Journal of Mathematics Education

December 2008, Vol. 1, No. 1, pp. 132-152 © Education for All

Using Performance Assessment in Secondary

School Mathematics: An Empirical Study
in a Singapore Classroom
Lianghuo Fan
Nanyang Technological University, Singapore

Yan Zhu
University of Hong Kong, Hong Kong

This article reports an exploratory study on using performance assessment in

mathematics instruction in a high-performing secondary school in Singapore. An
intact mathematics class participated in the study, and received chapter-based
performance tasks as intervention during regular mathematics lessons for about one
and a half school years. The performance tasks used included authentic and/or open-
ended tasks. The students’ academic achievements and attitudes in mathematics were
compared with a comparison class that did not receive the intervention. Both
quantitative and qualitative data were collected, mainly through questionnaire
surveys, performance task tests, conventional school exams, and interviews with
students and teachers. The results suggest that the students receiving the intervention
performed significantly better than their counterparts in solving conventional exam
problems, and in general they also showed more positive changes in attitudes
towards mathematics and mathematics learning. The students from the experimental
class also expressed positive views about the benefits of using performance tasks in
promoting their ability in higher order thinking, though no statistically significant
difference was detected between the two classes of students in solving unconventional
tasks before and after intervention. Overall, the results appear to support teachers’
using contextualised problems in real life situations and open-ended investigations in
students’ learning of mathematics.

Key words：performance assessment, alternative assessment, mathematics teaching

and learning, authentic questions, open-ended questions, Singapore mathematics
education.
Introduction
Over the last decades, the importance of assessment in education, and particularly in
mathematics education, has received much attention from educational policy makers,
mathematics education researchers, and mathematics teachers, among others, in many
countries (e.g., see Fan, 2006; Morgan, 2000; National Council of Teachers of
Lianghuo Fan & Yan Zhu 133

Mathematics [NCTM], 1995; Niss, 1993; Ruthven, 1994; Singapore Ministry of

Education, 2004a, 2004b).
In particular, relatively new assessment (or the so-called alternative assessment)
concepts and strategies have been widely advocated by educational reformers and
increasingly used by practitioners in classroom practices (e.g., see; Brookhart, Andolina,
Zuza, & Furman, 2004; Clarke, 1997; Hargreaves, Earl, & Schmidt, 2002; Kulm, 1994).
Along with new development, assessment is more viewed as an integrated part of the
teaching and learning process. As Pegg and Panizzon (2007/2008) noted, “The emphasis
on embedding assessment into the teaching and learning process is identifiable globally”.
A relatively large research project focusing on integrating new assessment strategies
into mathematics teaching and learning was recently conducted in Singapore schools. The
new assessment strategies under investigation in the project include those using
communication tasks, performance tasks, project tasks, and student self-assessment tasks.
Each new assessment strategy was implemented in two primary and two secondary
schools, in total 16 schools, for about one and half school years. The effects of using the
new assessment strategies on students‟ learning of mathematics were studied in both
affective and cognitive domains.
As part of the large project, the study presented herein focuses on the effects of using
performance tasks, including mainly authentic and open-ended ones, in mathematics
instruction on students‟ learning of mathematics in one participating secondary school.
The study was intended to address the following three research questions:

(a) What are the effects of using performance tasks on students‟ mathematics
performance as measured in solving unconventional problems?
(b) What are the effects of using performance tasks on students‟ mathematics
performance as measured in solving conventional problems?
(c) What are the effects of using performance tasks on students‟ attitudes
toward mathematics and mathematics learning?
It is hoped that the study can provide research-based evidence on the potential
influences of using performance tasks on students‟ mathematics learning so as to help
school teachers better align assessment practice with the desired educational goals and
hence improve the quality of teaching and learning.
Conceptual Framework and Perspectives
Performance assessment has been widely believed to have more pedagogical value
and it can reflect students‟ achievement more accurately than traditional multiple-choice
tests (e.g., see Kane, Khattri, Reeve, & Adamson, 1997). According to NCTM,
assessment is “the process of gathering evidence about a student‟s knowledge of, ability
to use, and disposition toward mathematics and of making inferences from that evidence
for a variety of purposes” (1995, p. 3). Following this definition, „performance
assessment‟, or sometimes called „performance-based assessment‟, is an assessment
strategy by which the evidence about students‟ learning is gathered through students‟
work on performance tasks. Nevertheless, there is no consensus on its definition, or more
specifically, what performance tasks are. According to Buechler (1992), the emergence of
134 Using Performance Assessment in Secondary School Mathematics

the performance assessment movement was due to the fairly widespread dissatisfaction
with high-stake multiple-choice tests. Gripps (1994) claimed that, in the United States,
performance assessment was often regarded as any type of evaluation which was not
multiple-choice or standardised testing. However, such a definition is rather broad and it
covers almost all types of alternative assessment (e.g., project work).
In the Third International Mathematics and Science Study (TIMSS), performance
assessment was included as one important component in international comparison, and
was referred to as integrated and practical tasks targeting students‟ content and
procedural knowledge as well as their ability in using knowledge for reasoning and
problem solving (Harmon et al., 1997). The Wisconsin Education Association Council
(1996), at the root of the meaning of the word „performance‟, set their definition as the
one requiring students to demonstrate skills and competencies by performing or
producing something. The central idea in Stenmark‟s (1991) definition about
performance assessment is to assess what students actually know and can do. It is clear
that while researchers tend to differentiate performance assessment from traditional
assessment, they have different concerns and focuses. In other words, a variety of aspects
have been connected with the term performance assessment.
To be more applicable to the Singapore school education context, performance tasks
used in this study were mathematical tasks with the following two distinguishing
characteristics: (a) authentic in context, and (b) open-ended in approaches and answers.
Naturally, when students solved these tasks, they were required to demonstrate how they
performed, in other words, their thinking or working process.
The authenticity of a problem, according to the NCTM, is the degree to which tasks
are faithful, comprehensive, and complex, which can be found in important, real-life
performances of adults that are non-routine yet meaningful and engaging for students
(NCTM, 1995). It is believed that tasks with this feature could engage students in
applying knowledge and skills they have learned in the classroom to real-world
challenges, and help them appreciate the usefulness of mathematics.
The open-endedness of a problem includes two aspects: (a) multiple venues of
access or ways of solutions, and (b) multiple acceptable answers to the problem. It is
believed that solving open-ended problems is more challenging than close-ended ones
that students usually encounter in their school work, and normally requires higher-order
thinking. In fact, these two aspects are to a large degree lacking in traditional assessment
tasks, which consequently often have received criticism over the last decades (e.g., see
Howe & Jone, 1998; Wu, 1994).
As a result, all the performance tasks used in this study are contextualised, to a
different degree, in real-world scenarios; they can be approached in various ways and
ended with different answers (not just in different representation forms). Below is a
sample performance task, authentic in the Singapore social context. More examples of
performance tasks designed and used in the study can be found in Fan (2008).

