Chinese/English Journal of Educational Measurement and

Evaluation | 教育测量与评估双语季刊

Volume 2 Issue 3 Article 3

2021

Analysis of a Mathematical Problem-Solving Test on Speed and

Students' Strategies: A Mixed Item Response Theory Approach
Chunlian Jiang
University of Macau

Do-Hong Kim
Wayne State University

Chuang Wang
University of Macau

Follow this and additional works at: https://www.ce-jeme.org/journal

Part of the Curriculum and Instruction Commons, Educational Assessment, Evaluation, and Research
Commons, International and Comparative Education Commons, and the Science and Mathematics
Education Commons

Recommended Citation
Jiang, Chunlian; Kim, Do-Hong; and Wang, Chuang (2021) "Analysis of a Mathematical Problem-Solving
Test on Speed and Students' Strategies: A Mixed Item Response Theory Approach," Chinese/English
Journal of Educational Measurement and Evaluation | 教育测量与评估双语季刊: Vol. 2 : Iss. 3 , Article 3.
Available at: https://www.ce-jeme.org/journal/vol2/iss3/3

This Article is brought to you for free and open access by Chinese/English Journal of Educational Measurement
and Evaluation | 教育测量与评估双语季刊. It has been accepted for inclusion in Chinese/English Journal of
Educational Measurement and Evaluation | 教育测量与评估双语季刊 by an authorized editor of Chinese/English
Journal of Educational Measurement and Evaluation | 教育测量与评估双语季刊.
Chinese/English Journal of Educational Measurement and Evaluation
Vol. 2, No. 3, 21-34, October 2021 21

Analysis of a Mathematical Problem-Solving Test on Speed and

Students’ Strategies: A Mixed Item Response Theory Approach
Chunlian Jiang a , Do-Hong Kim b , and Chuang Wang a
a University of Macau
b Wayne State University

Abstract Keywords
The present study used the mixed item response theory (IRT) model to identify Word problem;
qualitatively distinct subgroups of sixth-grade students with respect to their Problem-solving strategies;
performance on word problems on speed. A total of 345 Singaporean students and Mixed IRT model;
361 Chinese students took a problem-solving test on speed. The mixed IRT analysis Arithmetic strategies;
revealed two latent classes — the algebra proficient group and the algebra novice Algebraic strategies
group. The algebra proficient group was more likely to use traditional algebraic
and arithmetic strategies to solve the problems, whereas the algebra novice group
was more likely to use model drawing, unitary, and guess-and-check strategies, in
addition to using traditional arithmetic and algebraic strategies. Findings of the study
indicate that a greater variety of problem-solving strategies could be encouraged in
upper primary schools to help students make connections among these strategies, in
particular, between these strategies and the abstract algebraic strategies, and finally
to achieve a successful transition from arithmetic to algebra learning.

Introduction appeared in mathematics textbooks (Mayer, 1981). Our

Since the 1980s, mathematical problem solving has previous study investigated the measurement properties of
been one of the main topics in mathematics education a mathematical problem-solving test on speed (Jiang et
(NCTM, 1989). It has become one of the objectives al., 2018), and the results revealed the misfit of some
of mathematics curricula in almost all countries around items to the unidimensional Rasch model and differential
the world. For example, it was incorporated into item functioning (DIF) between Singaporean and Chinese
Singapore syllabi for both primary and secondary schools students. The present study used a mixed item response
in 1992 (Ministry of Education, 1990a, 1990b) and theory (IRT) model to determine whether two or more
is still the central focus of the pentagonal framework distinct latent classes could be identified, which differed
of the Singapore mathematics curriculum (Curriculum with respect to parameter estimates and strategy use.
Planning and Development Division [CPDD], 2012, 2019,
1 Literature Review
2020). Developing students’ abilities to solve application
problems has always been one of the objectives of China’s 1.1 Mathematical Problem Solving in Mathematics
mathematics curricula at both primary and secondary levels Curricula in Singapore and China
(National Institute for Curriculum and Textbook Research, Mathematical problem solving is at the center of the
1999). Word problems, as a bridge linking the real world Singapore mathematics curriculum framework (CPDD,
and the mathematical world, are widely used to promote 2019, 2020). One of the three aims of mathematics
students’ understanding of mathematical concepts and to education in Singapore is to enable students to develop
develop their abilities to solve mathematical problems cognitive and metacognitive skills through a mathematical
(Verschaffel & De Corte, 1997). Speed word problems approach to problem solving (CPDD, 2012). Heuristics,
are such a kind of word problems that most frequently

CONTACT: Chunlian Jiang. [email protected]. Faculty of Education, University of Macau.

