int. j. math. educ. sci. technol., 2002 vol. 33, no. 6, 801±818 What e ect does the introduction of graphics calculators have on the performance of boys and girls in assessment in tertiary entrance calculus? PATRICIA A. FORSTER* Faculty of Community Services, Education and Social Sciences, Edith Cowan University, 2 Bradford Street, Mt Lawley, Western Australia 6050 e-mail: [email protected] and UTE MUELLER School of Engineering and Mathematics, Edith Cowan University, 100 Joondalup Drive, Joondalup, Western Australia 6027 e-mail: [email protected] (Received 4 June 2001 ) The paper reports an investigation of gender-related e ects in the Western Australian Calculus Tertiary Entrance Examinations for the period 1995±2000, three years before and three years after graphics calculators were introduced. Consideration is given to students’ total examination scores, scores on questions grouped by curriculum component and students’ actual use of graphics calculators on two questions from the year 2000 examination. Results show that over the six-year period the performance of girlsÐon the examination as a whole and in most curriculum componentsÐwas superior to that of boys at the lower end of the achievement scale, while boys recorded the best performance at the top end of the scale. The results are partially explained by the participation rate in calculus for girls being lower than that for boys. Where superior performance is recorded for girls it is frequently attributable to competence with analytic methods. Superior performance favouring boys typically occurred on questions where diagram played a role in the solution. 1. Introduction In 1998, graphics calculators became required equipment for Western Australian Tertiary Entrance Examinations (TEE) in mathematics and science subjects. Since that time we have investigated extensively the extent and nature of students’ use of the calculators in the Calculus examination [1±3]. We have also systematically explored and described the characteristics of examination questions and the nature of changes between questions set prior to and since the introduction of the graphics calculator [4]. In particular, we considered the opportunity for graphics calculator use, the role of a diagram in the solution process, whether or not a well-known algorithm was called upon and the curriculum component to which each question belonged. The ®ve main curriculum components * The author to whom correspondence should be addressed. International Journal of Mathematical Education in Science and Technology ISSN 0020±739X print/ISSN 1464±5211 online # 2002 Taylor & Francis Ltd http://www.tandf.co.uk/journals 802 P. A. Forster and U. Mueller in the Calculus TEE are Functions and Limits, Theory and Techniques of Calculus, Applications of Calculus, Vector Calculus, and Complex Numbers. The classi®cation of questions was a critical evaluation of our own practices as examiners, for one or both of us were on the examining panel for Calculus in the years 1996±2000. For the 1998 examination, we carried out a preliminary analysis of students’ total scores by gender and compared the results with examination outcomes for 1995±1997 [3]. There was no statistically signi®cant di erence between the mean total examination scores of females and males in any of the years. The 1998 examination contained one question that attracted widespread calculator use for generating a graph. A comparison of scores for it with scores for similar questions from 1995±1997 gave no strong indication that the introduction of the calculators had been discriminating in its e ect. For other questions in the 1998 paper, including on Function and Limits, Complex Numbers and Applications, work shown in scripts indicated that the minority of candidates (as low as 25% for one question) chose to use the tool where its use was an option. With the 1998 Theory and Techniques questions there was very limited scope for using the calculator, and no opportunity in the item on Vector Calculus. Thus, issues to be investigated with the introduction of graphics calculators to the Calculus TEE, in relation to gender, are opportunities to use the technology, the extent to which it is used by girls and by boys, the nature of its use, and the distribution of scores for girls and boys opting for graphics calculator approaches and for those who choose traditional analytic approaches. In this paper we explore these issues for the period 1995±2000. Speci®cally, we consider students’ total examination scores for the six years, scores as per the ®ve main curriculum components, and scores by gender and use of graphics calculators for two questions on Applications of Calculus from the year 2000 examination. 2. Background 2.1. The literature Graphics calculators appeared in 1985. The introduction to schools was gradual but accelerated upon their inclusion in public examinations. The calculators were allowed for the United States Advanced Placement Calculus in 1993 and the `A’ level examinations in the UK in 1994. In Australia, the calculators were introduced in examinations for the Victorian Certi®cate of Education (VCE) in 1997, in the Western Australian TEE in 1998, and are allowed in school-based assessment for university entry in Queensland. The introduction of the calculators to schools was accompanied by a research programme but perusal of journals indicates that relatively little attention has been paid to issues of gender. A repeated ®nding, however, is superior performance by both males and females in groups using graphics calculators, compared to groups without their use [5±7]. Other outcomes are as follows. Ruthven [5] recorded that boys scored better overall than girls in a group using the calculators (in a high-school setting) but girls’ performance was superior to that of boys on items requiring symbolic answers for given graphs. The better performance of girls was attributed to the availability of graphical checks. Checks potentially decrease anxiety, which is particularly relevant to girls for they exhibit less con®dence than boys under conditions of uncertainty (Fenema, cited in [5]). E ect of graphics calculators on performance of boys/girls in calculus 803 On items where graphs were supplied and