Sei sulla pagina 1di 6

AMERICAN JOURNAL OF SCIENTIFIC AND INDUSTRIAL RESEARCH

© 2015,Science Huβ, http://www.scihub.org/AJSIR


ISSN: 2153-649X, doi:10.5251/ajsir.2016.7.1.1.9

Computer programming skill and gender difference: An empirical study


Solomon Olalekan AKINOLA
akinola.olalekan@dlc.ui.edu.ng

Department of Computer Science, University of Ibadan, Nigeria


ABSTRACT

Women have been said to be poorly represented in computing most especially in computer
programming. However, there is a dearth of empirical research to support this evidence. This
study employed an experimental proof approach to verify the wide acclaimed gender difference in
computing. Two parallel student-gender groups (male and Female) were subjected to two
different computer programming problems. The problems were given to them to solve, starting
from the analysis phase to the final implementation. Their outcomes measured in terms of the
accuracy and efficiency of codes turned out from the experiments were then compared
statistically. Briefly, the results showed that gender difference may not come into play at all when
it comes to computer programming. Other factors like fear, interest or attitudes may however be
responsible for the low representation of females in computing.

Keywords: Gender difference, Computer programming skill, Empirical study, Collaborative


Programming, Pair programming.

INTRODUCTION United State citizens' estimated 2000 computer


programmers were women (Lockheed, 1993).
Computer Programming is known to be a complex
Interestingly, the percentage of female computer
task that is difficulty to do. Consequently, learning
programmers began to decline sharply in the 1980s
computer programming is one of the first and most
with advent of the personal computer or more
challenging tasks encountered by computing
specifically, when computer began to appear in
students. The difficulties faced by the students can
homes. Even as job opportunities for programmers
be perceived by the high degrees of failure and the
and software developers peaked in the 1990s, the
difficulties presented in the courses directly
number of women pursing such opportunities
dependent on the abilities to program, to develop a
continued to decline. According to U. S. Department
logical reasoning and to solve problems. In part, this
of Labor, Bureau of Labor Statistics as at 2012,
is the result of the difficulties found by instructors to
computer programming is very much a male
effectively guide the students during their
dominated field, with women making up only 22.5%
programming lab activities, due to the large number
of computer programmers nationwide, with not even
of students per class (Eustáquio et al., 2006).
half that percentage of women represented in
In the work of Dawn (2005), women have been highly software development and engineering jobs. The big
influential in the field of computer programming since question is “Does gender difference really affects the
its supposed inception in the 1830s. Ada Byron, programming skills of the professionals?”
Countess of Lovelace and famed collaborator of
Previous studies in this area of research have used
Charles Babbage, is often credited as being the
observation, interview or questionnaire instruments to
world's first computer programmer. Perhaps even
study the gender difference in computing. However,
more influential to the field was naval officer Grace
this study was designed to empirically proof that no
Murray Hopper contributed to the development of
gender is better than the other in programming task
COBOL (Common Business Oriented Language), the
using experimental proof approach.
first business-oriented programming language,
released in 1960 and still in use today. These two The rest of this paper is organized as follows.
women are the most notable from a tradition of Related works are presented in section 2 while the
female computer programmers that persisted research methodology is presented in Section 3. In
throughout 1960s. For example in 1960, 65% of the