Translators: Chunlian Jiang. [email protected]. Faculty of Education, University of Macau; Tianying Feng. [email protected].
300 Charles E. Young Drive North, Los Angeles, CA 90095.
 CEJEME 22

which are general rules of what students can do to tackle up 12% (133/1097) of algebra word problems that were
a problem (e.g., “using a representation” or “making a selected from ten standard algebra textbooks used in U.S.
guess”), are suggested in the mathematics syllabus for lower secondary levels. Moreover, the number of motion
primary schools (CPDD, 2012). A unit entitled “Solving problems was the highest among the 21 problem families
word problems” is often included at the end of a chapter in including motion, current, work, unit cost, coins, and so
primary mathematics textbooks (Lee et al., 2018a, 2018b). on. Word problems on speed are also frequently included
Mathematical problem solving also plays an important in international studies such as the Trends in International
role in the mathematics curriculum in China (Cai & Nie, Mathematics and Science Study (TIMSS) (Mullis et al.,
2007). Research on mathematical problem solving often 2012) and Programme for International Student Assessment
focuses on the study of multiple solutions of one problem, (PISA) (Organization for Economic Co-operation and
the use of multiple problems to teach one concept, and the Development [OECD], 2014). Word problems on speed
discussion of multiple variations of one problem to help were selected for the present study because they are
students form a coherent scheme of an important concept application problems of various mathematical concepts
including speed (Cai & Nie, 2007; Jiang et al., 2014). In from primary to university levels (Jiang, 2005, 2016).
an analysis of 264 articles published in educational journals Word problems on speed have been presented in the
in Macao in 2009-2018 and were related to mathematics textbooks used in many countries. In Singapore, up to
education, we found that nearly half (48.5%) were related to 3-step word problems involving speed and average speed
mathematical problem solving; such a percentage was much are included in sixth-grade mathematics as a unit (CPDD,
higher than that related to mathematics curriculum (15.5%) 2012). Word problems about speed are presented under the
and instruction (22.7%) (Cheung et al., under review). unit of ratio in seventh-grade mathematics (CPDD, 2019).
It seems that the Singapore mathematics curriculum In China, word problems about speed have been
developers adopted the “teaching of mathematical presented in different units from time to time to illustrate
problem solving” approach, whereas China’s mathematics the applications of various mathematical topics in the
curriculum developers used a “teaching mathematics real world. For example, they are first presented in the
through mathematical problem solving” approach. The second-grade mathematics textbooks when multiplication
two different approaches might provide different learning and division are taught to eight- to nine-year-old students
opportunities and experiences for their students. Our (People’s Education Press, 2013). Since then, various
previous comparative study (Jiang & Chua, 2010; Jiang word problems about speed are presented in third- to
et al., 2014) found that the Singaporean students used sixth-grade mathematics textbooks when four operations
more categories of strategies (e.g., model drawing method, on whole numbers, decimals, and fractions are taught.
guess-and-check, and unitary methods) for solving word Word problems about speed that can be solved using linear
problems. They might benefit from the teaching of a equations are presented in sixth and seventh grades when
greater variety of problem-solving heuristics included solving linear equations is covered. Word problems about
in their mathematics curriculum (Fan & Zhu, 2007). In speed that can be solved using fractional expressions are
comparison, Chinese students were more likely to use the presented in eighth grade when fractional expressions are
traditional arithmetic and algebraic strategies, and their taught.
selection of strategy categories was a bit limited. This is Word problems about speed are treated differently in the
probably because they learned algebra earlier, and there Chinese and Singapore mathematics curricula, which may
were no such clear requirements regarding the teaching of bring about differences in student performance between the
different kinds of problem-solving strategies in China as in two countries. Our previous comparative study between
Singapore. sixth-grade Singaporean and Chinese students (Jiang et
1.2 Word Problems on Speed in Mathematics al., 2014) found that the Singaporean students performed
Curricula in Singapore and China better in one problem involving fractions, whereas the
Chinese students performed better in seven problems with
Word problems are often used to develop students’ medium-to-large effect sizes in the test. Most of the
problem-solving skills. Word problems on speed were problems in which the Chinese students performed better
frequently included in mathematics textbooks. Mayer were algebraic problems. In the problem that Singaporean
(1981) found that motion (i.e., speed) problems took students performed better, the students successfully used
23 Jiang et al.

