Documentation in Google Docs:
Substaining Gender Balance … Presented at https://idc2018girls.wordpress.com/
This message gives feedback about Tappetina’s Empathy Game session during April 2018. For general information about Tappetina, the books, and past workshops, please refer to http://www.tappetina.org. In general the goal of Tappetina is to build on research in art and technology to motivate people to cooperate toward positive goals, like education, gender equality, health for all.
The goal of the Tappetina Empathy Game is twofold: we want to establish a Playground for positive Storytelling and we want to increase Understanding of emotions. The game is a prototype developed by Master student Sindre B. Skaraas. The research is developed and carried out by Letizia Jaccheri, Javier Gomez, and Kshitij Sharma.
For the April trial, in total 58 persons have played the game. The age range is from 20 to 64. 35 male and 23 female participated. 26 persons are from Norway, the remaining from 23 different countries in Asia, Africa, Europa, America. A total of 17 stories has been produced.
We use the Toronto Empathy Questionnaire to build our inquiry process about empathy and storytelling. We have run a usability study by running a SUS Questionnaire to seek the help of players to improve the game. Your suggestions indicate that the usability of the system is around the average, but we have a lot of improving possibilities, such as provide a better view of the whole story by removing the scrolling, include icons/emojis, change the typography/color scheme and solve network issues to reduce the lag and timeout errors.
We also looked into the data we collected from the players of Tappetina empathy and there were some interesting results:
- There is no difference in the empathy quotient of men and women.
- There is a small but significant difference of empathy quotient of people from different age groups. Older participants showed lower empathy quotients than the younger participants.
- The game did not reduce the empathy quotient of the players (there is a high positive and significant correlation of 0.70 between pre- and post- game empathy quotient), however, it also did not improve it. One reason for this could be the limited interaction time with the game and the fellow players.
Finally, we used machine learning algorithms (Support vector machines, Random Forest, k-nearest neighbour, and Artificial neural network) on the audio data collected during the game to predict each player’s empathy quotient. The results, from the best machine learning model, show that using the audio data we could predict the empathy quotient with an accuracy of close to 10%. This means that on a scale of -37 to 37, we would have an error of around ±6 points.