1
Am. J. Sci. Ind. Res., 2016, 7(1): 1-9

section 4, we present and discuss the results forming a solution, completing it as quickly as
obtained in the study. Section 5 concludes the paper. possible, and then submit it for judging. Women will
be more likely to devote more time to conceptualizing
Related Works: Women are severely
a solution, spend longer working on an individual
underrepresented in the field of Computer Science.
solution, and then submit it for judging. This
Attitude of women toward computing is one of several
difference means that women will have less
factors that might explain the low participation of
opportunity to try various solutions, in the belief that
women in computing. Programming is not alluring to
they have increased their accuracy by spending
females. This could be attributed to females' low
longer on a solution (Maryanne and Anthony, 2006).
confidence and programming abilities. Therefore,
their attitudes towards programming are more Thus, within the context of computer programming
negative than males as found by Korkmaz and Altun contests, women may be perceived as less
(2013). competent because they adopt less risky
approaches. This matter is even more deleterious,
Gender influences how people will perform within
given that stereotypical beliefs, such as “people view
computing contests, although it is presumable that
me as less competent because I am female” often
there are numerous other variables, such as
lead directly to depressed performance (Steele and
ethnicity, personality, experience, and so on. In
Aronson, 1995). Second, rather than gambling on
addition, casual observations, as well as various
their ability to guess the correct approach, women will
reports (for example, Helgeson, 2005), reveal a
take more time to explore one solution as opposed to
systemic underrepresentation of women in Computer
implementing a “fast and dirty” solution that may not
Science (Maryanne and Anthony, 2006).
work. This less risky strategy will allow less time for
Severiens and Dam (1994) performed a meta- trying a wider assortment of potential solutions, and
analysis of the gender effect on two learning style less time to attempt remaining questions. Third,
instruments, namely Kolb’s Learning Style Inventory women will not guess at a response as readily as
and Entwistle’s Approaches to Studying Inventory, men, nor direct team mates to consider a possible
and found a significant gender preference on solution, when they are not sure of themselves. They
subscales of these instruments. Men tended to prefer will also feel uncomfortable when, in a mixed-sex
the abstract conceptualization mode of learning more team, they are asked to pursue high-risk strategies
than women, according to Kolb’s instrument. Men (Maryanne and Anthony, 2006).
were likely to adopt a deep approach, be extrinsically
Anxiety can play important role in gender computer
motivated and achievement oriented. Although
programming. Women experience a decrease in self-
females are able and willing to compete, and do so
confidence when they perceive that their work will be
well in specific situations, the widely held notion that
compared to others for evaluation (Lenney, 1997).
they tend to be more cooperative and less
When there is no pressure (i.e., in a non-competitive
competitive than males does hold merit. In general,
environment), there is equivalent enthusiasm
findings from social and developmental psychology
reported between women and men for computing
indicate that females prefer cooperative activities,
activities (Astrachan, 2004).
while boys prefer competitive activities. A study by
West (2001) show that there was no difference in Self-efficacy has been described as a critical element
learning styles between female majors, female non- in accomplishing tasks successfully. In developing
majors, and male majors in computing. and validating a computer user self-efficacy (CUSE)
scale, Cassidy and Eachus (2002) found that males
According to Cashdan (1999) as cited by Maryanne
had consistently higher self-efficacy than females
and Anthony, (2006), women often feel pressured by
irrespective of whether or not a computer training
popular stereotyping to be considered as passive and
course had been attended before. Perceived
feminine, and thus actively hide their competitive
masculinity of the task did not seem to explain the
nature. In general, women tend to choose high
gender differences in computer self-efficacy.
probability, low payoff strategies, whereas men tend
Experience with computers and familiarity with
to choose, under the same situations, low probably
computer software packages were the two most
but high payoff strategies (Daly and Wilson, 2001).
important predictors of computer self-efficacy. In a
This difference in strategy may lead to
literature review of research on the gender digital
differences in programming contest performance. It is
divide, Cooper (2006) argued that the digital divide
predicted that men will be more likely to rush at
can be boiled down to a problem of computer anxiety.