interpretation was required, but no equation, Ruthven found there were no signi®cant di erences in performance between calculator and non-calculator groups, and no gender-related e ects within the calculator group. Boers and Jones [8], in the context of undergraduate Calculus where graphics calculator usage was encouraged in the course, recorded di erences in overall examination performance favouring females. Superior performance by females was due mainly to success in algebraic questions. In questions that required the construction of graphs, females only marginally outperformed males; a greater (although not statistically signi®cant) percentage of boys and girls chose the calculator option over pure algebraic, or mixed `algebraic and calculator’ approaches; and the percentage of boys who chose a pure calculator approach was greater than the percentage of girls. In particular, for an item that required the integration of algebraic answers (from early parts of a question) with the sketch of a graph, the percentage of males who were successful was greater than the percentage of females. Smith and Shotsberger [6], in the context of a college (undergraduate) algebra examination with graphics calculators included, also reported that the percentage of males using only a graphical approach in some questions was greater than the percentage of females. Moreover, a greater percentage of females than males used mixed graphical and algebraic approaches, partly for checking one method against the other. Dunham [9] found that students’ performance in a `graphics-calculator-enhanced college precalculus course’ improved signi®cantly for both sexes with respect to visual items. Furthermore, in a pretest males showed greater competence in visual items but this gap closed in the post-test (cited in [8]). Nimmons [7], using a comparative study design in a college algebra setting, found that on visual items both male and female students in graphics calculator groups scored signi®cantly higher than students in groups without the technology. Females in the calculator groups showed greater gains than males in visualization skills and in the level of understanding, which re¯ects Dunham’s ®ndings. However, Cassity’s [10] investigation of the relation of the variable `mathematical performance at the conceptual level’ to the variables spatial visualization, mathematical con®dence, utilization of the graphics calculator in the classroom, basic algebra ability and gender, reported no statistically signi®cant results relating to gender. Lawton’s [11] study showed an overall improvement of mathematics performance with the calculators but no signi®cant gender-speci®c di erences in performance. In addition to performance, student attitudes and con®dence have been subject to inquiry with regard to gender and again results are mixed. In brief, Dunham [9] recorded signi®cantly higher positive correlations between task-speci®c performance and con®dence for females than for males at the end of the graphics calculator enhanced course. In contrast, Lawton [11] found no signi®cant correlation between mathematical performance and mathematical con®dence for males and females. Almeqdadi [12] investigated student attitudes towards graphics calculators in a university calculus course and identi®ed signi®cant gender-related di erences favouring males, no signi®cant di erences between the low and high achievers, and no signi®cant interaction between gender, students’ achievement level and students’ attitudes towards the calculators. Merckling [13] recorded that use of the calculator in a high-school setting had a positive e ect on students’ 804 P. A. Forster and U. Mueller attitudes to learning mathematics, and that the result was particularly true for males. Dunham’s [9] interviews with students suggested that low-con®dence females and males relied on the calculator more than on algebraic approaches; and high-con®dence females were more likely to use an algebraic approach, while high-con®dence males were most likely to mix the methods. Boers and Jones [14] observed that low-achieving students relied more on the calculator than other students in examinations, and that this did not pay o . Nimmons [7] found that overall retention in graphics calculator groups and the retention of females in particular was greater than in non-calculator groups; and the calculator activities seemed to bring about lasting positive changes in female students’ dispositions to mathematics. Thus, in regard to gender-related e ects, it is established that the overall performance and understanding of both boys and girls can be enhanced by inclusion of the calculators. Other aspects which are particularly relevant to our inquiry include di erences in the extent to which algebraic and calculator methods are used, uptake of the calculator for checking, visualization skills and use of the technology according to ability level. To date, research ®ndings on these items are inconsistent and have commonly been based on small sample sizes or given little attention in the literature. 2.2. Data collection and analysis The Curriculum Council of Western Australia who administers the TEE supplied us with the marks for each candidate for each examination question and with summary statistics for the population, for the years 1995±2000. The summary statistics for 1995±1997 were based on total numbers of examination candidates, and for 1998±2000 on scores of students who attempted the questions. We retained the two approaches in our analysis by gender. In all cases, two-sample t-tests were carried out to test the statistical signi®cance of di erences in mean scores of boys and girls. Markers of the Calculus TEE in 2000, who were all experienced in teaching Calculus, were also a direct source of data. We asked them to record on a proforma, for a minimum of twenty students each, whether students had used graphics calculator or analytical methods in a question on rectilinear motion, and the part marks scored. Data were collected for 588 of the total 1886 scripts. Examination scripts from a school are spread among several bundles and bundles are allocated randomly to markers. Each script is marked by two markers, and di erences reconciled. The ®rst author, as a