How can emotions be measured?

To measure emotions depends on how one defines emotion - see question 1. Some researchers simply ask their participants, how they feel. This is interesting, but it does not capture a full picture. Somebody might answer that they feel slightly worried, but their heart is beating rapidly. Someone else might blush in response to a question, but report not feeling anything. As discussed on the "What are emotions" page, emotions are now considered to comprise several components, and so measuring emotions involves assessing very different things. For some a questionnaire suffices, for others complex machines are required.
SELF-REPORT -- SUBJECTIVE EXPERIENCE
Psychologists are well versed in designing questionnaires that tap into aspects of a person, such as certain skills or personality, that are relatively stable. There are ways to statistically test whether such measures are reliable and valid - so that for example they actually measure again and again the same thing, or that what they measure relates to the underlying construct. With emotional states this is much more difficult - some states are very fleeting and you can never test whether the same state would lead to the same self-report, because how do you get the same state? If you watch a movie once, it will never be the same after that. So you might still be scared watching Aliens for the 8th time, but it is not the same scared it was the first time around. How many emotions should you ask for? Happy, angry, sad, fearful, disgusted? What about jealousy? Boredom? Different emotion theories differ in how many different emotions they assume to exist. But the more questions you ask, the more the fleeting emotion is gone, or the memory of what happened earlier is tainted. So why not simply ask a person what they felt - no categories, no scales, just an open box? So let's say the person says, "I was relieved". I bet that in a parallel universe, where the same person would have been given scales, and there would have been a choice of "angry" the person might also have selected angry because of a reflection such as "I was relieved nothing happened, but I was also angry that it came to that situation - how could he do that?" - so asking about emotions may trigger interpretations and reflections about emotions and that messes with the data and their quality. Also, if you leave the choice completely open, strange things might happen. I remember a project with a Master student at Laval University. We had just experienced a massive ice-storm that affected everybody in the province. We had decided not to prompt people with the label of the emotion. Instead, we asked: "How intense was the emotion, when you realized you would have to leave your house for [several days]" - and then some similar questions - finally, the question "Which emotion was it?". Yes, some people said afraid, or happy, or content (?) but what if someone answers "plumbing" or "children"? And yes, these are real answers. Plausibly, this could refer to being afraid of the water pipes bursting due to the freezing temperatures, or the children being affected by having to leave the apartment. In this case, the feeling state seems to relate to something between fear and concern. But to categorize the responses post-hoc, the researchers then have to invoke their own interpretations. Not good.
We are currently preparing a set of web pages at my laboratory at Jacobs University and will discuss there how we are measuring emotions in the research my collaborators and I conduct.
SELF-REPORT -- SUBJECTIVE EXPERIENCE
Psychologists are well versed in designing questionnaires that tap into aspects of a person, such as certain skills or personality, that are relatively stable. There are ways to statistically test whether such measures are reliable and valid - so that for example they actually measure again and again the same thing, or that what they measure relates to the underlying construct. With emotional states this is much more difficult - some states are very fleeting and you can never test whether the same state would lead to the same self-report, because how do you get the same state? If you watch a movie once, it will never be the same after that. So you might still be scared watching Aliens for the 8th time, but it is not the same scared it was the first time around. How many emotions should you ask for? Happy, angry, sad, fearful, disgusted? What about jealousy? Boredom? Different emotion theories differ in how many different emotions they assume to exist. But the more questions you ask, the more the fleeting emotion is gone, or the memory of what happened earlier is tainted. So why not simply ask a person what they felt - no categories, no scales, just an open box? So let's say the person says, "I was relieved". I bet that in a parallel universe, where the same person would have been given scales, and there would have been a choice of "angry" the person might also have selected angry because of a reflection such as "I was relieved nothing happened, but I was also angry that it came to that situation - how could he do that?" - so asking about emotions may trigger interpretations and reflections about emotions and that messes with the data and their quality. Also, if you leave the choice completely open, strange things might happen. I remember a project with a Master student at Laval University. We had just experienced a massive ice-storm that affected everybody in the province. We had decided not to prompt people with the label of the emotion. Instead, we asked: "How intense was the emotion, when you realized you would have to leave your house for [several days]" - and then some similar questions - finally, the question "Which emotion was it?". Yes, some people said afraid, or happy, or content (?) but what if someone answers "plumbing" or "children"? And yes, these are real answers. Plausibly, this could refer to being afraid of the water pipes bursting due to the freezing temperatures, or the children being affected by having to leave the apartment. In this case, the feeling state seems to relate to something between fear and concern. But to categorize the responses post-hoc, the researchers then have to invoke their own interpretations. Not good.
We are currently preparing a set of web pages at my laboratory at Jacobs University and will discuss there how we are measuring emotions in the research my collaborators and I conduct.
Some relevant literature
Mauss, I. B., & Robinson, M. D. (2009). Measures of emotion: A review. Cognition and Emotion, 23, 209-237.
Coan, J. A. & Allen, J. J. B. (Eds.) (2007). The Handbook of Emotion Elicitation and Assessment. New York, NY: Oxford University Press.
Coan, J. A. & Allen, J. J. B. (Eds.) (2007). The Handbook of Emotion Elicitation and Assessment. New York, NY: Oxford University Press.