2
Am. J. Sci. Ind. Res., 2016, 7(1): 1-9

She further proposed a causal model incorporating second experiment involved 72 students in four
gender stereotype, attribution patterns, and collaborative team sizes, with 12 male and 6 female
stereotype threat as the predecessors of computer groups. Two different programming exercises were
anxiety, and computer anxiety as a predecessor of given the students to tackle during the experiments.
computer attitude and computer performance.
Variables: Two major independent variables were
Noreen (1985) investigated gender differences in
manipulated in the study: Gender and Team sizes.
planning and debugging behavior, group processes,
The only dependent variable measured was the
and achievement among junior high school students
outcomes of the individual groups in the experiments.
learning computer programming in small group or
The groups were scored based on the efficiency and
individual settings. Results from the study shows that
accuracy of the turned-out codes for 10 marks in
in neither study and in neither setting, group or
each of the experiments. Efficiency was determined
individual, did females and males differ in behavior or
via the total lines of code produced, use of proper
learning outcomes. Furthermore, the composition of
programming logic and control structures and use of
the group, the ratio of females to males, had little
Object Oriented concepts. The accuracy was
effect on behavior or achievement. Pioro (2006) in his
determined via adherence to problem specification by
work also found out that In general, the gender and
the groups.
academic major factors were not found to be
significant in computer programming. Stoilescu and
Experimental Set Up and Implementation: The
Egodawatte (2010) in their work on “Gender
artifacts used for this experiment were two practical
differences in the use of computers, programming,
programming questions (See the Appendix) that
and peer interactions in computer science
involved the use of Object Oriented Programming
classrooms” show that in computer programming,
(OOP) concepts, data structures as well as control
female students were not so involved in computing
structures. The teams individually worked
activities whereas male students were heavily
collaboratively to solve the problems right from the
involved.
analysis phase to implementation, using a single
Research Methodology computer system allotted to them.
In this work, we try to find out empirically if gender
The first programming exercise involved a payroll
difference will really affect the programming skill
program using files. The volunteered students were
acquisition by students using a collaborative learning
distributed into pair groups of same gender. Twenty-
approach as suggested by Mustafa (2013). He
two male and twelve female pair programmers
asserts that collaborative programming could be a participated in the exercise. The pair groups worked
chance to dissipate gender differences in attitudes independently on the problem with one member of
towards programming.
each group serving as the “Driver” and another
Subjects: The subjects used for this research were serving as the “Navigator”. Each pair group analyzed
the volunteered Students in 200 level of the the problem, designed the algorithm for the solution
Department of Computer Science, University of to the problem at hand, implemented the algorithm
Ibadan, Ibadan, Nigeria. It is appropriate to use with Java programming language and debugged the
students of Computer Science at 200 Level who are code together.
just learning how to program because the effect of
collaboration can easily be noticed without wasting The second programming task involved implementing
much resource. The Students got little or no prior a solution for a supermarket with Java Object
knowledge of Java programming. They were taken Oriented concepts and Database Connectivity with
through a 13-week teaching and practical classes in Java. The students were randomly organized into
Object Oriented Programming (OOP) concepts in collaborative teams of size four of the same gender.
CSC 232 (Structured Programming) Courses in the Twelve male and six female independent
first semester of 2014/2015 Session. collaborative teams of size four emerged. Again,
each group analyzed the problem, designed the
Experimental Design: Two experiments were algorithm for the solution to the problem at hand,
organized for the students. The first experiment was implemented the algorithm with Java programming
carried out by 68 students in pair programming language and debugged the code together. During
mode, with 22 male and 12 female pair groups. The the practical session, all the four members discussed

3
Am. J. Sci. Ind. Res., 2016, 7(1): 1-9

the problem together with one of them serving as the ANOVA statistics.
“Driver”.
RESULTS AND DISCUSSION
Each parallel teams was served with the practical
questions and a Report Form that was filled during Main Results: Table 1 shows the results of the
each stage of the programming task. The teams also outcomes (Mean Scores) of the groups obtained from
submitted their final codes as well as sample output the two experiments involving independent gender
screen shots if they were successful. The team groups. Figure 1 shows the chart for the outcomes.
identification number was filled on the Reporting form
Table 1: Mean Outcomes of the Independent Groups in
in order to identify each team. The two experiments the Two Experiments
(programming tasks) were run for a maximum of
three hours each in the Computer Laboratory of the Team Mean Outcomes
Department of Computer Science, University of Male 2-Person Team 7.0  0.17 SEM
Ibadan, Nigeria for two days; one per week.
Female 2-Person Team 7.1  0.22 SEM
The researcher served as the coordinator for the Male 4-Person Team 6.7  0.26 SEM
period of the experiments. Most importantly, the
researcher ensured that there was no interaction Female 4-Person Team 6.2  0.46 SEM
between different groups during the practical Note: SEM means Standard Error of the Mean
sessions such that ideas were not shared among the
independent groups. Scoring and analysis of the
codes were carried out using descriptive, t-test and

Fig. 1: Chart of the Mean Outcomes of the Independent Gender Groups


counterpart in the 4-person collaborative teams
From Table 1 and Figure 1, it could be inferred that (second experiment).
the female pair group performed marginally better
Further Statistical Analysis: The mean outcomes of
than the male counterpart. However, the 4-person
both males and females teams were tested using t-
male teams performed better than their female
test. Table 2 gives the result of the analysis.