model drawing and unitary methods to transfer the understand why students respond differentially to test items
multiplication of two fractions into the multiplication of and better understand potential sources of DIF.
whole numbers, which greatly reduced the difficulty level This study was guided by the following two research
of the problem (Jiang & Chua, 2010; Jiang et al., 2014). questions:
1.3 Studies Involving Word Problems on Speed 1. How many latent classes can be found based
A number of studies have included problems on speed as on the students’ performance on a mathematical
a specific rate model for multiplication and division (e.g., problem-solving test on speed?
Greer, 1992; Lamon, 2007). However, the problems used
in previous studies belong only to the simplest of the 13 2. How do students in the identified latent classes differ
motion problem categories identified by Mayer (1981). Our in terms of their uses of problem-solving strategies?
previous study investigated the psychometric properties of
a 14-item math test that included more diverse and more 2 Method
complex word problems on speed based on the Rasch’s 2.1 Test Instrument
IRT model (Jiang et al., 2018). Our analysis revealed
The instrument is a test that includes 14 word problems
the misfit of some items to the unidimensional Rasch
on speed (Appendix). It was developed based on Mayer’s
model and DIF between Singaporean and Chinese students.
(1981) classification of word problems on speed and an
Building on these findings, the present study was designed
analysis of word problems on speed in the textbooks
to determine whether there existed two or more distinct
used in China and Singapore. The following paragraphs
sub-populations (latent classes) which would differ with
provide brief descriptions of the 14 problems. The terms
respect to parameter estimates, resulting in measurement
“problems” and “items” are used interchangeably in this
non-invariance. We also examined whether there were
study because each item in the test is a word problem.
differences in the use of strategies between the students in
Problems 1-3. These three problems describe only one
the resulting latent classes, which, in turn, could identify
motion of an object. For a single motion, three variables —
potential sources of DIF. The differences might shed light
distance (D), speed (S), and time (T) — are involved and
on the impact of the instruction of mathematical problem
their relationship can be described as D = S × T . Given any
solving on student performance.
two of them, the third can be set as the unknown. Therefore,
If measurement invariance is violated, differences in
these are the basic word problems about speed.
scores between the Singaporean and Chinese students may
Problems 4 and 9. These two problems describe two
not reflect true differences in the mathematical concepts
motions of an object where the directions of the two
being measured. Predominant measurement models such as
motions can be assumed to be the same. Problem 4 can be
factor analysis and item response theory (IRT) assume that a
solved using arithmetic strategies. Problem 9 is a typical
single measurement model can be applied to all individuals
algebraic word problem like the Chickens and Rabbits
in a population. Measurement invariance studies using
Problem in the ancient Chinese mathematics book Sunzi
these approaches often use manifest groups such as gender
Suanjing (Horng, 2012).
or ethnicity. Limitations of these approaches are that group
Problems 8, 13, and 14. These three problems describe
membership must be known and it may not adequately
a round trip, where one object makes two motions with
model heterogeneity due to unknown characteristics. Mixed
the same distance but in opposite directions. Problem 8
IRT models can be used to investigate heterogeneous
can be solved using arithmetic strategies, whereas Problems
subpopulations that are qualitatively distinct with respect to
13 and 14 cannot be solved using the same methods. The
a measurement model. For example, mixed Rasch models
knowledge of inverse proportions could be used to obtain a
(Rost, 1990) extend the more traditional IRT models to
solution to Problem 13 but not to Problem 14 (Jiang, 2009).
take heterogeneity in the population into account, allowing
Problems 6, 10, and 11. These three problems describe
for different performance levels within latent classes.
two motions of two objects. In Problems 6 and 10, the two
Furthermore, in a mixed Rasch model, a unidimensional
objects are moving towards each other from two different
Rasch model holds within each latent class but with
points; in Problem 11, they are moving in the same direction
different sets of item difficulty parameters across the latent
with one ahead of the other. Arithmetic strategies can be
classes (Rost, 1990). Such kind of analysis might help us
used to solve Problem 6 but not Problems 10 and 11 if a
 CEJEME 24