marker, collected similar data from 214 scripts for a question on exponential growth. Inferring students’ use of graphics calculators from their written answers is a subjective exercise so the results relating to the rectilinear motion and exponential growth questions need to be read with this in mind. We draw on our previous classi®cation of characteristics of questions for the 1996±1999 examinations [4]. The classi®cation was extended for the present inquiry to include the questions from the 1995 and 2000 examinations. The characteristics we refer to were adapted from Senk, Beckman and Thompson [15] and are: (1) curriculum component; E ect of graphics calculators on performance of boys/girls in calculus 805 (2) the role of a diagram in the solution (whether students were asked to interpret a diagram, make one, or a diagram could have assisted the solution); and (3) the role of graphics calculators (questions were calculator active, where there was a de®nite advantage in using the tool; neutral, where the question could reasonably be answered with or without the tool; or calculator inactive). We each carried out the classi®cations separately and negotiated our di erences of opinion to reach agreement. Further, when we refer to use of a graphics calculator in the analysis we are referring to use of capabilities over and above those of a scienti®c calculator. 2.3. Examination conditions The Calculus TEE is a state-wide examination. It comprises one three-hour paper. Total marks are 180. All questions are compulsory. Graphics calculators are required equipment, and non-symbolic calculators and the Hewlett Packard HP38G with limited symbolic capabilities are allowed. Students can take four A4 pages of notes into the examination (two sheets written on both sides), which was introduced to address the issue of di erent storage capacities of the calculators. 3. Results 3.1. Participation and total examination scores Table 1 shows the numbers of girl and boy candidates and the mean total raw examination scores for girls and boys. The ®gures in table 1 indicate that about twice as many boys as girls participate in the Calculus TEE and that during the period from 1995 to 2000 the number of male candidates decreased slightly (3% for the period), while the number of female candidates decreased substantially (22%). Curriculum changes for 1995±2000 which could explain the declining candidature, in particular the greater decrease for girls, are: (1) increased use of technology in the teaching and learning of mathematics; (2) the removal of the Calculus as a prerequisite for some Western Australian university courses, with universities o ering bridging courses instead; (3) the change of the university entry requirement from six subjects with satisfactory performance to four. That the retention of girls has been a ected more than the retention of boys is most likely related to students’ career aspirations. There are three TEE Number of girls Girls’ mean mark Boys’ mean mark Number of boys 1995 1996 1997 1998 1999 2000 706.00 103.45 104.04 1387.00 660.00 104.97 104.19 1264.00 618.00 120.29 119.49 1269.00 562.00 99.01 98.55 1320.00 577.00 106.08 104.62 1358.00 549.00 102.37 97.82 1337.00 Table 1. Candidature and total raw examination scores for 1995±2000 Calculus TEE. 806 P. A. Forster and U. Mueller 5th Percentile 8.0% 95th Percentile 6.0% 4.0% 2.0% 0.0% -2.0% -4.0% 1995 Figure 1. 1996 1997 1998 1999 2000 Percentage di erence in scores between female and male candidates at the 5th and 95th percentiles. mathematics subjects in Western Australia. We note that only the candidature in Calculus has decreased, whereas the number of students studying Discrete Mathematics has increased and the number of students in Applicable Mathematics has remained stable. Of the three subjects, Discrete Mathematics requires the least strong mathematics background and Calculus the strongest. The ®gures in table 1 show also that the mean total raw examination scores in Calculus for girls were greater than the mean scores for boys in the ®ve years from 1996 to 2000, with the result from 2000 being statistically signi®cant at the 98% level of signi®cance. The year 2000 examination is the only one for which the di erence in mean scores (4.6 or 2.6%) was substantial (see table 1). Throughout the time period, at the lower end of the achievement scale girls obtained higher scores than boys, while boys consistently outperformed girls at the upper end. The performance gap between boys and girls at the 5th and 95th percentiles is shown in ®gure 1. This gap is much wider at the bottom end (5th percentile) than at the top end, with percentage di erences in favour of girls at the bottom end twice as large as those favouring boys at the top end. The percentage di erence between scores of girls and boys at the 5th percentile ranges from 3.9 to 6.6% and at the 95th percentile between 0.6 and 2.8%. In addition, progressive shifts occurred in the percentiles above which performance of boys is superior to that of girls (30th percentile in 1995, 40th in 1996, 65th in 1997, 70th in 1998, 65th in 1999 and 90th in 2000). These outcomes are explained in part by the lower participation in TEE Calculus by girls, where one aspect is that fewer girls at the lower end of ability take the subject. We sought also to explain the signi®cantly better mean performance of girls in 2000 in terms of the extent to which use of graphics calculators was possible in examination questions. However, there was no clear link. For 1998±2000 when graphics calculators were allowed, the percentages of part-questions which were technology active or neutral (1998, 27%; 1999, 39%; and 2000, 36%). Next we investigated the call on visualization as measured by the percentages of part-questions in which there was a role for a diagram in the solution. The ®gures, E ect of graphics calculators on performance of boys/girls in calculus 807 (1995, 32%; 1996, 28%; 1997, 24%; 1998, 47%; 1999, 45%; and 2000, 48%), did not explain the di erences in total scores for boys and girls in the year 2000 examination. The increase in the role of diagrams between 1997 and 1998, upon the calculators ®rst being introduced, would have been when an e ect on scores could have been expected. 