Use the information listed in Distribution by Type of Dwelling as shown below to (1)
construct a pie chart of the distribution of dwelling types in 2000; (2) predict and construct a pie
chart of the distribution in 2010 and defend your answer.
Lianghuo Fan & Yan Zhu 135

Distribution by Type of Dwelling

% 1980 1990 2000

HDB Flats1 68.5 84.6 88.0
Condominiums & Private 2.3 4.1 6.0
Flats
Private Houses 8.5 7.0 5.1
2
Others 20.7 4.3 0.9
1 2
Government-built apartments. Includes shophouses, attap/zinc-roofed houses,
other public flats and others.

This task introduces students to the knowledge about various housing types in
Singapore and the changes in the distribution over the last three decades. The task
requires students to predict the possible distribution in ten years later based on the
information given. To have a reasonable prediction, students need to use their knowledge
in constructing pie charts and apply their daily life experience to figure out a possible
changing trend in the housing types and relevant factors that may have influences on the
changes. Among the four housing types, the changes for condominiums & private flats
and private houses appear stable over the last thirty years, while the changes for the other
two types appear more rapid over the same period, which may in particular lead some
students to think of the possibility of the disappearance of the housing type others in the
next ten-year time frame. Given that there are no universal and standard answers to the
question, three performance criteria were given for the teacher to assess students‟
answers: (a) the trend of change must be consistent with the previous two time intervals,
(b) all the percentages in the pie chart should sum up to 100%, and (c) the arguments for
the predictions must be reasonable and practical.
Like in many other countries, developing students‟ ability in solving authentic and
open-ended problems in the teaching and learning of mathematics has received increasing
attention in Singapore. As a matter of fact, the Singapore mathematics syllabus
emphasises the importance of students‟ applying mathematics in solving real-life
problems and being engaged in open-ended investigations in mathematics instruction
(Ministry of Education, 2002). However, an analysis of two widely-used Singapore
secondary mathematics textbooks revealed that fewer than 2% of textbook tasks were
authentic and about 2% were open-ended (Fan & Zhu, 2000, 2007). Moreover, according
to Schoenfeld (1992), the beliefs that mathematics learning has little or no relation to the
real world and that any mathematics task has one and only one answer are commonly
held among students. In this connection, the study has both theoretical and practical
significance.
Research Methods Below we shall provide information about the participants,
including the school, students and their mathematics teachers, the instruments for data
collection, as well as the procedures of data collection and data analysis in this study.
Participants
As mentioned earlier, this study focused on one participating secondary school,
which is identified as a high-performing school, as it was randomly selected from the 50
136 Using Performance Assessment in Secondary School Mathematics

best performing secondary schools according to year 1999 to year 2002 GCE “O” Level
Examination results released by the Singapore Ministry of Education.
Thirty-eight Secondary One (Grade 7) students from one intact class of high ability
in the high-performing school were selected in this study to receive chapter-based
interventions on performance tasks during regular mathematics lessons for about three
school semesters starting from early 2004. A parallel intact class of 40 students was
chosen as a comparison group. Table 1 provides the profiles of the students and their
mathematics teachers from the two classes. No significant difference was found between
the two classes in terms of students‟ Primary School Leaving Examination (PSLE)
overall scores (t [76] = 0.81, p = 0.42) and mathematics grades (U [38, 40] = 747.00, p =
0.84).
As we can see from Table 1, the two classes were taught by two different teachers
with basically equivalent professional background since the beginning of the study in
2004. However, it should be noted that due to some unforeseen reasons, starting from
January 2005 the teacher teaching the experimental class had to take over the comparison
class as well. Given the change, the teacher was advised not to use the intervention tasks
in the comparison class so as to keep the teaching practices unchanged in both the classes
in terms of interventions.

Table 1
A Profile of Students and Mathematics Teachers in the Experimental and
Comparison Classes

Experimental Class Comparison Class1

No. of Students
Boys 24 25
Girls 14 15
Mathematics teachers
Gender Male Male
Length of teaching e
3 months 9 months
xperience
Qualification M.Eng, PGDE2 MSc

Note. 1 In year 2005, the teacher teaching the experimental class also took over
the comparison class; 2 PGDE stands for Postgraduate Diploma in Education.

The teacher teaching the experimental class received training and guidance on how
to use performance tasks in teaching before and during the intervention from the
researchers. Intervention tasks were carried out in the experimental class throughout the
intervention period of about one and half years, while the comparison class was taught as
usual during this period of time.
Lianghuo Fan & Yan Zhu 137

