This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Serious Games, is properly cited. The complete bibliographic information, a link to the original publication on http://games.jmir.org, as well as this copyright and license information must be included.
Games for health are increasingly used as (part of) health interventions and more effect research into games for health is being done. This online experiment questions expectancies of games for health by investigating whether a game for health prompt might be considered arousal congruent cognitive reappraisal and as such positively effects self-efficacy before gameplay.
The aim of this study experiment is to test whether a game for health prompt effects self-efficacy and other well-being measurements, as a first step into investigating if a game prompt is a form of arousal congruent cognitive reappraisal.
This study used an online, 2D, between-subjects experimental survey design with self-efficacy as the main dependent variable. Stimulus is an assignment for health-related problem solving concerning living with diabetes type II, introduced as a game (n=125) versus the same assignment introduced as a task (n=107). Measurements after prompting the game/task assignment include self-efficacy, positive and negative affect, expected difficulty, flourishing, and self-esteem.
The results indicate a small negative effect from prompting the game assignment on self-efficacy, compared with prompting a task assignment. This effect is mediated by the expected difficulty of the health game/task. No differences between the game and task groups were found in affect, flourishing, or self-esteem.
This experiment provides no support for the notion that a game for health prompt might be seen as arousal congruent cognitive reappraisal.
In “Get excited: Reappraising pre-performance anxiety as excitement” by Brooks [
While reading the work of Brooks, it seemed to me that prompting a game for health is also a call to be excited amidst anxious content and I wondered if a game for health might be considered arousal congruent cognitive reappraisal? If so, this perspective could help to explain some of the attraction to games for health and their effects on self-efficacy.
Excitement is a state of high arousal and positive affect. Anxiety is a state of high arousal and negative affect.
Affect | High arousal | Low arousal |
Positive | Excited | Calm |
Negative | Anxious | Bored |
Research question: Is prompting a game for health a form of arousal congruent cognitive reappraisal?
Health science has been embracing gaming as a meaningful way to communicate, educate, and as a mechanism to deliver treatment [
Some research on the effectiveness of serious gaming for health promotion revealed an overall increase in healthy lifestyle adoption across several health domains [
In order to create a full picture of the effectiveness of games for health, broader intervention characteristics should perhaps be evaluated, such as user experience and perceived relevance [
Cognitive reappraisal is a change in cognition which allows for the interpretation of an emotion-eliciting situation in such a way as to alter the emotional impact it has [
Self-efficacy is a persons’ belief in their capability to perform any task [
Playing games for health has shown to invoke positive feelings [
H1 Prompting a game for health will increase self-efficacy, when compared to prompting a task for health.
H2 Prompting a game for health will increase positive affect and decrease negative affect which will correlate with higher self-efficacy, when compared to prompting a task for health.
H3 Difficulty judgment will positively mediate the strength of the correlation between prompting a game for health and self-efficacy.
Besides affect, flourishing and self-esteem are also incorporated as a measure of well-being that are connected to self-efficacy. The concept of flourishing [
H4 Prompting a game for health will increase flourishing and self-esteem which will correlate with higher self-efficacy, when compared to prompting a task for health.
The health-related context for this experiment is that of “living with diabetes type II” which is selected as it is a prevalent lifestyle disease that is greatly impacted by health behavior, and health professionals have been especially interested in video games as a way to deliver diabetes self-management support [
As you can see in
Hypothesized relationships.
This is an online, 2D, between-subjects experimental design with self-efficacy as the main dependent variable.
The stimulus material contains references that are both health related and math related, in an attempt to increase a sense of difficulty. The opening page of the survey displayed the blue logo of the Diabetes Foundation and 2 blue snakes around a blue staff, a visual that is generally associated with the medical profession. On this page participants were confronted with a text asking them to solve issues related to living with diabetes type II. The text tells them that they will have to solve these unknown problems in either a
Thank you for participating in this research, it will take approximately 5-10 minutes. All your answers will be stored and analysed anonymously.
HEALTH GAME [TASK] On the next pages you will be asked to play a game [task] in which you have to solve several health related problems concerning living with Diabetes type II. This game [task] includes dealing with glycaemic control, caloric intake, measurement intervals and other issues. The game [task] requires no specific prior knowledge and you do not need to have Diabetes to participate.