3.2. Results by curriculum component In this section we report mean scores on questions grouped by predominant curriculum component. We classi®ed each question on the 1995±2000 examinations as relating to one curriculum component. Where parts of a single question related to di erent curriculum components, the question was classi®ed as belonging to the component which attracted most marks. This decision arose because we had marks for whole questions but not part-questions. Cumulative distributions by gender for the curriculum components Theory and Techniques of Calculus, Complex Numbers, Function and Limits and Applications of Calculus for the years 1995 and 2000 are shown in ®gure 2. They are representative of trends in the distributions for the six years 1995±2000. In reporting the results, we start with Theory and Techniques of Calculus, the component having the strongest trend in regard to scores by gender, and ®nish with Vector Calculus, the component having the least consistent results by gender. In all six years, di erences between girls’ and boys’ scores in questions relating to Theory and Techniques favoured girls (see table 2). In ®ve of these years the di erences were statistically signi®cant. The greatest absolute di erences in mean marks was 1.50 in the year 2000 (0.8% on the examination). With Theory and Techniques questions, the use of graphics calculators and diagrams can assist the solution, for example, a graph can be used to help with establishing where a function is di erentiable. However, for questions in this component there is typically a minimal call or no call on the use of graphics calculators and diagrams (see table 2). Thus, Theory and Techniques questions have mainly required analytic methods, both before and since the introduction of graphics calculators, and girls have regularly done better on them, which is consistent with the ®ndings of Boers and Jones [8]. Plots of the cumulative distribution of percentage marks for boys and girls for all years 1995±2000 highlight the consistency and show that the trend of superior performance by girls persists at all levels of achievement (see ®gure 2 for the 1995 and 2000 distributions). Complex Numbers is the component with the next most consistent outcomes in relation to gender after Theory and Techniques. In ®ve of the six years from 1995± 1995 Maximum score Girls’ mean mark Boys’ mean mark ¤ 26.00 ¤ 17.98 ¤ 17.32 Part-questions with a role for Graphics calculators Ð Diagram 17% ¤ 95% con®dence level, ¤¤ 1996 ¤¤ 1997 ¤¤ 1998 ¤¤ 1999 2000 ¤¤ 35.00 ¤¤ 24.33 ¤¤ 23.07 36.00 ¤¤ 29.80 ¤¤ 28.83 28.00 ¤¤ 20.34 ¤¤ 18.91 29.00 19.14 18.73 25.00 ¤¤ 19.06 ¤¤ 17.56 Ð 11% Ð 8% 0% 0% 22% 11% 0% 0% 99% con®dence level Table 2. Results for the population on Theory and Techniques of Calculus questions. 808 P. A. Forster and U. Mueller Distribution TTC 1995 1 Distribution TTC 2000 1 0.8 0.8 0.6 0.6 0.4 0.4 F 0.2 F 0.2 M 0 M 0 0 0.2 0.4 0.6 0.8 1 Distribution CN 1995 1 0 0.8 0.6 0.6 0.4 0.4 F 0.2 0.4 0.6 0.8 F 0.2 M 0 1 Distribution CN 2000 1 0.8 0.2 M 0 0 0.2 0.4 0.6 0.8 1 Distribution F&L 1995 1 0 0.8 0.6 0.6 0.4 0.4 F 0.2 0.4 0.6 0.8 F 0.2 M 0 1 Distribution F&L 2000 1 0.8 0.2 M 0 0 0.2 0.4 0.6 0.8 1 Distribution AC 1995 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 Distribution AC 2000 1 0.8 0.6 0.4 M 0 0.2 0.4 0.6 0.8 F 0.2 F M 0 1 0 0.2 0.4 0.6 0.8 1 Figure 2. Cumulative distribution of % raw scores in Theory and Techniques of Calculus (TTC), Complex Numbers (CN), Functions and Limits (F & L) and Applications of Calculus (AC). E ect of graphics calculators on performance of boys/girls in calculus 1995 Maximum score Girls’ mean mark Boys’ mean mark 1996 1997 ¤ 1998 1999 ¤ 809 2000 ¤¤ 35.00 14.94 14.59 41.00 23.76 23.54 38.00 ¤ 22.45 ¤ 21.54 42.00 21.38 21.4 34.00 ¤ 16.42 ¤ 15.6 36.00 ¤¤ 17.79 ¤¤ 16.42 Part-questions with a role for Graphics calculators Ð Diagram 56% Ð 46% Ð 38% 27% 55% 30% 70% 22% 56% ¤ 95% con®dence level, ¤¤ 99% con®dence level Table 3. Results for the population on Complex Number questions. 2000, di erences favoured girls and for three of these years the di erences were statistically signi®cant (see table 3). The greatest absolute di erence in mean scores was 1.37 in 2000 (0.76% on the examination). The introduction of graphics calculators led to noticeable changes in the types of questions. As indicated in table 3, there has been a role for diagram in the majority of complex number questions and this has principally involved the construction and interpretation of Argand diagramsÐdrawn without the assistance of a calculator, which is re¯ected in the relatively low ®gures relating to calculator use. With the introduction of the graphics calculators, the number of questions related to arithmetic and to conversion from algebraic to polar representation (or the reverse) decreased substantially. This resulted in lower average scores in 1998±2000 compared to 1995±1997. Girls mean percentage scores out of the maximum possible scores for this component (see table 3) were 51%, 48% and 49% for 1998±2000 and 43%, 58% and 59% for 1995±1997. Boys’ scores showed the same trend. The aggregated ®gures in table 3, however, do not imply performance by gender is linked to use of diagrams or to opportunities for calculator use. However, a question-by-question analysis of the 29 questions, over the six years, indicated that girls performed signi®cantly better than boys on nine questions. They variously required interpretation of a diagram, the construction of a diagram, optional use of a diagram, or included no role for a diagram; of these, one in 1999 was technology neutral in that it allowed a graphical solution to an equation. Boys performed signi®cantly better than girls on three questions. One of these, in 1998, required students to make a diagram and was technology neutral (the diagram could have been produced on the calculator); in the other two (in 1998 and 1999), use of an Argand diagram, generated without the technology, could have signi®cantly assisted the solution. Thus, for Complex Numbers, use/non-use of a diagram is a stronger predictor of signi®cance than opportunities to use technology. Where signi®cant di erences in performance occurred, all di erences on questions that did not involve a diagram (i.e. required analytic methods or computation) favoured girls. All di erences that favoured boys were on questions that allowed use of a diagram. With regard to the distribution of scores, there was virtually no di erence in performance between boys and girls at the upper end of the performance scale in any of years 1995±2000 (see ®gure 2 for the 1995 and 2000 plots). The trend of superior performance by girls was evident at lower levels of achievement, and was pronounced and extended highest up the scale for 2000, which again is explained by the declining participation in calculus by girls. 