When designing a survey, the term “scale” refers to our measuring stick — the answer choices we provide in order to quantify our respondents’ attitudes and behaviors. In surveys you commonly see this expressed as a, “five-point scale.” You might, for example, ask people how important it is that their automobile have a steering wheel. The scale in this case might be: Not important, Slightly important, Neutral, Fairly important and Very important. Those five levels of importance are what define that particular five-point scale, and while the example is silly, you get the idea. In surveys, two of the most commonly used five-point scales are importance and likelihood.
Importance Scales
Wondering what features are most important when looking at refrigerators or choosing a vacation package? The importance scale is used frequently because it’s a simple way of capturing your target market’s perceived preferences. It’s also a little tricky, because our respondents may over- or under-state the importance of a given factor, not out of dishonesty, but because there are subconscious influences at work.Think about it: how do you select a brand of orange juice from the crowded grocery shelf? Because you have a coupon? Because of the container’s deisgn? Because you’ve bought the brand before? All could be important, but not all will be conscious. Consumers may not think about it, but there’s a reason orange juice isn’t generally packaged in brown containers, and milk cartons are usually mostly white. So while they may not report that “container design” is important, it really is.With importance scales we ask about what’s important to our respondents, but we have to keep in mind that it’s hard for people to accurately self-report everything that may be impacting their behavior. Be aware of items that may be harder to self-report, and either leave them out or be careful about how you report their results.
Likelihood Scales
Likelihood scales are a useful and common way to gauge likely future behavior. How likely are you to purchase a new car in the next six months? How likely are you to refer our brand to your friends or family? This is very useful information, but we have to be careful what we ask about, and we have to be very careful about time-frame. Asking a consumer about their plans in the next month is much more likely to be accurate than asking them to predict a year in advance –many of us don’t yet know what we’re having for lunch today!A sure sign of a survey design newbie is asking about likelihood without specifying a time frame. I recently saw this question, “How likely are you to visit our store again?” Ummm…ever? This year? In the next 6 months? Yikes, what utterly useless data.Likelihood scales are very useful and practical in market research if we keep the timeframes specific and preferably short — generally less than three months, and certainly no more than six months for most product categories.
Are You Odd? How about Your Scales?
These scales, also known as Likert scales, can have either an even or odd number of possible responses. While 5 is the most commonly used, you will sometimes see 4 point scales, 7 point, and in some cases 10 or 11 point scales.Odd scales give you a “neutral” answer option for respondents to select. The advantage of this is that neutral may be a realistic attitude. For example, if you ask me about the importance of “trunk size” when purchasing my next car, my answer could honestly be neutral—it is neither important or Unimportant. If you give me a 4 point scale, I would be forced to express an opinion that may not be valid. The risk with odd scales is that a respondent may take the easy middle ground, rather than giving the question a little more thought by taking a stand “for” or “against”. In practice, many researchers choose the odd scale for most projects.
Conclusion
Both Importance and Likelihood scales have value, but only if their limitations are respected. People can’t always define what’s important to them, nor can they always predict their own future behavior reliably. That doesn’t mean that collecting this data isn’t valuable — there are product categories and instances where people can be accurate — but be aware that there can be inconsistencies between what people say and what they do.