CandyCrush engagement analysis based on posture

Research Overview

People tend to give away there feeling unintentionally with their posture, also known as body language.

This research is trying to find a correlation between the player's posture and the player's engagement level with a quantity approach. By recording the distance between players eye and eye tracker under the screen, we were able to draw some interesting conclusion.

Screenshot from gameplay video.

Data Collection Method

Participants are asked to play CandyCrush on PC for 8 min.

During the play session, the distance between players face and screen are recorded by an eye tracker attached on the bottom of the screen.

Eyetracking device

Participant is also required to speak out what is in his mind while playing the game. The recording will provide insight into the person's experience during the play session.

Data Analysis

Gesture Analysis

headmovement<- read.csv("Headmovement.csv")
headmovement<- headmovement[complete.cases(headmovement),]
library(dygraphs)

## Warning: package 'dygraphs' was built under R version 3.4.4

dygraph(headmovement,main = "Eye distance")

## Result

Trend Analysis

Player tend to lean forward gradually when they are engaging the game.

Spike Analysis

Head wobble away from the screen

During 70~75s, 104 and so on, the participant is highly engaged in the game and did something very satisfying.

Head wobble towards screen

During 222~243s,292,348 and so on, the participant is trying to solve a puzzle and focusing on the game.

Huge spikes

Huge spike represents a huge emotion change on participants part. Huge spike 196,447, the participant is very happy. During huge spike 313, the participant is very frustrated by the microtransaction prompt.

Change pose

During 260s, the participant changed her pose and moved away from the screen because he is confused about the game.

Conclusion

The posture provides a fascinating insight into what is going on in participants minds. I’m surprised how consistent head wobble pattern is. The posture tracking can work well when players are not speaking or have on time to reflect on their try emotion.

However, more testing is needed in order to see if the gesture pattern is consistent among all participant.