810 P. A. Forster and U. Mueller 1995 Maximum score Girls’ mean mark Boys’ mean mark 1996 1997 1998 1999 ¤¤ 2000 ¤ 15.00 8.84 8.68 35.00 22.16 22.48 27.00 19.94 20.15 26.00 14.08 13.85 33.00 ¤¤ 19.02 ¤¤ 18.13 28.00 ¤ 19.33 ¤ 18.74 Part-questions with a role for Graphics calculators Ð Diagram 33% Ð 36% Ð 45% 67% 67% 50% 50% 36% 45% ¤ 95% con®dence level, Table 4. ¤¤ 99% con®dence level Results for the population on Functions and Limits questions. Di erences between girls’ and boys’ scores on Functions and Limits questions favoured girls in four of the six years (see table 4) and were signi®cant in 1999 and 2000, after the introduction of the calculators, but the greatest absolute di erence in mean scores was small (0.5% on the 1999 examination). However, a number of aspects of the questions from the curriculum area Functions and Limits are of interest. In 1998, the year in which the calculators were ®rst allowed, there was ample opportunity for calculator use and this was accompanied by an increased role for diagrams in the solution process (see table 4). For example, graphs of functions produced on the calculator could be used to establish or check limiting values, continuity and di erentiability. Checking has been suggested as being particularly bene®cial to girls [5, 6], but the results by gender in table 4 lend little support for this. The opportunity to use diagrams and graphics calculators progressively decreased in 1999 and 2000, and at the same time the requirement to use algebra increased. Hence, the signi®cant di erences in scores that favoured girls in those years might again indicate superior performance by girls on analytical methods. However, the cumulative distributions by gender for 1999 and 2000 show that girls at the lower and middle levels of the achievement scale bene®ted most from the change in emphasis (see ®gure 2 for the 2000 plot). The distributions for years up to and including 1998 indicate minimal di erence in achievement across the scale of performance (see ®gure 2 for the 1995 plot). Lower con®dence/lower achieving students have been observed to choose calculator approaches when they were an option, more frequently than algebraic methods [9, 14] and there were still opportunities in 1999 and 2000 where this was possible. Thus, it is not clear to what girls’ superior results in Functions and Limits can be attributed. A question-by-question analysis did not illuminate the issue. Actual use of the calculators by gender in this component needs to be ascertained. In Applications of Calculus, di erences in performance favoured boys in four out of the six years, were signi®cant in two of those years (1995 and 1998), and substantial to the extent of 1% on the 1995 examination (see table 5). However, girls have closed the achievement gap at virtually all levels: while the cumulative distribution of percentage scores for 1995±1998 indicated superior performance by boys for the major part of the scale, in 1999 the plots for boys and girls were virtually identical and in 2000 girls recorded marginally superior performance for nearly the whole scale (see ®gure 2 for the 1995 and 2000 plots). E ect of graphics calculators on performance of boys/girls in calculus 1995 Maximum score Girls’ mean mark Boys’ mean mark 89.00 ¤ 53.27 ¤ 55.10 Part-questions with a role for Graphics calculators Ð Diagram 23% ¤ ¤ 1996 1997 1998 ¤ 811 1999 2000 59.00 27.89 27.68 61.00 37.31 38.27 73.0 ¤ 36.7 ¤ 38.4 72.0 46.5 46.6 77.0 39.7 38.6 Ð 22% Ð 25% 30% 55% 45% 45% 48% 57% 95% con®dence level Table 5. Results for the population for Applications of Calculus questions. Since the introduction of the calculators, opportunities to use them in Applications and the role of diagrams have been relatively high, but the ®gures in table 5 do not suggest any link between students’ performance by gender and these question characteristics. So, we conducted a question-by-question analysis for the six years to explain the di erences in performance. The style of the question depends on the Application, so we grouped the questions by Application for the analysis. In questions on simple harmonic motion and rectilinear motion, boys’ mean performance was better than that of girls in ®ve of the six years from 1995 to 2000. All signi®cant di erences (recorded in two years for simple harmonic motion and in ®ve years for rectilinear motion) favoured boys and occurred on questions where there was a role for a diagram. In most instances, the opportunity or requirement to use a diagram occurred in several parts of the questions. Sometimes the diagram was provided and at other times it could be generated on a calculator. The requirement to interpret a diagram was more consistent with signi®cant di erences than was using the technology. The questions on rectilinear motion where signi®cant di erences occurred were relatively hard questions, as measured by students’ percentage scores. They called on high-level interpretation, in particular where velocity and acceleration were involved, or when a solution required interpretation of a given context [2]. However, that signi®cant di erences occurred on hard questions and that these favoured boys was not consistent for the other Application questions. For questions on volumes of solids of revolution, di erences in performance favoured girls in four years and were signi®cant in those years. The role diagrams played was minor. Graphics calculators could be used to evaluate the volume integrals after their derivation. The other applications (including area under a curve, optimization, related rates and incremental change, continuous growth and decay) did not show consistently superior performance by one gender group and did not have signi®cant di erences favouring one gender. There is typically only one question on Vector Calculus in the Calculus TEE. There was little opportunity in the questions for use of a diagram for the years 1995 to 1999 or for the use of a graphics calculator in 1998 and 1999, indicating the call was mainly on analytic methods. However, unlike for Theory and Techniques, these characteristics were not associated with consistently better performance by girls: in fact when di erences were signi®cant they favoured boys (see table 6). 