EMOJIS AS AN (IMPERFECT) PROXY FOR EMOTIONAL RESPONSE IN SURVEYS 

By: Annie Low LinkedIn, Senior Research Manager and Chris Oatey LinkedIn, Research Director, 2CV 

It is increasingly understood that emotions play a big role in our decision-making. Behavioural science shows us that how people feel about something has more influence on their choices than what they say they think about it. The happier they feel about a product, campaign, or brand, the more likely it is that they will go back to it.

This can be at odds with traditional research approaches, which typically prioritise a read on a respondent’s ‘rational’ choice. Survey questions tend to focus on stated preference and direct evaluations to gauge how people will react. 

It’s no surprise, then, that the research industry has looked to go beyond the purely rational and capture emotional responses, too. At 2CV, we started looking at ways to measure emotion quantitatively. 

One area of exploration was emojis – specifically Google’s free-to-use Noto emojis. 2CV ran a multinational study to see if these could be used as a proxy for emotional response.

Why emojis?

We know that emotions can be hard to articulate in a survey setting. Respondents may not be fully aware of them, or they may be difficult to express with words. 

Enter emojis: they’re non-verbal, simple, and globally familiar. Research has shown that ‘smiley’-type emojis are used in similar ways across Eastern and Western countries (Guntuku et al, 2019). These images, or non-verbal responses, can allow people to express an emotion spontaneously without having to rationalise it first, as they might do in a traditional open-ended question.

In addition, emojis can be used to represent the seven universal emotions we all display: anger, disgust, fear, happiness, sadness, contempt, and surprise (Ekman, 1972). Neutrality is included too for occasions when there is no strong emotional response. Fortunately for us, there is no shortage of emoji expressions to cover these basic human emotions.

Matching the emoji to the emotion 

The challenge for us was clear: which of the many available emojis are most universally understood to indicate the Ekman emotions? After all, each emoji carries a specific emotional nuance and its usage may depend on cultural context.

We set out to validate our emoji choices through an 11-market study with 900 respondents, testing which versions of the emojis should be used in future surveys.

Each respondent was shown both version A and version B (Image 1). They were asked to select which emoji (if any) best represented a given emotion – for example, “Which emoji best represents happiness?”


Image 1: Emoji options shown to respondents for each Ekman emotion

Most emojis were obvious, but a few needed clarifying 

The results were mostly clear-cut and relatively consistent across markets. There was strong consensus on the choices for anger, surprise, neutrality, and disgust, while the two happiness options were roughly equally acceptable, with only country-specific preferences for either option. 

However, the remaining emotions proved to be slightly more contentious. For example, fear version B  was sometimes read as surprise, so it would not be a good choice to use. The disgust options  were also sometimes read as contempt in a few countries, though this ambiguity cleared up once respondents were presented with the full set of emojis and emotions.

Interestingly, contempt was the hardest emotion to capture. It’s a more complicated emotion (one that Ekman also grappled with), and people occasionally attributed both contempt emojis  to other emotions, such as anger or disgust. Other research has also found that expressions of contempt can differ across cultures, which may explain why it was the most ambiguous emoji overall.

Best practices moving forward 

With these findings in hand, we arrived at our final eight validated emojis (Image 2) that we have since started using in various projects at 2CV.


Image 2: Final validated emojis chosen for each Ekman emotion

We concluded that emojis can work as a proxy to gauge emotional response, but they remain an imperfect tool on their own. They work better alongside a few practices put into place to avoid any confusion when using them in a survey.

One such practice is to label the emoji clearly (taking special care with translations), as the image alone may communicate something else. All eight emojis should also be shown together, as respondents can better differentiate between emotions if they see all discrete options at once.

We also found it helpful to ask this question early in the survey and as close to stimulus exposure as possible, before respondents were fatigued or affected by extraneous emotions. Measuring the intensity of their emotional response (e.g. on a rated scale) also adds another layer of nuance, and this is ideally paired with a follow-up open-ended question to understand why they are experiencing that emotion.

We also acknowledge that some clients might feel that disgust version B is too graphic (though it was the more universally understood option), in which case version A could be used instead. 

There will always be slight differences between markets, but these emojis represent a good starting point to get a quick read on your audience’s emotional state. Understanding this is not only useful as a potential indicator of future behaviour, but it also adds colour and context to the data, with the bonus of enhancing the respondent experience. 

This article was first published in the Q3 2022 edition of Asia Research Media

Leave a Comment

Your email address will not be published. Required fields are marked *