Instruments and Data Collection

Four main instruments were designed for data collection in this study: questionnaires,
performance task tests, intervention task worksheets, and interviews with teacher and
students.
Questionnaires
Two questionnaires, one for the pre-intervention survey and the other for the post-
intervention survey, were designed to find out students‟ attitude toward mathematics and
mathematics learning as well as their experience with performance tasks (see Appendix).
Both questionnaires consist of two parts, Part I and Part II. For a comparison purpose,
Part I, containing 22 items, is the same for both questionnaires. Those items, focusing on
students‟ perceptions about mathematics and their learning of mathematics, can be
categorised into four subgroups based on the areas they measured: general view towards
mathematics and mathematics learning, anxiety level in mathematics learning,
perceptions of own performance in mathematics, and beliefs about the usefulness of
mathematics. A nine-point Likert-type scale ranging from “disagree totally” to “agree
totally” is employed for each item.
Part II in the pre-intervention questionnaire was intended to measure students‟
experience with various alternative assessment tasks. Out of the 6 items in total, 3 items
are particularly about their experience in solving performance tasks in their mathematics
learning before intervention. Each item is in a six-point Likert-type scale on frequency.
Part II in the post-intervention questionnaire focused on students‟ feeling about using
performance tasks in the learning of mathematics. There are totally 16 items, each using
the nine-point Likert-type scale.
A pilot test of the pre-intervention questionnaire was conducted in January 2004,
involving 56 Secondary One students from two other schools. The initial version of the
questionnaire was modified based on the results of the pilot tests. The final questionnaire
has a reliability of 0.85.
The pre-intervention questionnaire survey was conducted in February 2004 for both
the experimental and comparison classes with a response rate being 100% and the post-
intervention questionnaire survey was in May 2005 with a response rate being 81.6%.
Performance task tests
Similar to the questionnaires, two sets of parallel performance task tests, a pre-test
and a post-test, were designed. The pre-test was intended for the researchers to have a
better understanding about students‟ entry levels in mathematics problem solving, while
the post-test was conducted for the researchers to measure the possible changes of
students‟ ability in problem solving after three school semesters with or without being
exposed to performance tasks in mathematics learning for experimental and comparison
classes, respectively. Both tests contain three open-ended tasks, with one being also
authentic.
A pilot study of the pre-test was conducted in February 2004 with 35 Secondary One
students from one school. Based on the students‟ feedback, necessary modifications on
test tasks were made. The modified tasks were again piloted by a group of 36 Secondary
138 Using Performance Assessment in Secondary School Mathematics

One students from another school in March 2004. As a result, while about 60% of the
students felt that the tasks were challenging to them, all the students had no difficulty in
understanding the tasks. Some minor modifications were further made in finalising the
pre-test items.
By the way, students‟ performance in solving conventional tasks was measured
using normal school exam scores. With the participating teachers‟ assistance, we were
able to collect all the 78 students‟ PSLE overall scores and mathematics grades (Exam A),
year 2004 school mid-year mathematics exam scores (Exam B), year 2004 school final-
year mathematics exam scores (Exam C), as well as year 2005 school first mathematics
common test scores (Exam D). In addition, the pre-test, which focused on students‟
ability in solving performance tasks, was conducted in March 2004 with a response rate
being 97.4% and the post-test, parallel to the pre-test as mentioned earlier, was in May
2005 with a response rate being 82.4%.
Intervention task worksheets
For the purpose of integrating performance tasks into classroom teaching and
learning, the design of the intervention tasks strictly followed the stipulated school
scheme of work. Moreover, all the intervention tasks meet both the criteria as described
earlier: authentic as well as open-ended. For each chapter covered in the scheme of work,
one to two performance task worksheets were first crafted by the researchers and then
finalised jointly by the researchers and the participating teacher to better match the
students‟ background and the teacher‟s teaching plan.
In total, the teacher of the experimental class managed to carry out a total of 12
interventions during the three school semesters. Students‟ work on each intervention task
was collected by the classroom teacher and then handed to the researchers for evaluation.
After grading, a copy of students‟ work with researchers‟ comments was returned back to
individual students for their information and possible revision.
The researchers observed most interventions to monitor how the performance tasks
were carried out in the classroom. During those classroom observations, the interventions
were recorded with field notes, or audio/video taping. The observations were also useful
for the researchers to improve the design of future performance tasks.
Interviews with teacher and students
The interviews with teacher and students were conducted in late May 2005, after all
the surveys and tests mentioned above were completed. The purpose of the interview was
mainly for getting information about the participants‟ experience and understanding
regarding use of performance tasks as well as their opinions or suggestions on the use of
the new strategy in teaching and learning. While the interview questions for both teacher
and students are similar, understandably those for teacher are more from a perspective of
teaching and those for students are more from a perspective of learning.
A total of three interview sessions were carried out. The mathematics teacher from
the experimental class received an individual interview, while six students (two high
performing, two average performing, and two low performing) were recommended by the
teacher to attend the student interviews, in which these students were grouped in three for
Lianghuo Fan & Yan Zhu 139

each session. The two student sessions lasted about 30 minutes each and the one with the
mathematics teacher was about 60 minutes. All the interviews were recorded by audio
taping and filed notes.
Data Process and Analysis
The data from the two questionnaires were analysed using quantitative methods.
Descriptive statistics (e.g., frequency and percentage) was applied to describe students‟
overall perceptions about mathematics and mathematics learning. Mann-Whitney U tests
were used to examine the possible differences between the two classes of students in each
survey for the researchers to detect the impact of using performance tasks on the
experimental students‟ attitudes.
Students‟ work in the two performance task tests was graded based on task-specific
rubrics by two independent researchers. The inter-rater reliability was calculated by the
Intraclass Correlation Coefficient (ICC) on absolute agreement. As a result, the reliability
on three performance criteria (i.e., Approaches, Solutions, and Representation) over the
three tasks for the two tests ranged from 0.98 to 1.00, with an average being 0.99. Similar
to the analysis for the questionnaire data, the rubric-based grades from the performance
task tests were analysed by descriptive statistics to investigate students‟ overall
performance at class levels before and after the intervention period. Mann-Whitney U
tests were employed to identify possible differences between the experimental and
comparison classes in each test. Wilcoxon Signed-Ranks tests were used to detect the
change of students‟ grades from the pre- to post-tests. Moreover, possible differences on
the changes between the two classes were examined by Mann-Whitney U tests to identify
the potential relationship to the intervention program.
Students‟ PSLE overall scores and mathematics grades in the experimental and
comparison classes were compared by t-tests and Mann-Whitney U tests respectively to
measure the equivalence of students‟ academic background in the two classes. The
following three normal school exam scores were analysed by 2 × 2 ANOVA with time
(Exam B vs. Exam D; Exam B vs. Exam C; Exam C vs. Exam D) as a within-subject
factor and treatment (experimental vs. comparison) as a between-subjects factor to
investigate potential effects of using performance tasks on students from the experimental
class.
The interview data collected in an audio format were transcribed. Using qualitative
methods, the data allow researchers to analyse the teacher and students‟ views about the
new type of assessment strategy, which is not easy to be gained by questionnaire surveys
or achievement tests. Moreover, the evidence from the interview is helpful for the
researchers to triangulate what has been revealed in the above quantitative data so as to
strengthen the findings of the study.
Limitations of the Study
Like many other intervention-based studies in educational research, understandably
there were also difficulties and hence limitations in this study, given that it was
conducted in authentic classroom settings.
First, to investigate the effects of using performance tasks on students‟ learning of
140 Using Performance Assessment in Secondary School Mathematics