After reading the text, the survey begins on the next page. Throughout the survey the words “game” or “task” are repeated 12 times. A manipulation check is included after the measurements, asking the participants if they are about to play a game or perform a task (or they don’t know). At the end of the survey a short debriefing explains the purpose of the experiment.
To establish the necessary number of participants, a power calculation was performed with G*Power [
The experiment was hosted online for 2 weeks during the Diabetes Awareness Month. A post on my social media accounts invited people to participate in “research on health-related choices” and encouraged sharing a link to the research with others. Participants were also recruited by specifically targeting Twitter accounts that were game and diabetes related (
Recruiting participants on Twitter.
In a second round participants were hired on the micro-task market: Amazon’s Mechanical Turk. Such Mechanical Turks are best used in cases where there is a small task with a need for many users, when there is a verifiable answer, and there are no objections to a diverse and unknown group of participants [
Participants that passed the manipulation check and had no missing values were included in the analysis. In total, 232 participants were included in analysis (115 men and 117 women; average age 37.5 years, 125 game group and 107 task group). No significant differences were found between the game and task group on the variables gender, age, education level, or experience of diabetes.
The survey consisted of 4 screens. The first screen contained the introduction and the stimulus, the second contained the self-efficacy and expected difficulty measurement. The third screen contained demographic, flourishing, Positive and Negative Affect Schedule (PANAS), self-esteem, experience of diabetes measurement, and the manipulation check. The fourth and last screen contained the debriefing and a comment/question box, followed by my thanks, my name, and academic title.
In this experiment self-efficacy is measured by adapting 2 established measures: the 13-item reduced form Coping Self-Efficacy Scale (CSES) [
“Before we start - We want to ask you to give a confidence rating on the game you are about to do. How confident or certain are you that you can do the following things in the game on Living with Diabetes type II”. This phrasing is in line with the ‘Guide for constructing self-efficacy scales’ [
This experiment uses 2 CSES subscales: Problem-Focused Coping (PFC; 6 items, α=.90) and Stop Unpleasant Emotions and Thoughts (SUET; 4 items; α=.90).
Items of the CSE subscale PFC include the following:
Break an upsetting problem down into smaller parts
Sort out what can be changed, and what cannot be changed
Make a plan of action and follow it when confronted with a problem
Leave options open when things get stressful
Think about one part of the problem at a time
Find solutions to your most difficult problems
Items of the CSE subscale SUET include the following:
Make unpleasant thoughts go away
Take your mind off unpleasant thoughts
Stop yourself from being upset by unpleasant thoughts
Keep from feeling sad
For this experiment, the wording of the items from the SSESS was adapted from the evaluative form into an expectant form. Furthermore, 2 items based purely on self-efficacy of content knowledge were deleted, leaving 4 items (α=.94).
Items of the adapted SSESS include the following:
In the GAME/TASK, I expect to do well
I have no doubts about my capability to do well on this GAME/TASK
I am sure I can do an excellent job in this GAME/TASK
Even when the GAME/TASK questions are difficult, I know I can succeed
All 14 self-efficacy items are combined in 1 matrix (α=.95). Performing a principal component factor analysis on the full matrix of 14 items revealed 1 underlying component. All answers were given on a 0-10 range with 3 semantic anchors (
Self-efficacy answer format.
In order to measure emotional valence, the PANAS is used [
Another measure of well-being is taken with the 8-item Flourishing Scale [
The expected difficulty of the game or task is measured in 1 item by asking the participants how difficult they expect the task/game is going to be and assessing this on a 7-point scale with 2 semantic anchors (1=very easy, 7=very difficult). This measure is an adaptation of the After-Scenario Questionnaire [
Self-esteem is measured by a Single-Item Self-Esteem (SISE) Scale. The SISE was banked against the Rosenberg Self-Esteem Scale [
Participants were asked how familiar they were with diabetes and its challenges, choosing between 4 possible answers:
I have Diabetes
Someone close to me has Diabetes
I have no personal experience of Diabetes but I am aware of what Diabetes is and what the challenges are
I have no or very limited knowledge on this subject
Participants were also asked about their gender, level of education, year of birth, and if their native language was English or otherwise.
