Content uploaded by Veit Etzold
Author content
All content in this area was uploaded by Veit Etzold on Oct 16, 2020
Content may be subject to copyright.
Effects on Girls‘ Emotion During Gamification-Tasks
with Male Priming in STEM Subjects via Eye tracking
Tabea Wanner1, Tamara Wanner² and Veit Etzold1
1 Competence Center of Neuromarketing, University of Aalen, Germany
tabea.wanner@hs-aalen.de, veit.etzold@hs-aalen.de
² Course of Study Internet of Things, University of Aalen, Germany
tamara.wanner@hs-aalen.de
Abstract. In the war of talents, there is great potential in female workers. To gain
their interest in STEM-Subjects it is important to reach them during their school
time. One aim of our study was to find out which emotions teen girls (age 13-15)
have while doing (block) programming Gamifications-tasks and observe them
while doing binary code calculation and building a Calliope-mini-piano (no
Gamification-tasks). An eye tracker with an emotion-software (based on Facial
Action Coding System) measure their emotions during the Gamification-tasks.
The girls were first divided into different groups, then primed in different gender
stereotypes (Pro-STEM, Anti-STEM and no priming). It was noticeable that the
girls who were primed with Anti-STEM achieved better results in the program-
ming game. This could be due to the age of the girls (puberty). It could be ob-
served that the girls enjoyed the Gamification-tasks. The other tasks were hardly
noticed. Gamification is one key to reach them.
Keywords: Emotions, Gamification, Eye tracker, STEM, Girls in STEM
1 Introduction and Literature Review
In Germany, around 7.7 million STEM specialists (STEM= Science, Technology, En-
gineering, Mathematics) were employed subject to social insurance contributions in
2017. However, the number of STEM shortage jobs is rising. The shortage of skilled
workers with vocational training could increase and widen in the future [1]. Where else
can skilled workers be generated in Germany? Although the proportion of women in
STEM occupations is slowly rising, it is still well below an average at 15% in Germany.
The proportion of women in the next generation of academics is also only 28% [1].
This could be counteracted, e.g. with the Girls' Day.
On Girls' Day, schoolgirls can gain insight into occupational fields that girls rarely
consider in the process of career orientation. Technical companies and departments as
well as universities, research centers and similar institutions primarily offer events for
girls on Girls' Day and register these in advance in the Girls' Day website. By means of
practical examples, the girls experience in laboratories, offices and workshops how in-
teresting and exciting this work can be [2]. Since 2001 there have been 1.8 million Girls'
Day places [3].
The Neuromarketing Competence Centre together with the Internet of Things course
of Aalen University offer a course for girls. The theme was ‘Create your own dance
party’. Besides an interactive internet game 'Dance party', and other games on code.org,
a game with the Programming language Python on codecombat.com, a Calliope mini
(small minicomputer) and a mathematical problem (converting binary codes), there was
an eye tracker study in which the students were observed in an internet programming
game named 'Elsa'. The aim of the study was to examine how gender stereotypical
priming [4] affects girls. The eye tracker and the corresponding software Imotions,
which recognizes the basic emotions by facial analysis, are intended to recognize the
implicit impressions of the girls. The explicit impressions are queried by two question-
naires.
Women are more often implicitly associated with humanities, family and domesti-
city, equality or flat hierarchies than men. Men are more often implicitly associated
with natural sciences, mathematics, career, hierarchy and great authority [5] [6] [7]. In
a 2008 study by Ortner and Sieverding, female test subjects had to complete a task of
spatial imagination. Women performed better in the task if they imagined themselves
to be a stereotypical man compared to the idea of a stereotypical woman [8].
2 Research Goal
Can male priming possibly be used to make women better in these areas? We have put
forward three hypotheses:
H(1): If the participants are primed to ‘Anti-STEM’, the fear of programming is
greater than with ‘Pro-STEM Priming’. No particular reaction occurs with any
priming.
H(2): The priming reference to Anti-STEM serves a typical cliché and distracts
the participants from the actual task.
H(3): Anti-STEM Priming causes the participants to perform the programming
task worse than the ‘Pro-STEM’ primed or those who were not primed at all.