mathematics, this study involved one experimental class and one parallel
comparison class. Ideally, unlike the experimental class, the comparison class should not
be exposed to performance tasks during the intervention period of about three school
semesters. However, as mentioned earlier, for the last school semester, the mathematics
teacher of the experimental class had to take over the comparison class due to unforeseen
reasons, which was beyond the control of the researchers and the teacher. Although
having worked with the researchers, the teacher clearly understood and agreed that he
should not introduce the ideas of performance assessment (let alone intervention tasks) to
his teaching in the comparison class, the experience of the teacher working with the
experimental class could still, though likely unintentionally, influence his teaching in the
comparison class one way or another, which can, to an extent, affect the results of the
study.
Second, while this study introduced performance tasks to the teaching and learning
of mathematics in the experimental class, at the school level those students were still
assessed based on the traditional assessment practice for their official school performance
grading and reporting. In other words, the new assessment strategy was only introduced
at the classroom level, but not at the school level, which could affect students‟ motivation
in their working on the new assessment tasks, and therefore have negative influences on
the results of the study (see more discussions in the next section).
Third, according to the research design, the experimental students should be exposed
to the performance tasks in a systematic and scheduled way. However, due to some
unexpected school activities, it was often very difficult for the teacher to do so in
delivering the tasks to the students. In particular, in the first semester, the class only
managed to carry out one intervention task, nevertheless with continuous efforts of the
researchers, the teacher, and the school administrators, the situation was significantly
improved in the second and third semesters.
Given those limitations, we wish to remind the readers that the conclusions of study
should be taken with some caution. More generally, we wish to emphasise that the study
should be viewed as an exploratory one, which was also our intention when we designed
the study.
Results and Discussions
The main findings of the study were reported below, based on the three research
questions mentioned earlier.
Effects of using performance tasks on students’ mathematics performance in
working on unconventional problems
As said earlier, the pre- and post- performance task tests were targeted at
students‟ mathematics performance in solving unconventional problems. Similar to the
intervention tasks, all the tasks in the pre- and post- tests are open-ended in approaches
and answers, and one task in each test is contextualised in a real-life scenario. Moreover,
the tasks in the post-test were designed to be parallel (equivalent) to those in the pre-test,
hence the researchers can better measure the difference in students‟ performance and how
Lianghuo Fan & Yan Zhu 141

their experience with the new assessment tasks during the intervention affect their
performance in the post-test.
In terms of the overall scores, the Wilcoxon Signed Ranks tests showed that the
students from both the experimental and comparison classes made significant
improvement from the pre- to post-test (Experimental: Z = 3.60, p < 0.001, r = 0.621;
Comparison: Z = 3.88, p < 0.001, r = 0.75). No significant difference between the two
classes was detected in either test.
As all the tasks are open-ended in nature, it appears more meaningful to further
examine students‟ performance in terms of the effective strategies they employed, the
number of answers they obtained, and the solutions they represented. In fact, these
aspects are the three performance rubrics designed for evaluating students‟ performance
in the tests. A brief description of the three performance rubrics is listed in Table 2.

Table 2
A Brief Description of General Rubrics by Approaches, Solutions, and
Representation
Level 0 Level 1
Approaches  No attempt or  Strategy is ineffective and
(decision/strategy about  No evidence of could not lead to any
Approaching tasks)  a strategy correct answer
Solutions  No correct answer  Only ne orrect answer
(no. of answers obtained) obtained  obtained
Representation  No attempt or  Working is not clear and
(documentation of problem  Working is irrelevant hard to read
solving procedures)

Level 2 Level 3 Level 4

 Strategy could lead to correct  Strategy shows partial  Strategy is ffective
 answer but not systematic systematic pattern  That would lead to
 (e.g., guess and check)   acomplete set of Answers
 More than one correct  At least 50% of the full  A complete set of
 answer obtained answers obtained  answers obtained
 Working is not organised  Working is organised  Working is well
 so that the approach is not  and approach ispartially  organised and approach
 observable observable  is fully observable

It is believed that the analysis on the sub-domains can provide more in-depth
information on how students approach and solve such challenging unconventional
mathematics problems, especially those from the experimental class.
Regarding the approaches employed by the students, the data revealed that in most
cases, the students were able to use more systematic/effective methods in the post- than
pre-test. That is, more students received a mean score over 2 on this performance scale in
the post- than pre-test (Experimental: 79.4% vs. 42.1%; Comparison: 82.8% vs. 23.7%).
The improvements in both the classes reached a significant level (Experimental: Z = 3.93,

1
Effect size r is calculated when significant difference is detected. According to Cohen (1992, 1988), an r
value over .5 is considered to be „large‟, around .1 to be „weak‟, and around .3 to be „medium‟.
142 Using Performance Assessment in Secondary School Mathematics

p < 0.001, r = 0.67; Comparison: Z = 4.12, p < 0.001, r = 0.79), but no significant
difference was found between the classes in terms of the improvements.
As indicated earlier, all the tasks in the tests contain more than one correct answer,
as listed below:

Pre-Test Post-Test
Task 1 7 10
Task 2 56 25
Task 3 2 2

Since the last task in each test only had two answers, a task-specific rubric on
solutions was set for the two tasks, shown as follows:
Task 3 Level 0 Level 1 Level 2 Level 3 Level 4
Solutions ·No ·Only partial correct ·One ·Two ·Two
(no. of correct answer obtained, i.e., complete correct complete
answers answer getting correct central answer with central answers
obtained) obtained number(s) different numbers with
Or central with one different
·Answers obtained number complete central
just by switching obtained answer numbers
surrounding numbers obtained obtained
without changing the
central numbers