student does not know the formulae. Problem 10 asks for process. If a student showed 100% of the solution process,
the time taken for the two objects to meet; Problem 11 asks but the solution process contained errors in computation,
for the time taken for one to catch up with the other. the solution was also graded as 3. If a student’s solution
Problems 5, 7, and 12. These three problems describe included 50% of the correct process, it was graded as 2. If
three motions of one object, where the directions of a student’s solution included some (less than 50%) correct
the three motions can be the same. They also involve steps, it was graded as 1. If a student’s answer showed no
fractions to represent the relationships between distances understanding of the problem, it was graded as 0. A blank
of individual parts of the journey to the entire journey or response also received 0 points.
to the remaining journey after the first motion. This kind 2.4 Data Analysis
of problem was found in a popular workbook written by
Fong (1998). They were included to examine whether the The mixed Rasch model was used in this study.
students could apply the concept of average speed of two The fit of the mixed Rasch model of different class
motions to three motions of an object. The results from our solutions was evaluated by Akaike’s Information Criterion
previous study indicated that students could apply such a (AIC), Schwartz’s Best Information Criterion (BIC), and
concept (Jiang & Chua, 2010; Jiang et al., 2014). Bozdogan’s Consistent AIC (CAIC) implemented in the
Problems 1-3 are short answer questions. Problems 4-14 program WINMIRA (von Davier, 2001). The model with
are open response questions that ask participants to write the smallest fit index value was selected. Beyond relying
down their working process in the space provided below. on statistical criteria, we also considered the relative sizes
As previously stated, Problems 4-8 are arithmetic problems, of the latent classes and the magnitude of the parameter
and Problems 9-14 are algebraic problems. differences across latent classes. Any improvement in
model fit was evaluated in relation to its substantive
2.2 Participants significance.
A sample of 706 sixth graders from China and Singapore Once the best fitting model was identified, item
participated in the study. The 361 Chinese students difficulties and person estimates within each class were
were from three primary schools in Wuhan and the 345 examined to characterize the latent classes. The item
Singaporean students were from four primary schools in Q-index (Rost & von Davier, 1994) was used to examine
Singapore. Wuhan is the capital city of Hubei Province, the fit of the items within each latent class. The Q-index
located in Central China. Schools were selected from all ranges from 0 to 1, with 0 indicating perfect fit (perfect item
three districts in Wuhan. The schools were recommended discrimination), 1 indicating perfect misfit (perfect negative
by an officer who had worked with the Hubei Provincial item discrimination), and .5 indicating random response
Department of Education for five years. In Singapore, behavior. The standardized Q-index (a z-statistic) was used
classes were selected from each school to reflect the average to determine whether the item pattern deviation from the
academic level of the sixth graders in the school after model was statistically significant. The class membership
consultation with the principals about the mathematics probabilities for each student were also estimated. Based on
performance of each class. No calculators were allowed. these probabilities, each student was assigned to the latent
It took an average of 80 minutes for the students to finish class for which his or her membership probability was the
the test. All students had already learned the topic of speed highest.
at the sixth-grade level before the administration of the test. We also examined the strategies used by the latent
2.3 Scoring System classes for solving problems 4 to 14 (the first three
problems were short-answer problems). Five specific
Items 1-3 were short-answered questions. Students’ categories of problem-solving strategies were identified
responses to these items were scored dichotomously with (Jiang et al., 2014): arithmetic, algebraic, model-drawing,
1 for a correct answer and 0 for an incorrect answer. The guess-and-check, and unitary strategies. A no-strategy
other 11 items (Items 4-14) were rated on a 5-point scale category was included for responses where the student
from 0 to 4. A student would receive a grade of 4 if wrote nothing or only copied information from the problem
the student’s answer was correct and with an appropriate statement without any further work. The descriptions of
solution process. A student would receive a grade of 3 if the the six strategies are included in Table 1. The percentage
answer was correct but incomplete with 75% of the solution of students (SP) in each class using each strategy and the
25 Jiang et al.

Table 1
Descriptions of Problem-Solving Strategies

Strategy category Description

Arithmetic strategy The student writes down a mathematical statement involving one or more arithmetic
operations on the numbers given in the problem.
Algebraic strategy The student chooses one or more unknowns as variables and sets up one or more equations.
Model drawing strategy The solution is suggested by or follows a model or a diagram.
Guess-and-check The student uses the following process:
(a) Make a guess of an answer or the unknown in the problem based on an estimation;
(b) Check if the constraints given in the question or implied from some of the question
statements are satisfied. If all the constraints are satisfied, the guess is correct; the answer
has been obtained or can be worked out. All processes will end at this point. If the
constraints are not satisfied, the guess will be refined or adjusted, and another guess will be
made followed by another round of guess-and-check.
Unitary strategy The student finds the value equivalent to one unit of a quantity from an equivalence
statement and obtains the value equivalent to more units of the quantity using the value
for one unit just found.
No strategy Absence of a written response or only pieces of information taken from the question are
written down without any further work.

success rate (SR), defined as the percentage of students who unitary, and guess-and-check strategies are often used to
used a specific strategy to reach the correct answer, were help students to achieve a successful transition (Fong, 1994;
obtained in order to explain the difficulty levels of the items. Jiang, 2005). The strategy analysis will help us to better
Among the six strategies, only the unitary strategy might understand students at the transition stage and provide
be unfamiliar to us. Therefore, three solutions using this insights into the improvement of instructional design.
strategy for solving item 7 are given in Figure 1. In solving
Item 7, we can draw the intermediate conclusion that “ 74 3 Results
of the total distance is 36 kilometers” based on the givens 3.1 Model Fit and Item Difficulty Estimates
of the problem, and then we use 36 ÷ 4/7 to find the total
The fit values are listed in Table 2. Although the
distance of the whole journey. When using the unitary
one-class model had the smallest CAIC value, the two-class
strategy suggested in mathematics textbooks in Singapore,
model had the smallest AIC and BIC indices. On the basis
a student could first write a mathematical sentence like “4
of evaluating the statistical fit indices, the two-class model
units = 36 km.” The student then could find the distance for
with class sizes .57 and .43 appeared to fit the data the
one unit and the distances of all the individual parts of the
best. The mean person measure WLE (Warm’s modified
journey. The use of unitary strategy could help students
likelihood estimates) was higher for Class 1 (M = 1.81, SD
avoid difficulties with divisions involving fractions. A
= 1.01) than for Class 2 (M = 0.98, SD = 0.57). There were
similar process can also be used to help students understand
422 students in Class 1, among which 335 were from China
why we need to turn the divisor upside down when divided
and 87 from Singapore. There were 284 students in Class 2,
by a fraction. This is also the method that is used to deal
among which 258 were from Singapore and 26 from China.
with the division problem with a fraction divisor in the
Table 3 shows the means and standard deviations of
mathematics textbooks in Singapore. Similar approaches
item raw scores by the two classes. Class 1 performed
are also used for problems with ratios and proportions to
significantly better than Class 2 on ten items. Class 2
remove students’ learning difficulties.
performed better than Class 1 on two items; however, the
The sixth-grade students are at the transition stage from
differences did not reach a significant level. As previously
arithmetic thinking to algebraic thinking. Model drawing,
 CEJEME 26