4
Am. J. Sci. Ind. Res., 2016, 7(1): 1-9

Table 2: Independent Sample t-Tests


Data Hypothesis Tested p-Value Mean Values Decision
Mean Programming Ho: There is no significant
Outcomes of gender difference in the programming
pair groups outcomes of both genders in pair
programming 0.66 Male = 7.0 Ho accepted
Female = 7.1
p>0.05
Mean Programming Ho: There is no significant
Outcomes of gender 4- difference in the programming
Person groups outcomes of both genders in 4-
person collaborative 0.37 Male = 6.7 Ho accepted
programming teams
Female = 6.2
p>0.05

Table 2 shows that there is no significant difference insignificant at the end of the experiment. Both the
between male and female programming teams’ genders performed more or less at par in the two
outcomes (p>0.05) in both experiments. experiments conducted. This boils down also to say
that gender is not an issue at all in Computer
A One-Way Analysis of Variance (ANOVA) was programming.
carried out on all the results obtained from the two
The results suggest that some other factors like
experiments. Results of the analysis show that there
interest or fear may be responsible for low
were no significant differences among the treatment
representation of women in computing which is
groups (Males and females Pair versus 4-person
experienced today in students adopting programming
collaborative teams, p = 0.15).
as laudable computing profession. Attitude of women
DISCUSSION OF RESULTS toward computing is one among several factors that
might explain the low participation of women in
Observations in recent years show that women
computing. Programming is not alluring to females at
participation in the area of serious programming in
all especially in the developing parts of the world
the software industry is very low (Korkmaz and Altun,
such as Nigeria. This could be attributed to females'
2013 and U. S. Department of Labor, Bureau of
low confidence in Computer Programming.
Labor Statistics, 2012). This study was motivated by
Therefore, their attitudes towards programming are
this observation to probe empirically on whether
more negative than males as found by Korkmaz and
gender difference exists or not among student
Altun (2013).
programmers at the University of Ibadan, Nigeria.
Limitation of these Results / External Threat to
The hypotheses tested in this study reiterate the fact
Validity: The results obtained from this study are
that gender difference may not come into play in
limited to academic environment and therefore, not
computing. The means of their average effectiveness
generalizable for now. The experiment may need to
were very close to each other. The mean outcome is
be redone at industrial environments in order to verify
a totality of their performances in the analysis of the
the claims made in this study.
problem given, their accuracy and efficiency scores in
the programming task. This buttresses the fact that Conclusion: In this study, we have empirically
the effectiveness of both gender teams was not demonstrated that gender difference may not come
different. This result confirms the qualitative studies into play at all when it comes to computer
earlier conducted by Noreen (1985), West (2001), programming. Other factors like fear or attitudes may
Astrachan, (2004) and Pioro (2006) who showed that however be responsible for the low representation of
both genders can perform at par in computing. females in computing. In building the females’
confidence in computing, collaborative programming
In fairness to both genders, their mean outcomes
approach as was used in this study is highly
were marginally different and their difference quite