812 P. A. Forster and U. Mueller 1995 Maximum score Girls’ mean mark Boys’ mean mark 15.00 8.42 8.35 Part-questions with a role for Graphics calculators Ð Diagram 17% ¤ 95% con®dence level, ¤¤ 1996 ¤¤ 1997 1998 1999 ¤¤ 2000 10.00 ¤¤ 6.84 ¤¤ 7.42 18.00 10.79 10.71 11.00 6.46 6.39 12.00 ¤¤ 5.43 ¤¤ 6.23 14.00 6.62 6.79 Ð 0% Ð 0% 0% 0% 25% 25% 50% 50% 99% con®dence level Table 6. Results for the population on Vector Calculus questions. 3.3. Analysis of the year 2000 question on rectilinear motion In the following two sections we report the extent and nature of students’ actual calculator usage as ascertained by perusal of students’ examination scripts instead of relying on ®gures relating to possible calculator usage. First, we discuss a question on rectilinear motion and then consider a question on continuous growth. Both questions are from the year 2000 Calculus TEE. The question on rectilinear motion was attempted by 541 of the 549 girl candidates and 1310 of 1337 boy candidates. The mean score for girls was 6.05 and for boys was 6.67 out of the possible 11 marks. The di erence in mean was statistically signi®cant at the 99% level. Data were collected for 588 examination scripts. A chi-square test of goodness of ®t of the distribution of marks within the sample showed the sample to be representative of the results for the entire population at the 90% con®dence level. Technology usage was inferred when students provided an answer without including any written solution steps. The question is as follows A hot air balloon begins a 60 minute ¯ight by rising upwards from the side of a hill. Its vertical velocity v (metres per minute) is given by: v ˆ t…t ¡ 35†…t ¡ 60†=1400 where t is the time from the start in minutes. (a) What is the maximum upward velocity? (b) While the balloon is ascending, when is its acceleration the greatest? (c) When does the balloon reach its maximum altitude and how far above its starting point is it at that time? (d) Does the balloon land above or below its original elevation? Explain your reasoning. Sample data are summarized in table 7. Data collected for part (d) are not included in the summary because the use of the calculators was di cult to discern. Part (a) could be solved by graphing the velocity function with a graphics calculator and using its in-built capabilities to `jump’ to the maximum turning point of the graph. An analytic approach required expansion and subsequent di erentiation of the expression or di erentiation using the product rule, and was more time consuming and potentially more error prone than the graphical approach. Since justi®cation of the answer was not required, the graphical solution would have been the more practical approach to take. E ect of graphics calculators on performance of boys/girls in calculus Females Answer only Calculus working Males Answer only Calculus working a 813 Part a Part b Part c Number Average studentsa markb Number Average students a markb Number Average students a markb ¤¤ 89 69 1.70 ¤¤ 1.17 262 160 1.74 ¤¤ 1.46 ¤¤ ¤¤ 36 113 1.08 ¤¤ 0.75 136 268 1.43 ¤¤ 0.97 ¤¤ ¤¤ 27 118 2.63 ¤¤ 2.27 107 276 3.15 ¤¤ 2.23 ¤¤ Sample size: 161 females, 427 males maximum 2, 18b maximum 2, 18c maximum 4 99% con®dence level b 18a ¤¤ Table 7. Graphics calculator usage by gender for the 2000 rectilinear motion question. Part (b) could also be solved by relying entirely on in-built calculator functions. After entering the velocity function into the graphics calculator, the interval [0, 35] over which the balloon was ascending could be determined visually and the derivative (acceleration) could be plotted without calculating the symbolic form. The global maximum acceleration was needed and the time at which it occurred …t ˆ 0† could read from the graph. Alternatively, the interval [0, 35] could be determined by inspection, from the velocity function; and the problem could be treated as a constrained optimization problem, requiring investigation of all interior critical points and endpoints. The graphical approach had the advantage of directly focusing attention on the endpoints. For parts (a) and (b), sample data indicated that boys who chose to use their calculators were the group who obtained the highest scores on average, followed by girls who used their calculators, then boys and, last, girls who used analytic approaches (see table 7). Part (c) built on part (b) as it required the use of the interval on which the velocity was positive. Once the integral of the velocity function was set up it could „ 35 be evaluated on the calculator … 0 v dt ˆ 216:927 m†. The analytic approach was more cumbersome and error prone. The order of achievement was the same as for parts (a) and (b) except that, with analytic approaches, the mean score for girls was greater than the mean score for boys (see table 7). On all parts (a) to (c) of the question a clear majority of each sex chose to use numerical integration in preference to evaluating the integral over the velocity function exactly and a greater percentage of boys than girls chose to use it. This pattern of choosing/not choosing to use the technology matches the pattern recorded by Boers and Jones [8]. For part (d), the solution involved observing that the area enclosed by the velocity function and the horizontal axis was greater for [0, 35] than for [35, 60]; or „ 60 evaluating 35 v dt (by calculator or „ 60by hand) and comparing the result with the answer for part (c); or evaluating 0 v dt and noting that it was positive. For an analytic solution the integral expression from part (c) could be used. In the sample, part (d) was answered by 130 girls (mean score 1.93, out of a maximum of 3) and by 342 boys (mean score 2.15). We note that other aspects of this problem were 814 P. A. Forster and U. Mueller Question 13a-d Girls Boys a Question 13e Number studentsa Average markb 459 1329 6.29 ¤¤ 5.94 ¤¤ Question 13f Number students a Average markb Number studentsa Average markb 492 1133 1.91 1.98 497 1152 0.83 ¤ 0.87 ¤ number in the cohort girls ˆ 549, boys ˆ 1337 maximum 10, 13e maximum 3, 13f maximum 1 99% con®dence level b 13a-d ¤¤ Table 8. Performance for the population on the 2000 question on continuous growth. the need to make sense of the context of a hot air balloon in mathematical terms, and the possibility of mixed calculator/analytic approachesÐthese were coded as `Calculus working’ (see table 7). 