Figure 1
Three Solutions Using Unitary Strategies for Item 7

Table 2 Table 3
Model Fit Indices for One-, Two-, and Three-Class Models Class-Specific Mean Score and Standard Deviation for
Each Item
Model AIC BIC CAIC
One class 14625.82 14885.72 14942.72 Class 1 Class 2
Two classes 14326.04 14850.40 14965.40 Item (n = 422) (n = 284) t
Three classes 14351.77 15140.58 15313.58 M SD M SD

Note. AIC = Akaike Information Criterion; BIC = Item 1 0.99 0.084 0.94 0.231 4.00***
Bayesian Information Criterion; CAIC = Consistent Akaike Item 2 1.00 0.069 0.99 0.102 0.90
Information Criterion. Item 3 0.99 0.084 0.97 0.175 2.49*
Item 4 3.59 1.063 3.70 0.782 −1.52
mentioned, items 1-8 were all arithmetic items and items Item 5 3.53 1.067 3.48 0.967 0.73
9-14 were algebraic items. Class 1 students performed Item 6 3.47 1.098 2.64 1.349 9.01***
significantly better on all the algebraic items than Class Item 7 3.05 1.378 3.19 1.132 −1.41
2 students, but their performances were not significantly Item 8 2.83 1.597 2.19 1.476 5.35***
different on half of the arithmetic items. Therefore, we
Item 9 3.12 1.584 2.09 1.867 7.88***
named Class 1 as the Algebra Proficient (AP) Class and
Class 2 as the Algebra Novice (AN) Class. Item 10 3.90 0.383 1.11 1.600 34.45***
Table 4 shows the class-specific Rasch item difficulties Item 11 3.35 1.346 1.50 1.472 17.31***
and item fit indices (Q-index). The standardized Q-indices Item 12 2.08 1.696 1.29 1.180 6.79***
indicated that all the items fit well for both classes. The Q Item 13 2.20 1.947 0.88 1.616 9.41***
values for all polytomously scored items (items 4-14) were
Item 14 1.68 1.945 0.27 0.973 11.25***
close to zero, ranging from .04 to .24 for the AP Class and
from .16 to .22 for the AN Class. The Q values for three *p < .05; ***p < .001.
dichotomously scored items (items 1-3) were relatively
large as shown in Table 2, but the standardized Q-indices
were not statistically significant, indicating adequate fit.
27 Jiang et al.

