Women and computers. Effects of stereotype threat on attribution of failure

Abstract

This study investigated whether stereotype threat can influence women’s attributions of failure in a computer task. Male and female college-age students (n = 86, 16–21 years old) from Germany were asked to work on a computer task and were hinted beforehand that in this task, either (a) men usually perform better than women do (negative threat condition), or (b) women usually perform better than men do (positive condition), or (c) they received no threat or gender-related information (control group). The final part of the task was prepared to provide an experience of failure: due to a faulty USB-memory stick, completion of the task was not possible. Results suggest a stereotype threat effect on women’s attribution of failure: in the negative threat condition, women attributed the failure more internally (to their own inability), and men more externally (to the faulty technical equipment). In the positive and control conditions, no significant gender differences in attribution emerged.

Introduction

Computer skills are key qualifications in the professional world (Todman, 2000). Yet, women are generally found to use computers and the Internet to a lesser extent than men do (Cooper & Weaver, 2003). In addition, a social stereotype exists that men are more competent at computer use than women are. This study investigated whether stereotype threat can be a factor contributing to the differential attribution patterns in a computer task in college-age students and can thus eventually lead to lesser computer use or more ready resignation in the use of computers by women. We focused on the attribution patterns of men and women after failure as important prerequisites for their persistence to solve computer problems.

Although the access of women to computers has increased substantially since the 1980s and early 1990s (Imhof, Vollmeyer, & Beierlein, 2007), a gender gap in computer use still exists across all age groups and cultures, which seems to be strongest in high school and college students (Li and Kirkup, 2007, Papastergiou and Solomonidou, 2005, Whitley, 1997). Colley and Comber (2003) compared recent findings on computer experience and attitudes of students (of age 11–12 and 15−16 years) within a period of 10 years. Along with gender differences in computer attitudes, boys were still found to use the computer more frequently in non-school contexts (e.g., playing games) than girls did. Boys liked computers more and were more self-confident in using computers. In both studies, 15–16-year-old girls had the least positive attitude toward computers.

A similar difference was also shown by the OECD-report of the PISA study (2005), which compiled data about the educational conditions and performance levels of pupils worldwide. Despite the existing gender equality in computer access, 15-year-old boys owned a computer more often, used a computer more frequently for programming, playing games, or for entertainment, and spent more time on the Internet than coeval girls did. Girls and women, on the other hand, had less self-confidence in handling the computer, particularly in complex tasks such as preparing multimedia presentations. Given that computer use has a positive impact on school math grades (Rocheleau, 1995), these findings can imply disadvantages for girls (OECD, 2005).

What are the reasons for this gender gap? Factual barriers to access and differences in the use of computers have substantially decreased (Cooper, 2006, Cooper and Weaver, 2003, Todman, 2000) or even vanished (Wasserman & Richmond-Abbott, 2005). Furthermore, there are no physical or cognitive obstacles that might disadvantage one gender group over the other. Normally, men and women should be equally suited for carrying out professional skills in the computer domain. The possible reasons for women not living up to their potential are the lack of female role models who use computers (Marx & Roman, 2002), gender-specific differences in support from parents and teachers (Busch, 1996), and gender specificity of the vast majority of computer software (Cooper, 2006). Potential psychological obstacles inhibiting the computer use by girls and women are unfavorable computer attitudes (Anderson et al., 2008, Lee, 2003, Shashaani, 1997, Whitley, 1997), low or missing computer self-efficacy (e.g., Cassidy and Eachus, 2002, Jackson et al., 2001), particularly in high school students (Whitley, 1997), computer anxiety (Chua, Chen, & Wong, 1999), and unfavorable attribution patterns (Dickhäuser and Stiensmeier-Pelster, 2002, Nelson and Cooper, 1997, Rozell and Gardner, 1999). Our study focuses on the aspect of unfavorable attribution patterns.