3.4. Analysis of the year 2000 question on continuous growth The year 2000 question on continuous growth was attempted by 549 girls (mean score 8.75) and 1329 boys (mean score 8.38). The di erences between boys’ and girls’ performance were statistically signi®cant at the 95% level and girls achieved greater mean scores than boys. The question is given below, and table 8 shows performance by gender by part-question. The part-mark data were recorded for this question, at our request, on the central marks collection forms for the examination and provided to us by the Western Australian Curriculum Council. The size of P…t† of a population of bacteria in a culture at time t minutes is modelled by the equation dP=dt ˆ P ¡ P2 =1000 …1† (a) For which values of P is the growth rate dP=dt zero? (b) For which value of P is the growth rate greatest? ¡ (c) Show by di erentiating that P ˆ 1000=…1 ‡ Ce t † satis®es equation (1), for any value of the constant C. (d) Find C, given that at time t ˆ 0 the size of the population is 100. (e) Sketch a graph of P as a function of t. (f) What is the limiting size of the population as t ! 1? Part (a), worth 1 mark, was solved easily by factorization. It could have been solved graphically, but less e ciently, on the calculator. Part (b), worth 2 marks, was easily solved by calculating the derivative by hand, equating it zero and solving the equation by hand to obtain the maximum. Part (c), worth 5 marks, was a high level question and required sophisticated algebra. Part (d), worth 2 marks, required relatively simple algebra. Thus, the aggregated marks for parts (a) to (d) were associated in large part with algebraic competence. Part (e) required the generation of a graph (see ®gure 3). The given function is not part of the TEE Calculus and its unfamiliarity to students would have forced E ect of graphics calculators on performance of boys/girls in calculus f(x) 1000 900 800 700 600 500 400 300 200 100 (500, 2·19) point of inflection 5 Figure 3. Girlsa Boysa 815 10 15 20 x ¡ The graph of P ˆ 1000=…1 ‡ 9e t †. Adequate range (e) Correct y intercept (3) Correct shape (e) Correct limit (f) 45 63 53 67 51 64 75 90 a Number in the sample girls ˆ 57, boys ˆ 157; number who attempted (e) girls ˆ 47, boys ˆ 128; number who attempted (f) girls ˆ 48, boys ˆ 131. Table 9. Sample data on % of students with correct answers to (e) and (f) when attempting them. calculator generation of the graph. The process involved prediction of the limit to in®nity in order to obtain the range for an adequate screen display, thus algebraic thinking was integral to the graphing (unless a calculator with an autoscale facility was used). Then, when drawing the graph, students needed to be aware of the horizontal asymptote and draw it in or at least ensure that their graph did not drift above f …x† ˆ 100. The graph did not need to be interpreted, except for part (f ), but the limit required in (f ) was ideally obtained before the graph was produced. The 214 scripts marked by the ®rst author showed little if any working for the graphing, which indicates that the graphs were usually obtained using a graphics calculator. The sample data are summarized in table 9. They indicate that girls had more problems than boys in constructing the graph and in establishing the limit. About one third of girls and boys who identi®ed the limit did not obtain an adequate range, indicating that they failed to make the connection between the two quantities. Boys achieved marginally better results on the graph for part (e), and statistically signi®cant but not substantially better results for the limit (see table 8). Thus, for this question, again girls were stronger with analytic methods and boys showed superior performance with the part that involved a diagram. The mathematics in some parts of this question was challenging, but the context of bacterial growth did not in itself result in complexity, unlike the motion of the hot air balloon in the previous question. 816 P. A. Forster and U. Mueller 4. Conclusions There are a number of variables relevant to boys’ and girls’ di erential performance in the Calculus TEE. The most prominent is the much smaller number of girls choosing to study Calculus at the upper secondary level. This can explain why overall examination performance for girls at the lower end of achievement of the cohort is better than that for boys, and why girls’ mean scores were consistently higher than boys’ mean scores over the ®ve years from 1996 to 2000. However, boys continue to outperform girls at the top end of the scale. Secondly, in questions that required analytic solution methods, for example in questions from Theory and Techniques and Complex Numbers, analysis indicates that girls regularly scored better than boys. Signi®cant di erences in performance favouring boys occurred only on questions where diagrams played a role in the solution, for example in questions on Complex Numbers and Applications. Finegrained analysis of students’ actual calculator usage in two questions from the 2000 Calculus TEE supports these generalizations. Furthermore, results on the Complex Number and Applications components indicated the variable `role of diagram’ is a stronger predictor of outcomes by gender than `opportunities to use graphics calculators’. However, girls performed signi®cantly better than boys on some questions where a diagram played a role; and in some instances, diagrammatic or graphical approaches are only feasible if the calculator is