Table 4 7-9, whereas for the AN Class, item 11 was estimated to

Item Difficulty Estimates and Q-Indices Arranged in Order be more difficult than items 7-9. The following section
of Decreasing Difficulty on the strategy analysis for items 4-14 might provide some
explanations for the discrepancy in item difficulty between
Class 1 Class 2 the latent classes.
Item Item
Item difficulty
Q-index Item difficulty
Q-index 3.2 Differences in the Use of Problem-Solving
Strategies Between the Latent Classes of Students
Item 14 2.22 .04 Item 14 1.87 .22
Item 12 1.89 .07 Item 13 1.46 .22 In this section, we examined the mathematical
problem-solving strategies the two classes used. We
Item 13 1.87 .03 Item 10 1.37 .16
first summarized the differences in strategy use between the
Item 8 1.26 .09 Item 12 1.29 .22 two classes for the 11 open-response items (items 4-14).
Item 9 1.18 .07 Item 11 1.05 .22 We then examined the strategies used for individual items
Item 7 1.08 .06 Item 9 0.90 .18 in more details to illustrate the differences between the two
Item 11 0.91 .08 Item 8 0.69 .16 classes.
The percentages of cases (number of items [11] ×
Item 6 0.61 .12 Item 6 0.46 .21
number of students) where each class of students used each
Item 4 0.54 .12 Item 7 0.05 .22 strategy are presented in Table 5. These results provide
Item 5 0.52 .09 Item 4 −0.37 .22 an overarching picture of the significant differences in
Item 10 −1.59 .24 Item 5 −0.38 .22 strategy use between the two classes. Both classes used
Item 1 −3.21 .23 Item 1 −2.05 .55 arithmetic strategies the most frequently; the students in
the AP Class used them in a higher percentage of cases.
Item 3 −3.46 .79 Item 3 −2.60 .57
For the AP Class, algebraic strategies were used the second
Item 2 −3.82 .75 Item 2 −3.74 .69 most frequently. For the AN Class, model drawing was the
second most frequently used strategy, followed by unitary
and guess-and-check methods. The z-scores in the last
column indicated that the two classes were significantly
The item difficulty measures of the 14 items (except
different in the percentages of students using the five
items 8, 10, and 11) for the two classes were more or
strategies. However, there was no significant difference in
less within expectations. Items 1-3 were the easiest,
the percentages of students having no strategies between the
four arithmetic items (items 4-7) were at the intermediate
two classes.
difficulty level, and four algebraic items (items 9, 12-14)
were at the high difficulty level. In particular, item 14 was Table 5
found to be the most difficult item for both classes. Class-Specific Overall Strategy Use Across Items 4-14
There was a noticeable discrepancy in the item difficulty
order between the two latent classes for items 8, 10, and AP class AN class
Strategy category z
11. The difficulty estimate of item 8 for the AP Class (n = 4,642) (n = 3,124)
was just below items 12-14. However, for the AN Class,
Arithmetic strategy 65.92a 52.08 12.23***
it was at the intermediate difficulty level, close to items
4-7. The item difficulty estimate of item 10 for the AP Algebraic strategy 12.88 0.99 18.83***
Class was quite low and negative, indicating a relatively Model drawing strategy 4.63 17.99 −19.24***
easy item. Item 10, however, was the third most difficult Guess-and-check 5.60 10.69 −8.28***
item for the AN Class. Item 10 also showed the largest Unitary strategy 2.46 10.79 −15.39***
difference in item difficulty (2.96 logits) between the two
No strategy 8.51 7.46 1.66
classes. Our previous study (Jiang et al., 2018) using the
a The numbers were the percentages (% is omitted) of the
traditional Rasch model indicated that the item fit for item
10 was poor. Item 10 was also flagged as having DIF cases (number of items [11] × number of students) where
between the Singaporean and the Chinese students. For the the class of students used the specific kind of strategies.
AP Class, item 11 was estimated to be easier than items ***p < .001.
 CEJEME 28

Based on the item difficulty measures in Table 4, we in the AP Class (z = −6.92, p < .001). In total, nearly
classified items 4-14 into the following three groups: (a) 70% of the students in the AP Class used the two effective
items 4-7; (b) items 9 and 12-14; and (c) items 8, 10, 11. For strategies, whereas only about 60% of the students in the
each group, we examined differences in the use of strategies AN Class chose to use them. Nearly 4% of the students in
between students of the two latent classes. the AN Class used model drawing strategies while only two
Items 4-7. The percentages of students in each class students in the AP Class used it (z = −3.27, p < .01). The
using the different strategies for solving items 4-7 and the percentages of students having no strategies were similar in
success rates are shown in Table 6. Because the results for both classes.
items 4 and 6 are very similar, the following discussion will For item 12, the strategy use between the two classes
group them together, and similarly for the results for items was significantly different (χ 2 = 163.95, p < .001). The
5 and 7. students in the AP Class mainly used arithmetic and
For items 4 and 6, the majority of the students in algebraic strategies. However, for the students in the AN
both classes used arithmetic strategies, and the AP Class Class, the main strategies were model drawing and unitary
could use them more successfully than the AN Class. strategies. Although the use of model drawing and unitary
Higher percentages of the students in the AN Class used strategies could convert the division of fractions into whole
the model drawing strategy than those in the AP Class. numbers (Jiang & Chua, 2010), it seemed difficult for the
The percentages of students using algebraic strategies and AN Class, which led to a relatively lower success rate
having no strategies were all very low. compared with the use of all strategies by the AP Class.
For items 5 and 7, the strategy use between the two In solving items 13 and 14, for the AP Class, the two most
classes was significantly different (item 5: χ 2 = 309.22; frequently used strategies were arithmetic strategies (48.3%
item 7: χ 2 = 305.06; p < .001). The students in the and 37%) and algebraic strategies (27% and 29%). Close to
AP Class mainly used arithmetic and algebraic strategies. 10% of the students in the AP Class used guess-and-check
However, for students in the AN Class, the main strategies strategies for both items. For the AN Class, the majority of
were model drawing and unitary strategies. A key students used arithmetic strategies (66.9% and 57.7%). The
difference between these two kinds of problem-solving students in the AN Class also used guess-and-check (15.5%
approaches lies in the operations on fractions. In the use of and 8.8%) and model drawing strategies (5.6% and 8.4%)
the arithmetic and algebraic strategies, addition/subtraction when solving items 13 and 14. Based on the success rates,
and multiplication of fractions were involved. However, algebraic and guess-and-check strategies were two effective
when the model drawing and unitary strategies were used, strategies for items 13 and 14.
these operations were converted into operations with whole In summary, students in the AP Class performed better
numbers instead (Jiang & Chua, 2010; Jiang et al., 2014). than those in the AN Class on the four items because
These strategies provided the AN Class with an advantage they could use the effective guess-and-check and algebraic
in item 7 even though the performance differences between strategies, particularly on items 9 and 14. Although higher
the two classes were not significant. percentages of students in the AN Class used the model
Items 9 and 12-14. The percentages of students in each drawing and unitary strategies, their success rates were all
class using different strategies for solving items 9 and 12-14 very low.
and the success rates are shown in Table 7. Items 8, 10, and 11. Table 8 shows the percentages
For item 9, the strategy use between the two classes was of students in each class using the different strategies for
significantly different (χ 2 = 141.24, p < .001). The success solving items 8, 10, and 11 and their success rates.
rates revealed that algebraic and guess-and-check strategies For item 8, the majority of students in both classes used
were the two effective strategies for this item. The data the arithmetic strategies (83.2% for the AP Class; 82.7% for
in Table 7 showed that a higher percentage of students the AN Class). The second most frequently used strategy
in the AN Class used inappropriate arithmetic strategies for the AP Class was algebraic strategies, whereas it was
(z = −3.05, p < .01). In the use of two effective strategies, the model drawing strategy for the AN Class. The students
a higher percentage of the students in the AP Class used the in the AP Class used the arithmetic and algebraic strategies
algebraic strategies (z = 11.11, p < .01) than those in the more successfully than those in the AN Class.
AN Class, whereas a higher percentage of the students in Most students in both classes used arithmetic strategies
the AN Class used guess-and-check strategies than those for solving item 10 (AP Class: 94.3%; AN Class: 63.4%)
29 Jiang et al.