3 Research environment. Girls' Day Program and Eye
tracking-System and Software
A total of nine schoolgirls, aged between 13 and 15 from the 8th and 9th grades, came
to the Girls' Day. They all come from the same local high school. In advance, they have
received a consent and privacy statement for their parents to bring on Girls' Day to
participate in the study. One schoolgirl was not allowed to take part in the study because
her parents did not agree.
After the greeting of the participants, the girls had to complete the following tasks
in 3.5 hours: Dance party, Elsa and Angry Bird (browser-based programming games
from code.org), Calliope mini piano and binary code calculation. The girls could or-
ganize their time freely. The girls wanted to work in groups, so they formed three teams
of two and one of three. At the beginning, each participant received a number for the
eye tracker study (No. 2 to 10). They were called one after the other to participate in
the study.
The Tobii Pro X2-30 system from Tobii Technology was used to record the eye data.
This eye tracker can be attached directly to the laptop and is primarily used for non-
contact measurements. Consequently, the subject is in a fixed position, but after a suc-
cessful calibration he can move freely within a certain radius. The Tobii Eye-tracker
measures the subject's eye position with an accuracy of 0.4 degrees and a sampling rate
of 30 Hz. The eye tracker has several infrared lamps and a high-resolution video cam-
era. [9]
The iMotions software was used to record, process and analyze the eye tracking data.
This software is particularly suitable for usability studies for the investigation of web
pages, because the software has extensive analysis and visualization possibilities of eye
tracking data. It offers the possibility to capture and process the data quickly. In addi-
tion, meaningful visualization options and statistics are provided. The software was
used in version 7.1. [9]
Structure of the Study. The aim of the study was to see whether priming texts, i.e.
texts that can influence the person, have an effect on the result of the person.
The pupils were divided into three groups -independent of the group formation for the
tasks:
1st group Pro-STEM (test subjects no. 2, 3 and 5)
2nd group Anti-STEM (test subjects no. 6, 7 and 8)
3rd group no priming (subjects no. 9 and 10)
It was not told to them that they had been divided into these groups and that they
were primed.
The first group received a text in which priming should have a positive effect on the
result [10] [11]. The text shows that a girl is successful in the STEM subjects (mathe-
matics, computer science, science and technology).
For comparison, the second group received a text in which priming should have a neg-
ative effect on the result [8]. The text shows that a girl has no interest in STEM subjects
because "boys are better".
The third group received no priming text.
The procedure of a study is as follows:
1. Welcoming the test person and checking the declaration of consent
2. Queries: glasses/contact lens, eye problems
3. Adjustment of the monitor position on the test persons
4. Calibration of the eye tracker on the test persons
5. Conducting the eye tracker study with subsequent questionnaire
6. Farewell of the test person.
When asked whether the test subjects needed seeing aids, it came out that four test
subjects wear glasses and one of them suffers from an eye disease. All four pupils wore
glasses during the study. The quality of the eye tracker recording was between 56% and
91%. For those who did not wear glasses, the quality was between 95% and 98%.
4 Research Outcomes
In the following the emotions are compared before and during the task.
4.1 Emotions of the Respondents on the Introduction Texts before the Task
Pro-STEM Text. Three schoolgirls, here also called test subjects (No. 2, 3 & 5), were
given a Priming Text - Pro-STEM (Figure 1) [11]- to read at the beginning of the study.
Fig. 1. Cutout of Pro-STEM Text
The three test subjects mostly showed neutral or positive emotions when reading the
text. In respondent no. 2, there is a higher value at the beginning for the emotion ‘fear’.
At the end of the text, the diagram shows a small negative emotion, but this has nothing
to do with the test, since the respondent did not look into the camera but spoke to the
study director.
All three have recurring high values for 'joy'. The values are about 80 to 99.9
out of 100.
Each of the group member is ‘Surprised', but have different values (values be-
tween approx. 54 to 98 of 100).
Towards the end, higher values accumulate for negative emotions ('disgust’,
‘contempt’, ‘sadness’). These values lie between about 30 and 95 out of 100.
This could also be due to increasing time pressure at the end of the task.
Task at Anti-STEM. Three test subjects (No. 6, 7+8) have been given a Priming Text
- Anti-STEM (Figure 2) - to read at the beginning of the study.
Fig 2. Cutout of Priming Text Anti-STEM
The three test subjects showed neutral, negative or positive emotions when reading
the text. Test person no. 6 shows few positive emotions, but rather negative emotions
such as ‘contempt’ and ‘disgust’ at the beginning.