The research question investigates the idea that prompting a game to solve health-related problems might be a form of arousal congruent cognitive reappraisal, similar to the “get excited” prompt [
Looking at the participants in either the game or task group in
Further negative correlations with self-efficacy scores are shown with diabetes knowledge (
Correlation table.
Correlation | Positive affect | Negative affect | Self-efficacy | Difficulty judgment | Flourishing | Self-esteem | Diabetes experience | Education | Gender | Age | Language | MTurk |
Game (1)/Task(0) | –.100 | .007 | –.170b,c | –.150a,b | –.125 | –.044 | .087 | –.035 | –.070 | .007 | .020 | –.091 |
Positive affect | — | –.143a,b | .494b,c | .365b,c | .594b,c | .522b,c | –.236b,c | .155a,b | –.068 | –.001 | –.046 | .312b,c |
Negative affect | — | –.346b,c | .074 | –.385b,c | –.155a,b | .003 | .015 | –.052 | –.221b,c | .116 | .004 | |
Self-efficacy | — | .404b,c | .543b,c | .411b,c | –.224b,c | .035 | .007 | .000 | –.133a,b | .289b,c | ||
Difficulty judgment | — | .303b,c | .390b,c | .200b,c | .083 | –.108 | –.195b,c | .003 | .340b,c | |||
Flourishing | — | .569b,c | –.183b,c | .138a,b | .088 | –.020 | –.080 | .142a,b | ||||
Self-esteem | — | –.189b,c | .191b,c | –.099 | –.045 | –.130a,b | .296a,b | |||||
Diabetes knowledge | — | .102 | .055 | –.224b,c | .325b,c | –.325a,b | ||||||
Education | — | –.014 | –.148a,b | .266b,c | –.113 | |||||||
Gender | — | .043 | .038 | –.122 | ||||||||
Age | — | –.114 | –.095 | |||||||||
Language | — | –.391b,c |
a
bSignificant value.
c
To investigate the effect of prompting a game or task for health on self-efficacy (H1), a comparison was made of the average compound self-efficacy score (14 items) of the game versus the task group via independent samples
To investigate the relation between belonging to the game or task group and self-efficacy further, the average scores on these (sub)scales between the game or task group were tested simultaneously by a one-way between-subject ANOVA. For the SSESS significant differences between the game group (mean 8.12 [SD 2.21]) and the task group (mean 8.97 [SD 1.72]) were found, with the Levene test showing that the variances of these scores were not equal (
To investigate if the game group will have a lower negative affect and a higher positive affect when compared with the task group (H2), an independent sample
Investigating the subset of participants that answered Somewhat difficult, Difficult, or Very Difficult (n=79), a difference on the average self-efficacy score (H3) between the game (mean 7.00 [SD 2.38]) and task (mean 8.52 [SD 2.16]) prompted group was found, but this was not significant (
To investigate if belonging to the game group has a positive effect on flourishing (H4), an independent sample
No significant differences were found between participants in the game or task group on age (
In order to test the conceptual model as a whole, a mediation analysis was conducted using the PROCESS macro (model 4: 10,000 bootstrap samples). This model is used for both simple mediation models and parallel multiple mediator models [
Mediation model. Multiple mediator model with flourishing, self-esteem, expected difficulty, and both negative and positive affect as mediators on the effect of a game or task prompt on self-efficacy. Age, gender, language, education, and diabetes experience are included as covariates (paths not shown). *
This analysis shows that the direct effect path between a game and a task prompt on self-efficacy is significant (
Further testing of flourishing and positive and negative affect as moderators between game or task prompting and self-efficacy (including the same covariates as in the mediation model) did not result in any significant paths, and discounted these variables as moderators.
Average scores on several variables between the 2 differently recruited groups were compared and tested via independent
The anonymized data set “The Effect of a Game Prompt on Self-Efficacy Concerning Problem-Solving Challenges of Living with Diabetes type II” can be found online at the Open Science Framework [
There are several significant correlations between the variables measured in this online survey. Being confronted with the game for health stimulus correlates with a little less self-efficacy and with the expected content being judged a little more difficult, compared with being confronted with the task stimulus. The scores of the flourishing scale correlate with the PANAS, self-esteem, and self-efficacy as expected, confirming these measurements of well-being among themselves within this data set.