The analysis revealed that compared to the pre-test, the percentages of students who
stopped at obtaining one correct answer (i.e., average score  1.33) were much smaller in
the post-test for both the classes. In fact, the two classes of students made significant
improvement in getting multiple correct answers from the pre- to post-test (Experimental:
Z = 3.39, p < 0.001, r = 0.58; Comparison: Z = 3.79, p < 0.001, r = 0.73). However, a
between-class comparison did not display any significant difference regarding the
improvements as well as students‟ performance on the particular performance rubrics in
either test.
It is believed in the study that representation is also an important skill in problem
solving. Therefore, although it is not a focus of the intervention program, how students
represent their solutions in the performance task tests was examined. The results revealed
that the students generally did not have significant changes in their representation from
the pre- to post-test. Moreover, consistent with the results on the other two performance
rubrics, the two classes of students did not have significantly different performance on
the aspect of representation in either the test.
Overall, it was found that both the experimental and comparison students made
significant improvement from the pre- to post-test, especially in the aspects of using
effective strategies and getting multiple correct answers. At first, the results appeared
somehow disappointing to us as it did not favour the experimental class. However, the
interviews with the teacher and students revealed that students‟ working on performance
tasks overall had a positive influence on their learning of mathematics. For instance, one
Lianghuo Fan & Yan Zhu 143

student felt that the open-endedness of performance tasks gave him a different view of
maths, say, “math is not just doing homework questions”. One student stated that doing
the performance tasks “allow us think differently”. A few students commented that the
new type of tasks help them “to think out of the box” instead of “sticking to one way”. As
to the teacher, although he expressed some concerns about some weak students, he
maintained that such tasks are generally beneficial for his students‟ learning in
mathematics, especially for “strong class [students] … who have the potential in them by
their own”.
A further discussion among the researchers and the teacher, to some degree, provides
some explanations to the above seemingly inconsistent findings. According to the teacher,
the students from the experimental class knew well about the research and they were
clear that all their grades on performance tasks would not be counted into their school
records. It might have affected their motivation in the post-test so that it is possible that
these students did not treat the test as seriously as their peers from the comparison class
who were just given the tests without further information. On the other hand, the result
might also imply that developing students‟ ability to a higher level in solving challenging
performance tasks could take a longer time than we have expected. In this regard, further
study is needed before we can make a definite conclusion, which is beyond the scope of
the current study.
Effects of using mathematics performance tasks on students’ mathematics
performance in solving conventional problems
Concerning students‟ performance in solving conventional mathematics problems,
we used students‟ PSLE mathematics grades and end-of-semester assessment scores
throughout the intervention period. As reported earlier, there was no significant
difference in the students‟ PSLE overall scores as well as mathematics grades (Exam A)
between the experimental and comparison students, which provided an indicator of
equivalence between the two classes. In Exam B, the equivalence still remained (t [75] =
0.02, p = 0.99). In fact, till the year 2004 mid-year school examination, the experimental
class only managed to implement one intervention. Therefore, no great change for the
experimental class was expected. More interventions were carried out later on, as shown
in Figure 1.
10

8
5 interventions
6

4
6 interventions
2

0
Exam B Exam C Exam D
144 Using Performance Assessment in Secondary School Mathematics

Figure 1. Difference of the average scores between the experimental

and comparison class on school exams.

From the figure, we can easily find that the differences between the two classes
increase rapidly in the last two school exams. In particular, the two differences reached a
considerably significant level (Exam C: t [74] = 1.94, p = 0.06, r = 0.22; Exam D: t [74]
= 1.96, p = 0.05, r = 0.22). A repeated measurement analysis of variance between Exam
B and Exam D revealed that there was a significant interaction between time and
treatment effects (F [1, 74] = 8.39, p < 0.005, r = 0.32) and the effect size is about
medium, which is in favour of the experimental class. A further analysis revealed the
significant interaction actually occurred between the period from Exam B and Exam C (F
[1, 73] = 6.68, p < 0.005, r = 0.29), and in the next period (from Exam C to Exam D), the
experimental class held the superiority. It appears clear that the students from the
experimental class had an advantage over the comparison students when doing their
conventional school exam tasks, which suggests that students‟ exposure to performance
tasks is beneficial for them to solve conventional tasks, an implicit hypothesis we had at
the beginning of the study. It is also interesting to investigate how long the positive
influence would maintain, which is, however, beyond the scope of the study.
Effects of using performance tasks on students’ attitudes toward mathematics and
mathematics learning
The data about students‟ attitudes towards mathematics and mathematics learning
are collected through the pre- and post-questionnaire surveys. As said earlier, Part I in
both questionnaires was the same, with focus on their general perceptions about
mathematics and mathematics learning, and the items in Part I can be categorised into
four subgroups.
The first subgroup, consisting of six items, was about students‟ general views about
mathematics and their learning of mathematics. The data revealed that the two classes of
students overall provided positive responses to these items in both the surveys. However,
it was also found that the students in both the classes became more negative in the post-
than in the pre-survey in terms of average rating2.
In general, there were no significant differences between the two classes of students
in terms of their general views toward mathematics and mathematics learning, though the
differences were in favour of the experimental class in both the surveys. Nevertheless, the
difference between the two classes became smaller in the post- as compared to the pre-
survey. In particular, while the experimental students appeared significantly more willing
to spend time in studying mathematics than the comparison students in the pre-survey
(Item 16: U [38, 39] = 552.50, p < 0.05, r = 0.25), the responses between the two classes
had no significant difference in the post-survey (Item 16: U [33, 29] = 363.50, p = 0.10).
We think a possible reason for this change is that the students from the experimental class
2
This result was not surprising, as available research has found that students at lower grade levels often
have more positive views about mathematics because of a variety of reasons, for example, mathematics
becomes more challenging to students at higher grade levels (e.g., see Macnab & Payne, 2003; Wong, Lam,
Wong, Leung, & Mok, 2001).
Lianghuo Fan & Yan Zhu 145

had more opportunities to work on performance tasks, therefore they might have
developed a perception that they had made enough effort and spent enough time in
studying mathematics. Another possible reason is that these students might have
experienced more frustrations because of solving challenging performance tasks, which
were reflected in the interview. For example, one student stated that “[we] have to gather
a lot of information and use it and sometimes … we don‟t [know] where to put what in
the question” and another said that “you do a lot of work, you still cannot solve, you are
very irritating”.
The second subgroup, also consisting of six items, was about students‟ anxiety level
in learning mathematics. While the students from the experimental class gave overall
positive responses to all the relevant items in the two surveys, those from the comparison
class only provided positive response to all the items in the pre-survey, but not in the
post-survey. In particular, the comparison students expressed in the post-survey that they
were somehow under terrible strain in mathematics lessons (Item 2) and not confident
when it came to mathematics (Item 20).
The results also showed that the experimental students were consistently less anxious
about their mathematics learning than the comparison students, though both the classes
became more anxious from the pre- to post-survey. As can be found from Table 3, there
are statistically significant differences in students‟ responses to a number of items
between the two classes in the post-survey. Specifically, the experimental students were
significantly less stressed (Item 2), less afraid of (Item 6), less nervous (Item 17), and
more confident about mathematics (Item 20) than their counterparts and the effect sizes
on the four items ranged from 0.33 to 0.42 with an average being 0.38. In comparison, no
significant difference was found in the pre-survey.

