at the research and can see the scales being used. My theory is that people will consider 5 to be the mid-point regardless. I’d not really go over 7-point and even then, I think it only really works “properly” when it’s carefully laid out for the respondent.
A – I think it comes down to how the scale is presented. If it’s laid out clearly, so that respondents can visualise where the mid-point is (or isn’t in the case of a 10-pt scale), then it will be used more meaningfully than if it’s just a number being asked for.
A – The most notable example of an 11 point scale is Net Promoter Score (NPS) which runs from 0-10. My own view would be that the ‘0’ would have to mean something for it to be at the bottom of the scale, i.e. that whatever you are assessing can have a zero value, e.g. ‘How much confidence do you have the the earth is flat?’. Otherwise 1-10 makes more sense (to me).
A – There are massive cultural differences — well, National ones anyway. In my experience/recollection, Americans averaged about 2 whole points higher than French on a 10-pt scale — all else remaining equal.