Do Colours Affect Survey Responses?

By Briana Brownell

Does a Yellow Checkbox Give You Better Brand Equity?

The meaning of colours in branding and marketing is a popular topic: blue means you’re trustworthy and yellow means fun, but do yellow checkboxes mean that respondents will give you better scores in brand aspects like approachability or likeability? Maybe…as researchers love to say, more research may be required.

Even though research on the impact of colour in surveys is pretty slim, many survey design guides warn that colour can potentially influence survey responses. Most of the research surrounding the effect of colour in surveys has been in regards to response rates in mail surveys, and unfortunately, there are not many conclusive results about whether colour has a significant effect online. One study found that added elements like pictures don’t seem to cause detrimental results in online surveys. It’s not all good news though. Other experiments showed that question types might affect responses to perceptual questions, and in some of these instances, colour has played a role.

How Can Colours Affect Your Research?

  • Colour can make a survey less clear. A difficult-to-read or difficult-to-complete survey will have lower response rates and potentially misleading results if respondents misunderstand the survey questions or if the answer options are difficult to read.
  • A coloured mail survey could be conspicuous – or look like junk. Both positive and negative effects have been found for coloured mail surveys. If colour makes a survey more noticeable, it can serve as a reminder to complete and increase response rates. However, if a survey is confused for junk mail, response rates can decrease.
  • A coloured scale can affect rating questions. Colour can influence the perception of a scale’s spread and influence results on perception-based rating questions such as agree-disagree scales or numerical rating scales. When the gradient of the colours from one end to the other is amplified, respondents perceive the scale as more severe and give more moderate ratings.

Inventory Questions Are Pretty Safe
Inventory questions such as “Who is your current telecommunications provider?” or “In what year were you born?” do not appear to be affected by the question design or colours used because they have an objectively true answer. As long as the question design and layout are clear and the design doesn’t cause confusion, there is no evidence that the survey’s design affects the responses.

Perceptual Questions May be Affected
Perceptual questions, on the other hand, may be affected by various factors concerning the question style. I know what you’re thinking: we already know that. Very true. Perceptual questions should always be taken with a grain of salt and considered a comparator rather than an absolute measure, whether they’re rainbow coloured or black on white.

Sliders seem to have some interesting effects on survey responses. Both the initial placement of the slider and the size of the slider matter: a wide slider discourages respondents from answering at the extremes, and a slider with an initial placement in the middle discourages a neutral response (respondents prefer to move it rather than leave it where it is). Colouring may also matter in the interpretation of the scale, if the colours used affect the respondent’s perception of the measurement.

Overall, using colour and changing design seem to be okay as long as they are consistent. Think of using the different question styles as using different anchor points and treat them this way in the analysis.

Entering the Era of Grayscale Research Surveys?
Colour and visual elements might be a fun addition to your survey as long as you don’t go overboard: clarity is key to collecting quality data. Remember that researchers see far more surveys than respondents will: make sure it’s not you who is bored with the formatting.

It’s a safer bet to keep the wilder stuff for the inventory questions. Consistency should be a key priority in tracking work (I’m sure I’m the first one to ever recommend that!). Questions that are going to be compared should be in the same format. There, I said it: researchers, here’s your excuse to feed your addiction and give respondents a few pages of item-bank radio button grids.