Fig. 3. Priming Texts Anti-STEM Summed
In summary, the emotions of the priming group Anti-STEM (Figure 10) show that
negative emotions occur more frequently than positive emotions.
No Priming Text. The last group of the study consists of two probands, No. 9 and No.
10. This group had no priming text
The two test subjects showed different emotions while reading the text. Proband No. 9
felt ‘fear’, ‘contempt’, ‘sadness’ and ‘disgust’. Proband no. 10 only showed the emotion
‘fear’ at the beginning of the text.
Fig. 4: Emotions No Priming Text Summed
With this background H(1) can be seen as confirmed.
4.2 Emotions of the Test Persons During the Task
All test persons of the three groups played a programming game after the introductory
text. It is a block programming game with Elsa and Anne from the Disney film 'The Ice
Queen - Completely Frozen'. The game is available at Code.org [12] “Code.org® is a
nonprofit dedicated to expanding access to computer science in schools and increasing
participation by women and underrepresented minorities.” [13]. The game consists of
different levels. For each level, the respondents must complete a task. Predefined blocks
have to be put together to solve the task. Either the figure Elsa or the figure Anne show
the result on the left side of the browser. The levels consist of Elsa or Anne skating a
certain figure, e.g. a snowflake, through the block programming with the ice-skate, so
that the figure is created.
The respondents have twelve minutes to complete as many levels as possible. There
are two explanatory texts (at the beginning and in the middle) to explain block pro-
gramming. After twelve minutes, the browser is closed and the task is finished. Then
the test subjects have to fill out a small questionnaire.
Task at Pro-STEM. Test subjects took approximately one to two minutes to read the
first explanatory text for the block programming. Until then, they had hardly shown
any emotions, except subject 2 with ‘joy’.
All three have recurring high values for 'joy'. The values are about 80 to 99.9
out of 100.
Each of the group member is ‘Surprised', but have different values (values be-
tween approx. 54 to 98 of 100).
Towards the end, higher values accumulate for negative emotions ('disgust’,
‘contempt’, ‘sadness’). These values lie between about 30 and 95 out of 100.
This could also be due to increasing time pressure at the end of the task.
Task at Anti-STEM. Results of the website recording of the Anti-STEM group:
Proband No. 6 has predominantly 'joy' in the task (values approx. 93 to 99 of
100). There are only a few outliers with contempt (values approx. 70 to 86 of
100) and disgust (values approx. 7 to 14 of 100).
Proband no. 7 shows hardly any emotions. If then short sequences of 'surprise'
(values 14 of 100), 'fear' (values 30 to 43 of 100), 'contempt' (value 4 of 100)
and 'disgust' (values 34 to 44 of 100).
Proband no. 8 shows mixed emotions. In addition to almost consistently high
values for 'joy' (values: approx. 60 to 99.9 of 100), she also shows almost con-
sistently medium-high values for 'fear' (values approx. 32 to 83 of 100). Besides
there are in a few places high values with 'sadness' (values approx. 80 to 99 of
100), 'surprise' (values approx. 70 to 90 of 100) and 'disgust' (value approx. 75
of 100).
Task at No Priming Text. Results of the website recording of the probands with no
priming.
Test person no. 9 was often 'angry' (values between approx. 19 and 93 of 100).
However, she also had 'joy' with higher values (values between 96 and 99 out
of 100), a recurring 'sadness' at a lower level (values between 15 and 23 out of
100). From time to time, there were higher values for 'contempt' (about 72 to
98 out of 100) and 'disgust' (about 44 to 94 out of 100).
Test person no. 10 on the other hand, did not show so many emotions. There
were higher values for 'joy' (values between 76 and 98 of 100), 'contempt' (val-
ues between 69 and 96 of 100) and 'disgust' (values between 31 and 99 of 100).
4.3 Analysis of the Attention Distribution by Heatmaps
During the evaluation of the attention distribution, the mean values on the respective
AOIs (Areas of Interest) from the eye tracking analysis are calculated and illustrated by
heat maps. The red and orange colors indicate a focus of attention, the green color,
which is more widely distributed, indicates a reduced cognitive load on the test subjects.