used. The di erences in performance over the six years 1995±2000 were substantial to the extent of 2.6% on the 2000 examination, the only year in which di erences in overall performance were statistically signi®cant. The greatest source of di erence in that year was the component Theory and Techniques (0.8% in the mean favouring girls) and this was greater than the e ects relating to other components in all six years, except in 1995 Applications contributed 1% which favoured boys. After Functions and Limits, Applications has allowed most opportunity for the use of graphics calculators. While data on actual calculator usage by gender are not available for Functions and Limits, sample data for the rectilinear motion application in 2000 indicated that the majority of boys and girls chose to use the technology when it was an option and scored better than students not choosing it; and a greater percentage of boys than girls chose it and scored better than the girls. There are important time and accuracy advantages arising from choosing appropriately to use the calculator in the context of the Calculus TEE. Complexity introduced by context was another aspect of the questions on rectilinear motion. In conclusion, our study revealed trends in overall performance and in some curriculum components in the Calculus, and ®ne-grained analysis highlighted the trends. Anomalies are that increased opportunity for visual methods upon introduction of the calculators has not been accompanied by higher mean total scores in the examination for boys; and that Vector Calculus questions have called mainly on analytic methods, yet have not shown a pattern of superior performance by girls. Further investigation into actual calculator usage by gender, at di erent levels of achievement, and taking into account the level of di culty of questions are directions for future research which might further explain the discrepancies. Because of the anomalies and the unequal sizes of the boys’ and girls’ groups, we are hesitant to make broad generalizations as a result of the study, nor do we claim our ®ndings are applicable across all mathematics subjects. However, if con®rmed in other contexts, the trends that we have identi®ed have important implications for the equitable setting of examinations which screen students’ entry E ect of graphics calculators on performance of boys/girls in calculus 817 into university courses, where entry for some courses is highly competitive. A mix of question types that overall are not likely to favour one sex over the other would be recommended. Implications for preparing students for such examinations where graphics calculators are included are that girls might need to be encouraged to embrace the use of the technology and the visual approaches that it allows, and competence with analytic methods might need to be emphasized to boys. Acknowledgements We sincerely thank the markers who contributed to our inquiry and the Curriculum Council of Western Australia for their support and permission to copy the two questions from the year 2000 Calculus TEE paper. Copyright for these questions belongs to the Curriculum Council. Comments in this paper are not to be taken to represent views of the Curriculum Council. References [1] Mueller, U., and Forster, P. A., 1999, Proceedings of the Asian Technology Conference in Mathematics (ATCM) 99, edited by W.-C. Yang, D. Wang, S.-C. Chu and G. Fitz-Gerald (Blacksburg, VA: ATCM) pp. 86±95. Available: http://www.runet. edu/¹atcm/atcm98.html [2] Mueller, U., and Forster, P. A., 2000, Proceedings of the Asian Technology Conference in Mathematics (ATCM) 2000, edited by W.-C. Yang, D. Wang, S.-C. Chu and G. Fitz-Gerald (Blacksburg, VA: ACTM) pp. 394±405. [3] Forster, P. A., and Mueller, U., 2001, Int. J. Math. Educ. Sci. Technol., 32, 37±52. [4] Forster, P. A., and Mueller, U., 2000, Graphics calculator usage in the West Australian tertiary entrance examination of calculus. Paper presented at the annual conference of the American Educational Research Association (New Orleans, LA) ERIC ED 443423. [5] Ruthven, K., 1990, Educ. Studies Math., 21, 431±450. [6] Smith, K. B., and Shotsberger, P. G., 1997, School Sci. Math., 97, 368±376. [7] Nimmons, L. A., 1997, Spatial ability and dispositions toward mathematics in college algebra: Gender-related di erences (graphing calculators). [CD-ROM]. Abstract from: Proquest File: Dissertations Abstracts Item: AAT 9804393. [8] Boers, A. M., and Jones, P. L., 1993, Contexts in mathematics education, Proceedings of the 16th Annual Conference of the Mathematics Education Group of Australasia, edited by B. Atweh, C. Kanes, M. Carss and G. Booker (Brisbane: Queensland University of Technology) pp. 123±128. [9] Dunham, P., 1991, Mathematical con®dence and performance in technology-enhanced precalculus: Gender-related di erences. Doctoral Dissertation, The Ohio State University, 1990). Dissertations Abstracts International, 51/10, 3353. [10] Cassity, C. L., 1997, The relation of gender, spatial visualization, mathematical con®dence, and classroom graphing calculator utilization to conceptual mathematical performance: Learning with technology. [CD-ROM]. Abstract from: Proquest File: Dissertations Abstracts Item: AAT 9805262. [11] Lawton, T. C., 1999, E ects of graphing calculators on students’ con®dence and performance in college algebra. [CD-ROM]. Abstract from: Proquest File: Dissertations Abstracts Item: AAT 9939118. [12] Almeqdadi, F. A., 1997, Graphics calculators in calculus: An analysis of students’ and teachers’ attitudes. [CD-ROM]. Abstract from: Proquest File: Dissertations Abstracts Item: AAT 9732653. [13] Merckling, W. J., 1999, Relationship(s) between the perceptual preferences of secondary school students and their achievement in functions using a graphing calculator. [CD-ROM]. Abstract from: Proquest File: Dissertations Abstracts Item: AAT 9918684. 818 P. A. Forster and U. Mueller [14] Boers, A. M., and Jones, P. L., 1992, The graphics calculator in tertiary mathematics. Paper presented at the 15th Annual Conference of the Mathematics Education Research Group of Australasia, Sydney. [15] Senk, S. L., Beckmann, C. E., and Thompson, D. R., 1997, J. Res. Math. Educ., 28, 187±215.
© Copyright 2026 Paperzz