When the relationship between self-efficacy after a game prompt or a task prompt was tested, scores in each group are significantly different. The game group scores an average of 8.08 on self-efficacy, while the task group scores on average 8.66. This scoring indicates answers between 0=cannot do at all and 11=certain can do. Even though a difference in scoring between groups is found, the difference in scores is small and the average scoring in both groups represents a high amount of self-efficacy.
A significant difference between game and task group participants holds on the State Self-Efficacy Scale and PFC Subscale, but disappears in the scores on the subscale for stopping unwanted emotions or thoughts. That no effect could be found on this subscale might be due to the limited timespan of the survey (5-10 minutes) which is likely not long enough to raise the issue of consciously controlling ones’ emotional and cognitive state.
The expectation that a game for health prompt would be followed by more positive affect and less negative affect was not found. No significant difference in affect was found at all between the game and task groups. Neither could any significant difference in the scores on self-esteem or flourishing be found between the task and game groups.
When looking at the subset of participants that judged the expected content to be difficult, no significant effect of either the game or task stimulus could be found on self-efficacy. However, because many participants did not judge the expected content as difficult, this analysis relies on a smaller number (n=79) which might explain the lack of a robust finding. The results might be indicating a trend that a drop in self-efficacy following a game prompt will get bigger if the content is expected to be more difficult.
Through running a mediation analysis it becomes clear that the difficulty judgment fully mediates the connection between participating in the game prompt or task prompt group and the score of self-efficacy. This mediation indicates that a game for health prompt creates the expectancy of slightly more difficult content compared with the task prompt, which influences the relationship between the type of prompt and self-efficacy in the direction of a game prompt being followed by a little less self-efficacy.
No differences were found between the game and task group in flourishing and self-esteem average scores. Although negative and positive affect as well as flourishing show a significant relation to self-efficacy scores in the mediation model, neither mediation nor moderation can be established and not much can be said on the connection between these variables from these data.
The use of MTurks for online surveys and experiments is getting more widespread [
This experiment has no heart rate measure as an indication of arousal, which is a practical limitation of doing online research. Subjective measures of arousal do exist (such as a self-report scale that might be used online), but research indicates that such measures did not match physiological data collected via electromyography and skin conductance [
Although mediating effects are found from difficulty judgments, participants on average expected this assignment to be “neutral,” meaning neither difficult nor easy. Future research might investigate game or task prompting where the judgment of the expected content is on the “very difficult” side of the scale.
No measurement of game literacy is included in the study; this information might provide interesting correlations with expectations of playing games for health. Future research might look at the level of experience with 3 game categories: entertainment games, serious gaming, and games for health. A further investigation of lay-beliefs and expectancies of these 3 categories before any gameplay seems warranted.
The aim of this study was to try and establish a first foothold into investigating whether prompting a game for health might be considered arousal congruent cognitive reappraisal. As far as this one small study can indicate anything, it appears to indicate that this is not the case. Games or gamification in health care context have been shown to increase self-efficacy [
Prompting health-related content as a game is followed by slightly less self-efficacy (H1), mediated by an increase of the expected difficulty (H3) between “neutral” and “somewhat difficult,” when compared with the assignment as a task. This could be interpreted as the view that games are expected to be more challenging in a negative way. Those who wish to use gaming or gamification for diabetes type II–related interventions, or more broadly in a health-related setting, should be aware of this.
There is no difference in affective state found following a game or a task prompt (H2) and no difference is found between the game and task groups in flourishing and self-esteem (H4). Together, this provides no support for the notion that a game prompt might be seen as cognitive reappraisal.
Coping Self-Efficacy Scale
Positive and Negative Affect Schedule
Problem-Focused Coping
Single-Item Self-Esteem
State Self-Efficacy Subscale
Stop Unpleasant Emotions and Thoughts
This research was started as part of the PhD program at the Faculty of Rehabilitation Science of the Technical University Dortmund, but was concluded independently. I am grateful to Tamara Bouwman, Daniel Oberski, Christian Roth, and my anonymous reviewers for their time, effort, and comments.
None declared.