Table 6
Percentages of Students (SP) in the Two Classes Using Different Strategies and Success Rates (SR) for Items 4-7

AR MD US AL NS
SP SR SP SR SP SR SP SR SP
Item 4 AP class 93.8 86.4 2.8 100.0 0.0 —a 1.9 87.5 1.4
AN class 92.3 83.6 7.4 71.4 0.0 — 0.0 — 0.4
Item 6 AP class 86.5 76.4 10.0 88.1 0.0 — 0.2 100.0 3.3
AN class 70.4 34.0 28.5 60.5 0.0 — 0.0 — 1.1
Item 5 AP class 69.7 82.0 7.6 78.1 12.1 94.1 7.8 72.7 2.8
AN class 10.9 48.4 35.6 71.3 51.4 76.0 1.8 60.0 0.4
Item 7 AP class 65.6 64.3 10.0 83.3 10.0 83.3 6.4 29.6 8.1
AN class 10.6 23.3 49.6 60.3 37.0 61.9 1.8 60.0 1.1
Note. AR = arithmetic strategy; MD = model drawing strategy; US = unitary strategy; AL = algebraic strategy; NS =
no strategy. SP = percentage of students using a kind of strategy (strategy percentage); SR = percentage of students who
used the given strategy and solved the problem correctly (success rate). % was omitted.
a No SR could be calculated.

Table 7
Percentages of Students (SP) in the Two Classes Using Different Strategies and Success Rates (SR) for Item 9 and Items
12-14

AR MD US GC AL NS
SP SR SP SR SP SR SP SR SP SR SP
Item 9 AP class 18.7 53.2 0.5 50.0 0.0 —a 31.5 97.7 38.4 92.6 10.9
AN class 28.5 6.2 3.9 27.3 0.0 — 57.7 95.7 2.1 83.3 7.7
Item 12 AP class 43.4 39.9 13.7 67.2 4.3 44.4 0.0 — 15.9 61.2 22.7
AN class 28.2 3.8 33.8 22.9 27.8 7.6 0.0 — 1.8 0.0 8.5
Item 13 AP class 48.3 36.3 2.6 72.7 0.7 66.7 9.2 87.2 27.0 87.7 12.1
AN class 66.9 4.21 5.6 18.8 2.5 28.6 15.5 90.9 0.4 100.0 9.2
Item 14 AP class 37.0 16.0 1.2 0.0 0.0 — 9.2 92.3 29.1 87.8 23.5
AN class 57.7 0.0 8.1 0.0 0.0 — 8.8 68.0 1.1 0.0 24.3
Note. AR = arithmetic strategy; MD = model drawing strategy; US = unitary strategy; GC = guess-and-check; AL =
algebraic strategy; NS = no strategy. SP = percentage of students using a kind of strategy (strategy percentage); SR =
percentage of students who used the given strategy and solved the problem correctly (success rate). % was omitted.
a No SR could be calculated.

and item 11 (AP Class: 84.6%; AN Class: 61.3%). both classes, the second most frequently used strategy was
However, the AP Class had much higher success rates guess-and-check strategies (AP Class: 4.3% for item 10
than the AN Class. The solving procedures provided by and 7.3% for item 11; AN Class: 14.4% for item 10 and
the students of the two classes might be quite different. 21.1% for item 11). The success rates of the AP Class
The students in the AP Class used formulae for solving were much higher than those of the AN Class. For the AN
the two items, whereas students in the AN Class made Class, the third most frequently used strategy was model
irrelevant operations on the givens (Jiang et al., 2014). For drawing strategies (item 10: 7.0%; item 11: 4.9%). Higher
 CEJEME 30

