From Fieldwork Came Coding 1 Emotion Question In A Questionnaire

After some fieldwork and telephone surveys for a research agency that conducts emotion research at companies and organizations, I received a long-term assignment to code 1 question from a customer satisfaction questionnaire for emotions once a week. This survey question required participants to select a photo from a range that they felt best described their customer service experience. After choosing the photo, they could explain the photo. This method is also used to assess employee satisfaction among other things.

Explanation Of Categorizing A Photo

I coded the explanation of the photo into the 8 emotion categories of Plutchik's emotion wheel. In addition, I was not allowed to look at the answers to the other questions in the questionnaire, as this could influence my choice for an emotion category. It was purely about the customer's explanation of the choice for the photo. The photo reflected their customer service experience. I could choose from the following emotions:

• Joy

• Trust

• Fear

• Surprise

• Sadness

• Disgust

• Anger

• Anticipation.

I have done this coding work on the photo question every week since 2017 up to the fall of 2022 for Dutch, English and French answers.

From EMDR Taking Emotion Out Of The Event To Emotion Wheel From Plutchik Looking For The Emotion In The Answer

When coding the photo explanation, as I said before, I had to classify the answer into the 8 emotion categories. I had to give the question a charge. This was the reverse of the EMDR therapy I've had where the emotion was taken out of the event by retelling the event over and over again. Sometimes the emotions dripped from the explanation of the question and sometimes it didn't. It was precisely when the emotion was immediately apparent from the answer that I had to pay close attention to the signal words in order to choose the right emotion category.

Now A Robot Is Doing My Work

In the beginning I talked to the client about the division into the emotion categories and the words to watch out for. It often involved:

• The length of answers

• The verbs used

• Etc.

I learnt a lot about emotions by coding the answers into emotion categories and grading within the emotion category into 3 gradients:

• Low

• Normal

• High.

Over time, I saw how the Joy category became more and more common. I can therefore conclude that such a questionnaire works to improve customer service. Sometimes with compliments or High Anger I indicated that the answer had to be looked at. This was necessary to improve service and prevent a recurrence of an incident.

AI (robot)

I coded within a system that gradually expanded and became more and more automated. I even had to check occasionally whether the AI chose the right categories and gradient. By correcting, the robot (AI) learnt just like me and other people. The only difference between me and an AI is that a robot works more systematically.

And Then I Handed Over My Work To A Robot (AI)

Every week I coded French and Dutch answers and the AI learned from that. When applying this questionnaire to employees, it was already the AI that coded and a human (me or someone else) who checked it. In autumn it was time for the AI to take over my tasks. Benefits of using AI:

• Technology has taken another step forward

• Saving money: No more human hours

• Saving time: As soon as the questionnaire arrives it is automatically coded. Also outside office hours

• The margin of error of AI is smaller than with humans because the robot is coded according to the system of the categories I mentioned earlier

• The bias will also be less because a person remains a person with emotions who assesses an emotion question from their own view of life. 2 people can differ from each other, and 2 robots will not differ from each other.

Periodic checking remains necessary because it concerns an open question, and the answer does not always contain the signal words on which to score.