5 tips to optimize your facial expression analyses
Are you involved in emotion recognition and facial expression analysis? These 5 tips will guarantee the best results!
1. The whole face should be visible
You may normally offer your test participants something to eat or drink during the test to make them feel comfortable. However, when eating or drinking, the test participant blocks part of the face with his/her hand and the cup/glass/spoon. So, make sure the test participant finishes his/her cup of coffee before you start the test… unless, of course, the purpose of the test is to see a reaction to a new food or beverage!
There are still steps you can take to minimize interference: if testing a soda, for example, test participants could drink through a straw, which would ideally be transparent and colorless.
A tip to consider: Some people have a natural tendency to touch their face without realizing it (like the lady in the picture). You may want to warn the test participants to keep their arms on the table throughout the test in order to get the best possible results.
2. Make sure the test participant looks into the camera
Rotation of the face is OK with automatic facial analysis, but it should be limited (<40°). That means that you should sit opposite of the test participant and not next to him/her if you want to ask questions during the test. In a social interaction setting (medical student – patient or consultant - client interaction) you would ideally position the camera on the conversation partner’s shoulder. With this set-up people almost look straight into the camera when they talk to their partner. In practice this may not be possible, but if you put the camera a short distance away, at the same height, with direct sight line to the participant’s face, you have a working solution.
A tip to consider: In a taste-test study, you can use a cup or bowl holder so the dish is closer to the face and the test participant does not need to tilt the head downward as much to taste the drink/food.
A tip to consider: If asking clients to read instructions or answer a questionnaire, you can present this information on a screen with the relevant text and attach the camera to the screen, or use an on-board webcam.
3. Test people in their natural environment
When you include emotional expression analysis in your research you of course hope to see some displayed emotions. “All my test participants looked neutral throughout the test” is an often-heard complaint. It may help to test people in their home or other familiar environment (restaurant, school, office, etc.). Running your tests in the test facilities in your office has the obvious advantage that you can control the test conditions much better than out in the field.
The downside, however, is that people may react differently to your stimuli. Particularly when interested in emotions, this is a major factor to take into account. Noldus has created a new software, FaceReader Online, to allow testing to occur at home. With FaceReader Online, participants can view an advertisement, or series of images/ads, at home, with all results sent directly to you upon completion.
A tip to consider: Testing people via an online service also allows collection of data from people in remote locations, which can help to gather more diverse demographic data.
4. Real products are best
What approach is best to use when testing your product? You can show test participants pictures of your product, videos/commercials featuring your product, or you can give them the real product to view. What we have seen is that real products elicit greater responses. While having the real product at hand to view it is preferable to seeing a photo, real-life use is even more preferred. So, if you wanted to know whether your test participants like your new healthy flour mix, for example, you can ask them to bake pancakes in a test kitchen, and observe the displayed reactions during the baking process and the consumption phase.
A tip to consider: Making the environment as natural as possible when doing this type of testing will elicit the most “real” emotional response!
5. Present more than one stimulus
If you want to test your new product (commercial, beverage, etc.), it is best to embed it in a series of other products. In this way the test participants do not concentrate fully on the focus product and will respond in a more natural way. This also allows the participant to evaluate your product compared to similar, and to allow the emotion response to be evaluated compared to other products.
A tip to consider: Be sure to randomize the order in which you present the products.