In an early study, Deaux and Emswiller (1974) found that the (successful) performance of male stimulus persons was more likely to be attributed to their skills (internal attribution), whereas the performance of the female stimulus persons was more likely to be attributed to luck (external attribution). Studies that tried to replicate these findings have yielded diverse results (e.g., Hill and Augoustinos, 1997, Swim and Sanna, 1996). Some studies focused on gender-specific attribution patterns in the self-attributions of success and failure (Burgner and Hewstone, 1993, Nelson and Cooper, 1997, Rozell and Gardner, 1999, Whitley et al., 1986). For example, Dickhäuser and Stiensmeier-Pelster (2002) conducted two studies to investigate the gender differences in self-related attribution patterns in the computer domain. They provided different scenarios, such as “imagine that you cannot open a file you have previously saved on a disk”. Participants then had to name the reasons for this event. The authors confirmed the hypothesized pattern: men preferred the external causal attribution “faulty disk”, women preferred the internal attribution “lack of own knowledge”. In a second study, the authors found that lack of own knowledge or ability resulted in a higher levels of shame and lower expectations of success. Thus, the affective valence of the computer was directly related to attribution patterns. However, will attribution patterns be differentially visible between men and women, when the stereotype related to “women and computers” is activated to different degrees? This research question can be addressed by choosing a stereotype threat approach to gender in the computer domain.

Smith, Morgan, and White (2005) have demonstrated that there is a stereotype that women do worse at computers. The stereotype threat paradigm (Steele, 1997, Steele and Aronson, 1995) puts forth that people belonging to minority groups suffer from performance impairments when a negative task-relevant stereotype concerning their ingroup becomes salient. When minority group members perform a difficult task in an area, in which the ingroup is considered weak, they feel at risk of confirming the stereotype. This psychological pressure can lead them to underperform. A growing body of research has shown this effect for the domain of “women and math performance” (e.g., Spencer, Steele, & Quinn, 1999). Women, who were told that they were about to take up a math test, where men usually perform better than women do, typically underperformed in the test. In contrast, women typically perform better when told that in this test there are no gender differences (Spencer et al., 1999). For the stereotype threat to be effective, a number of conditions need to be fulfilled: (a) individuals need to show a certain degree of identification with the domain (DI) on the one hand, and, on the other hand, the social group that is subject to the threat; (b) the task needs to have a high level of difficulty to truly differentiate between the stereotyped groups; and (c) the stereotype needs to be of a certain relevance to the situation, in which participants encounter themselves while taking the test (Steele, 1997).

To date, the stereotype threat theory has been only rarely applied to gender in the computer domain. We found two studies: Cooper (2006) had students organize a presentation in Power Point and showed that female students primed with their identity as a woman immediately before a computer task (by writing a report about social life on campus, dating, partying, etc.), performed worse than those who were primed with their identity as a student (i.e., by writing a report about courses and assignments). Mauch (2008) showed that implicit threat caused by the group gender composition had direct effects on the performance at the computer: women performed best in same-sex groups, whereas men performed best in mixed-sex groups. So far, no study has addressed the influence of stereotype threat on computer-related attribution patterns.

The goal of our study was to test the differences in attribution of failure after stereotype threat in 16–21-year-old students of the “Gymnasiale Oberstufe” in the German school system (i.e., 11th to 13th grade). The study innovatively tested the psychological influence of stereotypes on gender differences in the computer domain, and attribution patterns of failure following the threat. For the main manipulation, we implemented the negative threat condition in the conventional manner, by informing the participants that men usually outperformed women in this computer test. In addition, we introduced a positive condition to investigate whether the positive information about women’s test results can prevent a self-handicapping attribution of failure in women. A number of studies showed that the stereotyped group’s performance increased after having been informed that their social group is expected to perform well on the task (e.g., Cadinu et al., 2003, Stangor et al., 1998). In our study, we wanted to know whether this would also be valid for women’s use of more beneficial attribution patterns after failure. Finally, we provided no threat-related information to the control group.

In a quasi-experimental 2 × 3-design (sex × threat condition), we tested the following hypotheses: (a) women in the negative threat and in the control condition attribute failure in a computer-related task internally, to their own inability, and (b) women in the positive condition and men in all three conditions attribute the failure externally, to the faulty technical equipment (Deaux and Emswiller, 1974, Dickhäuser and Stiensmeier-Pelster, 2002). In our analysis, we controlled the influence of computer self-efficacy, computer-related intrinsic motivation, and practical computer knowledge at baseline.