Table 3
Comparison Between the Experimental and Comparison Classes
on Anxiety Level Items
Pre-Survey (U) Post-Survey (U)
Item 2 654.50 293.50**
Item 6 675.00 297.00**
Item 10 677.50 357.00
Item 14 600.00 403.50
Item 17 570.00 322.00*
Item 20 645.00 278.00**
Note. * p < 0.05, ** p < 0.01; The U values in the tables are obtained by Mann-
Whitney U-test, which examines the differences in the ranked positions of ratings
between the experimental and comparison classes.

The fact that the experimental students have been exposed to performance tasks
appears to be one reason for such a result. As the performance tasks are generally more
challenging than normal school mathematics tasks, the experimental students then had
more opportunities to be engaged in higher-order thinking via working on those tasks.
Therefore, these students became less anxious about mathematics for both challenging
tasks as well as normal school tasks.
146 Using Performance Assessment in Secondary School Mathematics

The third subgroup of items was about students‟ perceptions about their own
performance in mathematics. It also comprises six items. The data revealed that the
students were happy with their own performance, however unlike experimental students,
the comparison students indicated in general that they did not like solving challenging
mathematics problems in the post-survey (Item 21). A comparison of the two classes of
students‟ responses to this item showed that the experimental students were significantly
more willing to attempt challenging mathematics tasks than the comparison students in
the post-survey (U [33, 29] = 333.00, p < 0.05, r = 0.30) but no significant difference was
found in the pre-survey (U [38, 40] = 671.00, p = 0.37). It appears that the experience
with performance tasks did help the experimental students to develop a positive
disposition towards working on challenging tasks.
However, we also found that while the experimental students had significantly
stronger beliefs that they could do well in mathematics (Item 15) than the
comparison
students in the pre-survey (U [38, 39] = 546.50, p < 0.05, r = 0.26), such a difference
did not show again in the post-survey (U [33, 29] = 398.50, p = 0.25). This result is not
surprising to us, because the performance tasks the experimental students had worked on
during the intervention period were generally not easily solvable, and this fact would
likely lead the experimental students to better appreciate the challenging nature of
mathematical tasks and hence give a more modest answer, compared to their counterparts
in the comparison class whose responses may only refer to the normal school
mathematics tasks they had encountered.
The fourth subgroup comprised four items designed to examine students‟ beliefs
about the usefulness of mathematics, which is related to one particular feature of the
performance tasks used in this study: authenticity in the task context. The results showed
that the students from both the classes provided overall positive responses on all the
relevant items in the two surveys and no significant between-class differences were
detected in either survey. However, compared to the pre-survey, the students‟ responses
in the post-survey became more negative for both the classes. Such a change is quite
understandable, as in general mathematics became more abstract and appeared further
away from students‟ daily life when students moved to higher grades. In particular, we
noticed that for the experimental students, the negative change on the item about the
meaningfulness of studying mathematics reached significant level (Item 12: U [38, 33] =
436.00, p < 0.05, r = 0.30). This result may be related to the fact that while the
experimental students were given many opportunities to work on performance tasks
which involved real-life application of mathematics knowledge, the skills they learned
from the new assessment strategy, however, were seldom assessed in their formal school
examinations. This inconsistent practice may lead students to believe that working on
performance tasks was somehow a waste of time. The interviews also revealed a number
of such thoughts. For example, one student commented that “some [tasks] are not
relevant to our normal maths” and the teacher told us that “none of these are tested in the
academic tests, … so … the kids also sometimes question whether …they need to spend
so much time on these … the first question they ask … „is this counted in CA1?‟” These
Lianghuo Fan & Yan Zhu 147

results reminded us of the importance of aligning teaching and assessment, especially for
those students who are test-oriented.
Part II of each of the questionnaires was targeted on students‟ experience with new
assessment strategies, including authentic tasks as well as open-ended tasks. Particularly
for the experimental students, items were added to measure their perceptions about
performance tasks in the post-survey.
The data from the pre-survey showed that both the experimental and comparison
classes had overall similar experience in doing the tasks with the aforementioned features.
Basically, they worked on the tasks with the relevant features either on a monthly basis or
a weekly basis. The comparison students‟ responses to the same items in the post-survey
were not significantly different from those in the pre-survey, which indicates that the
teaching practice in the comparison class remained unchanged in terms of the use of
performance tasks and it is consistent with the research design.
Regarding the experimental students‟ new experience with performance tasks, the
results from the post-survey revealed that they generally accepted well the specific
features of the performance tasks, including multiple approaches of the tasks (Item 26)
and the authenticity in task contexts (Item 30 & Item 31). Moreover, the students
believed that doing performance tasks helped them to be more creative (Item 27) and
systematic (Item 32). However, it seems that, to a degree, the students were still
uncomfortable with the open-endedness in final answers. In the interviews, some students
also told us about their confusions. For example, one student told us, “because you have
found one of the answers, and then if we check with other people for the answers, we
thought that either one of us was wrong”. It is understandable that in their previous
school experience, students were often merely required to provide one and only one
correct answer to a task and they had already been used to such practice in solving
mathematical tasks and felt comfortable with it.
The experimental students generally felt that doing performance tasks was very
challenging. More than 60% of the students claimed that they had to think harder in doing
the tasks (Item 28), 58% believed that it was time-consuming (Item 35), about 21% felt
lost in doing the tasks (Item 29), and 36% needed hints when working on those tasks
(Item 33).
In terms of the usefulness of doing performance tasks, the results showed that the
majority of the students did not have negative views toward such experience; in particular,
they believed that doing those performance tasks could help them in learning
mathematics (Item 25) and made them learn mathematics better (Item 37). However,
about one third of students felt that working on performance tasks was a waste of time
(Item 38) and only slightly more than one tenth of the students were willing to take more
performance tasks in their future learning (Item 36). Such a result could be related to the
fact that, as mentioned earlier, performance tasks were not included in the formal school
exams. Therefore, some of the experimental students were unable to “see” the immediate
benefit of doing the tasks for at least it does not seem to help them to get higher marks in
the conventional school tests. Consequently, these students did not fully see the
usefulness of doing performance tasks in their mathematics learning and hence became
unwilling to have more in future study.
148 Using Performance Assessment in Secondary School Mathematics

