Why do students drop computing science?
Matthew Barr & Maria Kallia
- Eccles and Marsh Framework
- Our Participants
- Our findings may be summarised as follows
- Video presentation
- Song of the Day
In many parts of the world – including Scotland – there is a shortage of Computing Science graduates. And we are particularly short on female Computing Science (CS) graduates. There are lots of reasons for this, and prior work has found that students who drop Computing Science cited a lack of time, poor study skills, and prioritisation of other subjects as the reasons for doing so, for example. However, these reasons are underpinned by a range of other factors, such as the perceived difficulty of the subject.
And, of course, gender is a big factor here. There’s the “macho geeks problem”, for example, referring to the geeky “know it all” male culture that dominates CS classrooms. Meanwhile, Taylor-Smith et al. note that the exclusion of women from CS is a self-perpetuating problem, reinforced through gender stereotypes of computing. Finally, there’s the absence of support that might help ensure that female students continue to study the subject.
Eccles and Marsh Framework
To understand why our students drop CS, in our recent study, we draw from two theoretical models. The first is Eccles’ expectancy-value model of achievement which may be used explain students’ choices, persistence, and performance in a subject. The second is the Generalized Internal/External Frame of Reference Model, developed by Marsh, which concerns students’ domain-specific self-concepts of ability.
The basic premise underlying Eccles’ model is that an individual’s choice to work within a certain domain, and their motivation to perform a task, can be explained by two factors: our expectancy of success, and subjective value we place on the task. This subjective task value can include the utility of the task in meeting our personal goals, our enjoyment of the activity, and the perceived cost of participating in it – this includes the emotional costs and the time taken away from other activities. In short, the model predicts that students’ choices regarding their occupation or course choices are influenced by the value they assign to the tasks available to them.
The Generalized Internal/External Frame of Reference Model is about self-concept, or how we view ourself. And in educational psychology, self-concept is often characterized as our individual beliefs about our strengths and weaknesses, and is a major predictor of academic choices. Self-concepts are shaped by both external and internal frames. The external frame is the comparison students make between self-perceptions of their ability in one subject and the perceived ability of other students in the same subject. Through the internal frame, students assess their self-perceived abilities in one subject against their self-perceived abilities in another subject. So, we can say that students use their perceived performance and that of their classmates to create frames of reference for self-evaluation.
In our study, we surveyed current third and fourth year students at the University of Glasgow who had taken CS in first or second year, but were no longer enrolled in any CS courses. The survey asked respondents to explain why they didn’t continue to study CS, and how their gender affected their experience of studying CS, if at all. The responses to each of open questions were analysed by both of us using a deductive thematic approach, with themes drawn from the models previously described.
We identified eight themes in the data, including two emergent themes that weren’t easily explained by the Eccles or Marsh models: Social factors and the Difficulty of the subject.
|Eccles Model||% of participants who referred to each theme|
|Expectancy of success||13%|
|Marsh Model||% of participants who referred to each theme|
|Emergent Themes||% of participants who referred to each theme|
The Utility theme refers to the degree of usefulness of Computing Science to students’ future goals. Meanwhile, the Cost theme reflects how the respondents’ decision to continue with CS limited access to other activities or required a lot of time and emotional effort. Our analysis revealed two sub-themes here: time, referring to the time and effort of engaging with CS and affect, referring to the emotional impact of engaging with CS. Here, Female and male students placed similar importance on the cost of engaging with CS.
The intrinsic value theme groups together factors referring to students’ interests in studying CS for reasons intrinsic to computing, like enjoying the subject. Slightly more female respondents referred to this theme than males. Overall, slightly more female students (18%) referred to expectancy of success than men (10%). The Attainment theme includes students referring to their self-image, personal values and beliefs – it relates to students’ identity. Most of the references to this theme were reported by female students (27%) and only 5% by male students.
The Comparisons theme is derived from Marsh’s framework, including both the internal and the external frames. And, while in the framework described by Marsh, the Internal frame focuses on achievement comparisons between subjects, our students included other comparison measures such as the time and effort required by two subjects – i.e. the cost. Here, more female students (36%) made subject comparisons than male students (19%). Interestingly, the external frame, which refers to comparisons between a student’s perceived ability and that of their peers, was only reported by female students; male students did not refer to comparisons with their peers’ abilities as a factor in dropping CS.
The emergent Social theme refers to factors related to having a supportive environment and a social network that students can draw upon to ask for help or to make them feel that they belong to CS. In comparison with other themes, social factors are significantly more important for female respondents than male respondents. In total, 55% of female students reported social factors affecting their decision to drop CS in comparison to only 5% of male respondents. The second emergent theme, Difficulty, reflects the perceived difficulty of CS as a subject. Interestingly, the percentage of female students who reported that the course’s difficulty was one of the main factors for not continuing with CS was more than double
the percentage of males.
Our findings may be summarised as follows
(although we suggest reading the paper!):
- Subjective task value played a much more significant role than the expectancy of success.
- Utility and cost, are more important than attainment value.
- Students’ comparisons between themselves and their peers, and between CS and other subjects, are also a factor.
- Course difficulty and social concerns were also emphasised by our participants.
- Comparisons with peers, social concerns, perceived subject difficulty, and issues of attainment associated with self-concept all play a more significant role in female students’ decision to drop CS.
There’s more work to be done here, especially in understanding why female students – and perhaps other minority groups – drop Computing Science. In order to delve a bit deeper, our future work will be based on more in-depth semi-structured interviews, designed with reference to the models and the preliminary findings presented here.
Matthew Barr and Maria Kallia. 2022. Why Students Drop Computing Science: Using Models of Motivation to Understand Student Attrition and Retention. In Koli Calling ’22: 22nd Koli Calling International Conference on Computing Education Research (Koli 2022). Association for Computing Machinery, New York, NY, USA, Article 17, 1–6. https://doi.org/10.1145/3564721.3564733
Jacquelynne S Eccles. 1994. Understanding women’s educational and occupational choices: Applying the Eccles et al. model of achievement-related choices. Psychology of Women Quarterly 18, 4 (1994), 585–609.
Päivi Kinnunen and Lauri Malmi. 2006. Why students drop out CS1 course?. In Proceedings of the Second International Workshop on Computing Education Research (ICER ’06). Association for Computing Machinery, New York, NY, USA, 97–108. https://doi.org/10.1145/1151588.1151604
Herbert W Marsh. 1986. Verbal and math self-concepts: An internal/external frame of reference model. American Educational Research journal 23, 1 (1986), 129–149.
Andrew Petersen, Michelle Craig, Jennifer Campbell, and Anya Tafliovich. 2016. Revisiting why students drop CS1. In Proceedings of the 16th Koli Calling International Conference on Computing Education Research (Koli Calling ’16). Association for Computing Machinery, New York, NY, USA, 71–80. https://doi.org/10.1145/2999541.2999552
Ella Taylor-Smith, Camilla Barnett, Matthew Barr, and Carron Shankland. 2022. Participant-Centred Planning Framework for Effective Gender Balance Activities in Tech. In Proceedings of the 2022 United Kingdom and Ireland Computing Education Research Conference. Association for Computing Machinery, Dublin, Ireland, 7. https://doi.org/10.1145/3555009.3555016