Combining Cognitive Styles Matters for Female Software Designers

Overcoming society’s complex problems requires novel solutions. Applying different cognitive styles can promote novelty when designing software aimed at these problems. Through an experiment with 80 software design practitioners, we found that female practitioners who had a preference for more than one cognitive style (intuition and rationality) produced the most novel software features of all participants. THERE IS CONSENSUS in the software engineering community that practitioners sometimes rely on their intuition when designing software. Despite this, the emphasis in the software development process has generally been on promoting a rational cognitive style through rationalized processes, tools, and techniques. Meanwhile, the potential benefits of an intuitive style have been largely ignored [1]. One such benefit of intuition is novelty [2], which is crucial for tackling complex societal problems such as inequality, climate change, and health. To address this gap, we carried out an experiment in which software design practitioners with different cognitive styles designed software features for a mobile application to address a widespread health behavior problem. We found that female practitioners produced more novel software features than male practitioners, especially when they were both highly rational and highly intuitive. Our study highlights the importance of considering and combining cognitive styles when designing new software features, but shows that female practitioners may uniquely benefit from combining intuitive and rational cognitive styles. WHY FEATURE NOVELTY? When designing software for a complex problem, software design practitioners, like product designers and requirements engineers, create new features to (partially) solve the problem. These practitioners tend to start off by sketching various ideas for a design on a whiteboard or piece of paper [3]. They will then cycle back and forth between their understanding of the problem and their idea(s) for a potential feature, updating these concurrently as they go along. Nowadays, software solutions naturally lend themselves to addressing societal problems. However, the reality is that such problems demand substantial levels of novelty in software features [4]. COGNITIVE STYLE AND GENDER MEET FEATURE NOVELTY Software design practitioners, like all people, have different cognitive styles. Cognitive style describes differences in how people obtain, organize, and process information [5]. Intuition and rationality are two such cognitive styles. A practitioner designing a software feature through an intuitive style might do so quickly, and have a gut feeling that their solution is the right one. IEEE Software Published by the IEEE Computer Society © 2020 IEEE 1 ar X iv :2 01 2. 05 56 3v 1 [ cs .S E ] 1 0 D ec 2 02 0 Conversely, a practitioner using a rational style would arrive at a particular feature more slowly, justifying their solution in the context of available requirements. Both styles can be used by a practitioner at any time, in a particular order or even simultaneously. Still, all people tend to usually rely on one or both styles in a specific configuration [6], known as their dispositional style. Both intuition and rationality have been positively related to novelty. Intuition has been shown to result in more novel solutions through holistic information processing and promoting associative thinking [2]; the “big picture”. Rationality enables practitioners to assess details, and to analytically compare potential solutions [6], [7]. Nevertheless, whether this is specifically true for software design practice remains to be seen. Although dispositional style is not inherently gender-specific, it has been shown that the interaction between gender and job type can influence preference for intuition [7]. Given that female practitioners are often underrepresented in software engineering [8], endure unique barriers to entering the field [9], and are subject to a number of different biases [10], we were particularly curious about whether the novelty of software features designed by software design practitioners would vary based on their gender and dispositional style. When speaking of male and female in our study, we take gender to be a self-identification construct, which may or may not align with biology or presentation [11]. Given these potential associations between cognitive style and feature novelty, and gender differences in style preference, our study investigated whether certain combination(s) of cognitive style and gender led to higher software feature novelty. STUDY DESIGN We conducted an experiment with practitioners to enable some control, while still maintaining real-world applicability. Practitioners, whose primary task involves high level design of features in any software engineering role, were recruited through the online platform Prolific. Such participants are familiar with the complexity of the task, and comfortable with producing rough, wireframe-like sketches. First, participants took part in a feature design task. Afterwards, the same participants were randomly assigned to evaluate the novelty of ten features designed by others. We chose to focus on the health issue of obesity as our context, being a well-known issue that participants would at least be familiar with. Feature design task Participants were given an explanation of the problem, and instructed to design at least one feature for a mobile application. They were then given 15 minutes to sketch their software feature(s) on a piece of paper and provide suitable explanations, using a basic template as per [3]. Figure 1 presents a selection of the designed features. Afterwards, we asked the participants to note which of their features, if they designed more than one, solved the problem best. The participants then photographed or scanned their features for upload. To measure participants’ dispositional cognitive style, we used the REI-10 (rational experiential inventory), which consists of five statements about participants’ use of intuition and five about their use of rationality [6], measured on a 7-point scale from “completely disagree” to “completely agree.” We dropped one item from the rationality scale that reduced the scale validity. We collected participant’s self-identified gender in the same section in which we asked control questions about work-relevant experience, industry role, age, and familiarity with the obesity problem. Participants were paid four English pounds for completing this part of the study. Feature evaluation task After completing the design task, participants were contacted again, and randomly formed into groups of five participants. Each group evaluated the same ten randomly selected features (always excluding their own). For each feature design sketch, participants were required to answer the question, “How novel is this feature when compared with existing features from applications in the market?” Answers were recorded on a fivepoint scale ranging from “not novel at all” to “extremely novel”. Participants were paid two English pounds for completing the evaluation of ten features.