Conclusions and Implications

This study aimed to investigate the effects of integrating performance assessment
into regular school mathematics teaching and learning. In the study, performance tasks
were defined as those contextualised in real-world settings, approachable using various
methods, and with multiple acceptable answers. Thirty-eight Secondary One students in
an intact class from one randomly selected high-performing Singapore school received a
three-school-semester intervention, with the other forty students as an intact comparison
class. By using questionnaire surveys, performance task tests, students‟ normal school
exam scores as well as interviews with teacher and students, the researchers investigated
the possible impact of using performance tasks on the experimental students‟
mathematics learning in both academic and affective aspects.
Regarding students‟ academic achievement, the study looked into students‟
performance in both the conventional assessment (school exams) and unconventional
assessment (i.e., performance task tests). The results in the normal school exams showed
that the changes in students‟ performance across three continuous school semester tests
were significantly preferable in the experimental classes. Moreover, the favourite changes
occurred after the intervention program had been implemented about one school year,
where the experimental class completed 7 interventions, and maintained till the
interventions ended. Although it is hard to attribute the positive result solely to students‟
experience with performance tasks, it appears clear that the students from the
experimental classes did benefit from being exposed to performance tasks.
In the unconventional tests, the students from both the experimental and comparison
classes performed significantly better in the post- than pre-tests, not only in terms of their
overall scores but also in specific performance domains, including using effective
strategies and obtaining multiple answers. However, no significant difference was
detected between the two classes regarding their progresses. As suggested earlier, one
possible reason for such a result is that the students from the experimental class well
knew that their performance in the test would not be counted into their school records,
which might have limited their performance in the test. The result might also imply that
developing students‟ ability to a higher level in solving challenging performance tasks
could take a longer time than we have expected. In this regard, further evidence is needed
before we can make a definite conclusion. Nevertheless, it is clear that no negative effect
of using the new assessment strategy was found on students‟ performance.
Consistent with many other researchers‟ findings (e.g., Macnab & Payne, 2003;
Wong, Lam, Wong, Leung, & Mok, 2001), the students in this study in general become
more negative toward mathematics and mathematics learning in the post- from the pre-
survey. On the other hand, the study also revealed that the changes were generally in
favour of the students from the experimental class, especially in the anxiety level about
mathematics. However, it appears not expected that the changes on the view about the
usefulness of mathematics were preferable to the students from the comparison class,
while contextualisation in real life was one important characteristic of performance tasks,
and the experimental students expressed their appreciation for the specific features of
performance tasks in Part II of the post-survey as well as interviews. One possible reason
Lianghuo Fan & Yan Zhu 149

for the seemingly contradicting results is that the students may find that what they
experienced in the performance tasks seldom appeared in their regular school
mathematics learning. To them, working on normal school mathematics tasks was more
important, as it would really be tested. Correspondingly, the experimental students may
have an even stronger feeling that the mathematics they encountered in the regular school
learning was farther from their daily life.
The results from the study suggested that teachers and students were capable of
handling performance tasks. Although the effects of using the new strategy in some cases
were not obvious, the study did observe positive effects on students‟ academic
achievement and their anxiety level about mathematics learning. Moreover, it is clear that
no negative impact was observed. In short, the overall results appear to support teachers‟
using contextualised problems in real life situations and open-ended investigations in
students‟ learning of mathematics.
Finally, we would like to point out that this study was an initial step for us to explore
the possible effects of using performance tasks on both teachers‟ teaching and students‟
learning of mathematics. More research on the impact of using such new strategies on
teaching and learning in various aspects is needed. Moreover, given the complexity of the
practice of teaching and learning, there is a long way for us to go to fully understand how
the new assessment strategy can be effectively used to improve the quality of teaching
and learning, particularly with different students (e.g., with different abilities).
Acknowledgement
The study reported in this article was funded by the Ministry of Education of
Singapore through the Centre for Research in Pedagogy and Practice (CRPP), National
Institute of Education, Nanyang Technological University, Grant No. CRP24/03FLH. All
the views expressed are solely those of the authors and do not reflect the views of the
CRPP. The authors also wish to thank Mr. He Ming for his technical assistance in the
final stage of completion of this article.
References
Brookhart, S., Andolina, M., Zuza, M., & Furman, R. (2004). Minute math: An action
research study of student self-assessment. Educational Studies in Mathematics, 57,
213-227.
Buechler, M. (1992). Performance assessment. Policy Bulletin, No. PB-B13.
Clarke, D. (1997). Constructive assessment in mathematics: Practical steps for
classroom teachers. Berkeley, CA: Key Curriculum Press.
Fan, L. (2006). Making alternative assessment an integral part of instructional practice. In
P. Y. Lee (Ed.), Teaching secondary school mathematics: A resource book (pp.343-
354). Singapore: McGraw Hill.
Fan, L., & Zhu, Y. (2000). Problem solving in Singaporean secondary mathematics
textbooks. The Mathematics Educator, 5(1/2), 117-141.
Fan, L., & Zhu, Y. (2007). Representation of problem-solving procedures: A comparative
look at China, Singapore, and US mathematics textbooks. Educational Studies in
Mathematics. 66(1), 61-75.
150 Using Performance Assessment in Secondary School Mathematics