Fig. 5. Heatmap of the Priming Text Pro-STEM
The heatmap of the Pro-STEM group (Figure 5) shows that the respondents are pri-
marily focused on the text describing the subsequent task (bottom paragraph). The fo-
cus of the eyes is less on the lower part of the priming section (middle section), which
states that the former Girls' Day participant uses the programming languages in her
daily work.
Fig. 6. Heatmap of the Priming Text Anti-STEM
Fig. 7. Heatmap of the Text with No-Priming
The heatmap of the Anti-STEM group (Figure 6) shows that the respondents are
primarily focused on the priming text (top paragraph). Here is the description of a for-
mer Girls’ Day participant who is not responsible in the STEM area and prefers ‘fe-
male-associated things’. The focus is then on the text for the subsequent task.
The heatmap of the text without priming (Figure 7) shows a focus at the end of the
text.
Therefore H(2) can also be seen as confirmed.
4.4 Comparison of the Results of the Task
The Pro-STEM group did not achieve more than level 6. In the group Anti-STEM, the
test subjects no. 6 and no. 7 have passed the 6th level and have mastered the 8th or 9th
level. Test person no. 8 did not complete the 6th level. In the group without priming,
subject no. 9 reached the 7th level, but subject no. 10 did not complete the 6th level.
When comparing the priming groups, it is noticeable that the Anti-STEM group suc-
cessfully completed more levels in the same time than the Pro-STEM group.
4.5 Questionnaire
After the Girls' Day, the test subjects were asked about their experiences with technol-
ogy, their hobbies, the work of their parents and siblings in a questionnaire. These val-
ues were compared with the data of the study. The following results were obtained:
Girls who indicated that they had already had something to do with technology suc-
cessfully completed more levels than other test subjects.
Test person no. 7, who sees herself in the future to 90% in a technical, mathematical,
scientific or IT-related occupational field (highest value of all test persons) also reached
the highest level of all participants (successfully completed level 9).
A high average daily use of the Internet does not necessarily lead to a good result in
the test. 62.5% of the participants said that their interest in technology increased after
the Girls' Day. In no case did the Girls' Day negatively influence their interest in tech-
nology. Knowledge of block programming had no effect on the success of the test. None
of the participants did not like the task, but only 37.5% of the participants would like
to learn more about block programming.
5 Discussion
The organizers were able to make the following observations during Girls' Day:
Questions related to math were not dealt with precisely, simple tasks could not
be solved (e.g. dividing a three-digit number by two).
Inaccurate reading of tasks: The Calliope mini piano was only tested to see if
things conduct electricity. The girls said, they already know the answers. How-
ever, their answers were only partly correct. The following parts were provided
for testing: a paper clip, a banana, a wooden clothespin, a wire, a plastic wire,
a piece of cloth and an eraser. They found it exciting that the banana was con-
ductive and then took a closer look at the piano. The others simply plugged the
things into the crocodile clips and looked to see if there was any sound coming.
A positive example are the dance parties of Code.org: the girls were very happy
to program their own dances with blocks. They use the time intensively to com-
bine their hobbies like dancing or movies/cinema with programming.
Most of the time they spent programming the dance party, than the other pro-
gramming games. Less time were spent for the Calliope mini and only little
time was used for the calculation with the binary code.
The questions were cribbed from each other, so the question cards could not be
included in the evaluation.
Block programming codes could be reproduced and questions could be an-
swered right. Python codes could also be applied.
In order to get the girls to program, they need something playful in which they can
participate themselves, such as dancing. The tasks should not look technical; with ‘real’
interactions, the girls could be persuaded to participate. Like, ‘I enter a code and then
something happens’.
In the priming texts, the test subjects predominantly showed the following emotions:
Pro-STEM: Positive emotions
Anti-STEM: Negative emotions
No priming: negative emotion
The Anti-STEM Group has completed its tasks most successfully. Therefore the
H(3) cannot be regarded as confirmed. Accordingly, in this study it could not be found
that male priming improves women in certain areas [8]. However, this may also be due
to the age of the test subjects and priming at that age is the opposite. The girls in the
study are in their puberty and behave accordingly. The reaction to the Anti-STEM text
could be a defiant reaction (‘I'll show them how it works’). This may also indicate that
young women at this age can no longer be primed so easily and that priming takes place
earlier, e.g. in childhood. For this a separate test series would be promising.