5
Am. J. Sci. Ind. Res., 2016, 7(1): 1-9

encouraged and recommended to the tutors in Korkmaz Ö. and Altun H.(2013). Engineering and CEIT
teaching programming to students, as it could be a student’s attitude towards learning computer
chance to dissipate gender differences in attitudes programming. International Journal of Social Science,
towards programming (Mustafa, 2013). 6(2):1169-1185.
Lenney E. (1977). Women’s self-confidence in achievement
ACKNOWLEDGEMENT: The researcher specially setting. Psychological Bulletin, 84, 1–13.
acknowledges the 200 Level students of the
Department of Computer Science, University of Lockheed Marlaine E.(1993). Women, Girls, and
Ibadan, Nigeria for their sincere and faithful Computers: A First Look at the Evidence Sex Roles
participation in this study. The study was carried out 29, 115-121.
as part of their regular teaching and practical periods Maryanne FISHER & Anthony COX(2006) Gender and
in the First Semester of 2014/2015 session. Programming Contests: Mitigating Exclusionary
Practices (Internet)
REFERENCES http://www.maryannefisher.com/wp-
Astrachan O. (2004). Non-competitive programming content/uploads/2011/01/Informatics_education.pdf.
contest problems as a basis for just-in-time teaching. Downloaded on 15th June, 2013.
In: Proc. ASEE/IEEE Frontiers in Education
Conference.Savannah, Georgia, pp. T3H/20–T3H/24. Mustafa Baer (2013). Attitude, Gender and Achievement in
Computer ProgrammingMiddle-East Journal of
Cashdan, E. (1999). Are men more competitive than Scientific Research 14 (2): 248-255, 2013 ISSN 1990-
women? British Journal of Social Psychology, 37, 213– 9233.
229. Noreen M. Webb (1985). The Role of Gender in Computer
Cassidy, S. & Eachus, P. (2002). Developing the computer Programming Learning Processes, Journal of
user self-efficacy (CUSE) scale: Investigating the Educational Computing Research, Volume 1, No. 4,
relationship between computer self-efficacy, gender pp. 441 - 458
and experience with computers. Journal of Educational Pioro B. T. (2006). Introductory computer programming:
Computing Research, 26(2), 133-153. gender, major, discrete mathematics, and calculus,
Cooper, J. (2006). The digital divide: The special case of Journal of Computing Sciences in Colleges, Volume 21
gender. Journal of Computer Assisted Learning, 22(5), Issue 5, May 2006, USA, pp. 123 – 129.
320-334. Severiens, S. E. & Ten Dam, G. T. M. (1994). Gender
Daly, M., and M. Wilson (2001). Risk-taking, intrasexual differences in learning styles: A narrative review and
competition, and homicide. Nebraska Symposium on quantitative meta-analysis. Higher Education, 27(4),
Motivation, 47, 1–36. 487-501.

Dawn Patitucci (2005). Gender and Programming Steele, C., and J. Aronson (1995). Stereotype threat and
Language Preferences of Computer Programming the intellectual test performance of African Americans.
students at Moraine Valley Community College. In Journal of Personality and Social Psychology, 69,
Master Degree (Project) submitted to Department of 797–811.
Occupational and Technical Studies Old Dominium Stoilescu, Dorian; Egodawatte, Gunawardena (2010).
University.digital.lib.odu.edu:8000/dspace/bitstream/.../ Gender differences in the use of computers,
326/1/patitucci_dawn.pdf Downloaded 20th July, 2013. programming, and peer interactions in computer
Eustáquio São José de Faria , Juan Manuel Adán-Coello, science classrooms, Computer Science Education,
Keiji Yamanaka (2006). Forming Groups for Volume 20, Number 4, pp. 283-300(18)
Collaborative Learning in Introductory Computer U.S. Department of Labor, Bureau of Labor Statistics
Programming Courses Based on Students’ (2004). Women in the Labor Force: A Databook
Programming Styles: An Empirical Study. 36th (Report 973), http://www.bls.gov/cps/wlf-tables11.pdf.
ASEE/IEEE Frontiers in Education Conference. Downloaded 20th August, 2012
http://fie2012.org/sites/fie2012.org/history/fie2006/pap West, Margaret Lynne Perkins (2001), "Gender
ers/1170.pdf Downloaded on 23rd May, 2013. underrepresentation in beginning computer
Helgeson, V. (2005). Psychology of Gender (2nd edition). programming courses", PhD Dissertations. Paper
Prentice-Hall, Upper Saddle River, New Jersey. 2646. http://aquila.usm.edu/theses_dissertations/2646,
downloaded in May 2013.

Potrebbero piacerti anche