# To Mid-Point or Not To Mid-Point, That is the Question #MRX

Image via Wikipedia

Here’s the question: Do you use a mid-point or not?
.
The standard five point scale gives you two degrees of positivity, two degrees of negativity, and one degree right in the middle that can be interpreted as neutrality, uncertainty, or whatever the responder feels like using it as. The six point scale provides three degrees of positivity, three degrees of negativity, and no way for anyone to waffle. Is there a right way to set your Likert scale?
.
Sometimes, researchers really want to know which side of the fence you’re on. When you go to vote, you aren’t there to tell both candidates that they’re doing a good job or that they are both equally horrible. You are there to pick one, the winner. However, when given the choice between a Crunchie bar and a Snickers bar, it is certainly possible that you will end up buying one of each. And I would whole heartedly stand behind that decision. Sometimes a mid-point is logical, other times it’s just not so clear. Here are a few things to consider.
.
1) Does the survey refer to something that people truly have to choose between? Is it reasonable for me to buy Dove and Herbal Essences, or will I really only buy one Dell computer.
.
2) How much do you want to annoy survey participants? Put your feet back in the shoes of a normal everyday person and think about how it feels to answer a survey without midpoints. You know you hate it. You know responders hate it. And hate equals decreased data quality. Hate equals lower response rates. Hate equals increased costs of panel recruitment. Are these risks worth that one point?
.
3) Do you know how to use decimal places? Remember, you aren’t using a 5 or 6 point scale. Because you really aren’t concerned about individual responses, but rather averages of hundreds of responses, you’ve got a bazillion decimal places on your side. 1, 2, and 3 may reflect the positive side of a 6 point scale, but so does 1 to 2.4 on a 5 point scale. You can cut your midpoint, top box or bottom box anywhere you like with decimals.
.
What is my choice? In almost every case, I stick with odd numbered scales. To be more specific, 5 point scales.

Related Articles

### 4 responses

1. Hi Annie – I did a post a few weeks ago on mid-points, see http://zoomerang.wordpress.com/2009/04/06/debate-use-a-mid-point-or-not for more on the topic.

As I mentioned there, research suggests a forced response will yield roughly the same proportion of agree/disagree as a scale with a midpoint.

For example:

Results with a midpoint: 30% con, 50% neutral, 20% pro

Results without a midpoint: 60% con, 40% pro

Using (or not using, as the case may be) a mid-point comes down to what your objectives are and how important it is to know whether you’ve got some people who are wavering.

-Donna

2. I use odd number scales for the reasons you mention. However, your post got me to thinking about how I should interpret consumer intent behind the mid-point.

If I don’t know whether a consumer is being uncertain, ambivalent, or truly neutral with the mid-point on the scale, how confident can I be with the insights from their data?

I realize this nuance is lost in the aggregation of the sample, but still, you got me thinking…

Great post BTW

3. Hi Annie – I like the 5 point scale myself. I was having a discussion with a colleague about this and he likes 7 point scales because he believes that some people will never choose the extreme box and therefore you will never get a top box score on a 5 point scale, but you can get top 2 on a 7 point scale = the most positive/negative, while leaving room for those less committed. Your thoughts?

4. Lisa Klein Pearo

I agree 100% although I do think one can easily get away with 7-points without much cost. In addition, if you force people to “choose a side” how reliable is their slight leaning anyway? I’m much more inclined to relyon the validity of a “neutral” than a forced 4 or 5 on an 8-point scale.