6 Conclusion
In order to bring programming with less negative emotions closer to the girls, they must
be brought to the STEM subjects earlier, as it becomes more difficult during puberty to
get them enthusiastic about these topics. Only hip things like Disney movies or current
charts in connection with games (keyword gamification) can make it enjoyable for girls
to deal with the topic. In addition, learning must take place in schools, as they are not
motivated to learn programming in their free time. In the school they should teach ac-
cording to a smart education of universities: "1. adaptation, 2. sensing, 3. inferring, 4.
self-learning, 5. anticipation and 6. self-organization and re-structuring [14, 15].
7 Next Steps
The future steps are a) research with a larger group of females, b) compare with males
in the same age and c) research with younger females (and males) and compare it with
the other results. Also, one can analyse at what age which methods for STEM-Subjects
are better.
References
1. Statistik der Bundesagentur für Arbeit Berichte: Blickpunkt Arbeitsmarkt–MINT -Berufe,
Nürnberg, September 2018 https://statistik.arbeitsagentur.de/Statischer-Content/Arbeits-
marktberichte/Berufe/generische-Publikationen/Broschuere-MINT.pdf. (access: 2019-05-
26).
2. Girls’ Day (2019). Was ist der Girls’ Day: https://www.girls-day.de/Footer/Haeufige-Fra-
gen (access: 2019-05-16)
3. Reker, J. (2019). Girls' Day-Auftakt 2019: Mädchen für MINT-Berufe begeistern, in
https://www.girls-day.de/Footer/Presse/Pressemitteilungen/Girls-Day-Auftakt-2019-Ma-
edchen-fuer-STEM-Berufe-begeistern (access: 2019-05-26).
4. Chatard, A., Guimond, S. & Selimbegovic, L. (2007). ‘How good are you in math?’ The
effect of gender stereotypes on students' recollection of their school marks. Journal of Ex-
perimental Social Psychology, 43 (6), 1017-1024.
5. Fine, C. (2010). Die Geschlechter Lüge. Die Macht der Vorurteile über Frau und Mann.
/Delusions of Gender. The Real Science behind Sex Differences. How Our Minds, Society,
and Neurosexism Create Difference. W.W. Norton & Company, New York, London.
6. Mast, M. S. (2004). Men are hierarchical, women are egalitarian: An implicit gender stere-
otype. Swiss Journal of Psychology, 63 (2), 107-111.
7. Rudman, L.A. & Kilianski, S.E. (2000). Implicit and explicit attitudes towards female au-
thority. Personality and Social Psychology Bulletin, 26 (11), 1315-1328.
8. Ortner, Tuulia & Sieverding, Monika. (2008). Where are the Gender Differences? Male
Priming Boosts Spatial Skills in Women. Sex Roles. 59. 274-281. 10.1007/s11199-008-
9448-9.
9. iMotions. Tobii X2-30, https://imotions.com/tobii-x2-30/, (access: 2019-05-23), p. 1.
10. Alexander D. Stajkovic, Edwin A. Locke, and Eden S. Blair: A First Examination of the
Relationships Between Primed Subconscious Goals, Assigned Conscious Goals, and Task
Performance. Journal of Applied Psychology 91, 2006, S. 1172-1180
11. McIntyre, R.B. & Lord, C.G. & Gresky, Dana & Ten Eyck, L.L. & Frye, G.D.J. & Bond
Jr, C.F.. (2005). A social impact trend in the effects of role models on alleviating women's
mathematics stereotype threat. Curr Res Soc Psychol. 10. 116-136.
12. Code.org Frozen (2019). https://studio.code.org/s/frozen/stage/1/puzzle/1'. (access: 2019-
05-26).
13. Code.org (2019). https://code.org/international/about (access: 2019-05-26).
14. Uskov, V.L., Bakken, J.P., Pandey, A.: The ontology of next generation smart classrooms.
In:Proceedings of the 2nd International Conference on Smart Education and e-Learning
SEEL-2016, 17–19 June 2015. Sorrento, Italy, Springer, pp. 1–11 (2015)
15. Uskov, V.L, Bakken, J.P., Pandey, A., Singh, U., Yalamanchili, M., Penumatsa, A.:
Smart University taxonomy: features, components, systems. In: Uskov, V.L., Howlett, R.J.,
Jain, L.C. (eds.) Smart Education and e-Learning 2016, Springer, pp. 3–14, June 2016, 643
p., ISBN:978-3-319-39689-7 (2016)