Fan, L. (Ed.) (2008). Performance assessment in mathematics: Concepts, methods, and

examples from research and practice in Singapore classrooms. A resource booklet
presented for the 2008 seminar organized by the Curriculum Planning and
Development Division of the Ministry of Education for the Heads of Departments of
Mathematics in secondary schools, Singapore.
Gripps, C. V. (1994). Beyond testing: Towards a theory of educational assessment.
London: Falmer Press.
Hargreaves, A., Earl, L., & Schmidt, M. (2002). Perspectives on alternative assessment
reform. Educational Researcher, 39(1), 69-95.
Harmon, M., Smith, T. A., Martin, M. O., Kelly, D. L., Beaton, A. E., Mullis, I. V. S., et
al. (1997). Performance assessment in IEA’s Third International Mathematics and
Science Study. Chestnut Hill, MA: TIMSS International Study Center, Boston
College.
Howe, A. C., & Jones, L. (1998). Engaging children in science (2nd ed.). Upper Saddle
River, NJ: Merrill.
Kane, M. B., Khattri, N., Reeve, A. L., & Adamson, R. J. (1997). Assessment of student
performance. Washington, DC: Studies of Education Reform, Office of Education
Research and Improvement, U.S. Department of Education.
Kulm, G. (1994). Mathematics assessment: What works in the classroom. San Francisco:
Jossey-Bass.
Macnab, D. S., & Payne, F. (2003). Beliefs, attitudes and practices in mathematics
teaching: Perceptions of Scottish primary school student teachers. Journal of
Education for Teaching, 29(1), 55-68.
Morgan, C. (2000). Better assessment in mathematics education? A social perspective. In
J. Boaler (Ed.), Multiple perspectives on mathematics teaching and learning (pp.
225-242). Westport, CT: Ablex.
National Council of Teachers of Mathematics. (1995). Assessment standards for school
mathematics. Reston, VA: Author.
Niss, M. (1993). (Ed.). Cases of assessment in mathematics education: An ICMI study.
Dordrecht, The Netherlands: Kluwer.
Pegg, J., & Panizzon, D. (2007-2008). Addressing changing assessment agendas: Impact
of professional development on secondary mathematics teachers in NSW.
Mathematics Teacher Education and Development, 9, 66-80.
Ruthven, K. (1994). Better judgment: Rethinking assessment in mathematics education.
Educational Studies in Mathematics, 27, 433-450.
Schoenfeld, A. H. (1992). Learning to think mathematically: Problem solving,
metacognition, and sense making in mathematics. In D. A. Grouws (Ed.), Handbook
on mathematics teaching and learning (pp. 334-370). New York: Macmillan.
Singapore Ministry of Education. (2002). Mathematics syllabus (lower secondary).
Singapore: Curriculum Planning Division.
Singapore Ministry of Education (2004a). Assessment guides to primary mathematics.
Singapore: Author.
Singapore Ministry of Education (2004b). Assessment guides to lower secondary
mathematics. Singapore: Author.
Lianghuo Fan & Yan Zhu 151

Stenmark, J. (1991). Mathematics assessment: Myths, models, good questions, and

practical suggestions. Reston, VA: National Council of Teachers of Mathematics.
Wisconsin Education Association Council. (1996). Performance assessment. Madison,
WI: Education Issues Series.
Wong, N.-Y., Lam, C.-C., Wong, K.-M. P., Leung, F. K.-S., & Mok, I. A.-C. (2001).
Students‟ views of mathematics learning: A cross-sectional survey in Hong Kong.
Education Journal, 29(2), 37-59.
Wu, H. (1994). The role of open-ended problems in mathematics education. Journal of
Mathematics Behavior, 13(1), 115-128.
Appendix

Students’ Attitude toward Mathematics and Mathematics Learning Survey Items

A. Part I (for both Pre- and Post-invention surveys)

1. I enjoy doing mathematics.

2. I am never under a terrible strain in a math class.
3. I am sure I can learn mathematics well.
4. I believe mathematics is useful.
5. Mathematics is hard for me.
6. I am not afraid of doing mathematics.
7. I can get good grades in mathematics.
8. It is important to know mathematics nowadays.
9. Mathematics is interesting to me.
10. I am unable to think early when doing mathematics.
11. I am not good at mathematics.
12. Studying mathematics is a waste of time.
13. I don‟t have good feelings about mathematics.
14. I feel lost when trying to solve math problems.
15. I don‟t think I can do well in mathematics.
16. I like spending time on studying mathematics.
17. It makes me nervous to even think about having to do a math problem.
18. I will use mathematics a lot as an adult.
19. I don‟t like to attend math lessons.
20. I have a lot of confidence when it comes to mathematics.
21. I like solving challenging math problems.
22. I would rather have someone give me the solution to a difficult math problem than to
have to work it out for myself.
B. Part II (for pre-intervention survey; only items relevant to performance tasks are
included)
24. In the last school term, my math teacher encouraged me to solve math questions in
different ways.
27. In the last school term, how many math questions did your teacher ask you to do that
have more than 1 correct answer?
152 Using Performance Assessment in Secondary School Mathematics

28. In the last school term, how many math questions did your teacher ask you to do that
have nothing to do with real life situations?

C. Part II (for post-intervention survey)

23. I like to solve mathematics questions which have more than one correct answer.
24. Doing mathematics performance tasks is difficult to me.
25. Doing performance tasks helps me to learn mathematics.
26. I like to do mathematics questions which could be solved using different methods.
27. Doing mathematics performance tasks help me to be more creative in problem
solving.
28. I have to think harder when I am doing mathematics performance tasks.
29. I feel lost when I am doing mathematics performance tasks.
30. I like to do mathematics questions which involve the real world.
31. Doing mathematics performance tasks helps me see more connection between
mathematics and daily life.
32. Doing mathematics performance tasks helps me to become more systematic when I
am solving mathematics problems.
33. I need hints to help me do mathematics performance tasks.
34. I am good at doing mathematics performance tasks.
35. Doing performance tasks takes me more time than doing other mathematics questions
usually done in class.
36. I would like to have more mathematics performance tasks for my mathematics
lessons.
37. Doing mathematics performance tasks makes me learn mathematics better.
38. Doing mathematics performance tasks is a waste of time.

Authors:
Lianghuo Fan
Nanyang Technological University, Singapore
Email: lianghuo.fan@nie.edu.sg

Yan Zhu
University of Hong Kong, Hong Kong
Email: yanzhu@hkucc.hku.hk