Table 8
Percentages of Students (SP) in the Two Classes Using Different Strategies and Success Rates (SR) for Items 8, 10, and
11

AR MD GC AL NS
SP SR SP SR SP SR SP SR SP
Item 8 AP class 83.2 64.4 1.4 50.0 0.0 —a 10.9 67.4 4.5
AN class 82.7 29.8 13.4 55.3 0.0 — 2.1 16.7 1.8
Item 10 AP class 94.3 91.2 0.2 100.0 4.3 100.0 0.9 100.0 0.2
AN class 63.4 21.7 7.0 0.0 14.4 19.5 0.0 — 15.1
Item 11 AP class 84.6 81.5 0.9 75.0 7.3 83.9 3.1 100.0 4.0
AN class 61.3 17.2 4.9 14.3 21.1 55.0 0.0 — 12.7
Note. AR = arithmetic strategy; MD = model drawing strategy; GC = guess-and-check; AL = algebraic strategy; NS =
no strategy. SP = percentage of students using a kind of strategy (strategy percentage); SR = percentage of students who
used the given strategy and solved the problem correctly (success rate). % was omitted.
a No SR could be calculated.

percentages of students in the AN Class had no strategies 11 (z = −4.30, p < .001). This is surprising because
than those in the AP Class. All these factors contributed our previous study indicated that there were significant
to the items having relatively higher difficulty levels for the differences between the percentages of Singaporean and
AN Class than for the AP Class. Chinese students having no strategies for almost all items
Item 10 was relatively more difficult than item 11 for except items 4 and 14 (Jiang et al., 2014). Although there
the AN Class. This is probably because the computation was an overlap between the latent groups and manifest
involved and the answer (2 hours 5 minutes) in item 10 is groups, the result obtained in the current study indicated
more complicated than that of item 11 (7 hours). that students’ cognitive processes that are involved in
In summary, the analysis of strategy use by the two mathematical problem solving and lead to group differences
classes explained the differences in item difficulty estimates in item measures (i.e., measurement invariance, DIF) might
between the two classes. In solving the four arithmetic not be perfectly explained by the manifest groups.
items (items 4-7), the majority of the AP Class used
arithmetic strategies, whereas students in the AN Class 4 Conclusion, Discussion, and Implications
used the model drawing and unitary strategies in addition The results of this study indicated that two latent classes
to the arithmetic strategies. In solving the four algebraic were present in the data. The heterogeneity in measurement
items (items 9 and 12-14), higher percentages of students revealed in this study provides further explanations for the
in the AP Class used the guess-and-check and algebraic item misfit and DIF between Singaporean and Chinese
strategies with high success rates, which brought them students found in the previous study (Jiang et al., 2018).
better performance. In solving the three items (items 8, For the two classes, one consisted of algebra proficient
10, and 11) with a discrepancy in item difficulty order students, and the other consisted of algebra novice students.
between the two classes, the success rates of strategy use A separate set of item difficulty parameters was required for
by students in the AN Class were all very low. On the each class, and their item difficulty orders showed unique
other hand, students in the AP Class could use arithmetic patterns beyond what could be obtained using traditional
strategies with high success rates, leading to the discrepancy IRT methods. For both classes, items that could be solved
in performance and item difficulty. by directly using formulae were the easiest. Furthermore,
The z-tests were also conducted to compare the arithmetic items were at the intermediate difficulty level,
percentages of students having no strategies in the two and algebraic items were at the high difficulty level.
classes; it was found that there were only significant Another contribution of this study is the instrument
differences for items 10 (z = −8.05, p < .001) and that could be used in fifth- to seventh-grade mathematics.
31 Jiang et al.

Although the study was conducted with sixth-grade been used to represent the other four unknowns (D1 , T2 ,
students, the easy and the intermediate difficulty items D2 , and T D) (i.e., D1 = S1 × T1 = 15x; T2 = T T − T1 =
could be used with fifth-grade students and the high 6 − x; D2 = S2 × T2 = 75(6 − x); and T D = D1 + D2 =
difficulty items could be used with seventh-grade students. 15x + 75(6 − x)). The students might have forgotten what
This study revealed a big jump in difficulty levels from relationships they have used for deducing other variables in
arithmetic to algebraic word problems on speed, which is this long process and be stuck on what relationship to base
significant for mathematics teachers because they need to on for setting up the equation. If a student can produce a
give students more time in classroom instruction to allow complete guess-and-check cycle as the one shown in Figure
for the transition from arithmetic to algebraic thinking. 2, the student can be guided to deduce other variables
This is a contribution of the current study to the field of (Kieran, 1996) and eventually form an equation based on
mathematical problem solving. the constraint(s).
Mathematics educators often suggest using strategies,
such as the model-drawing, unitary, and guess-and-check Figure 2
strategies, to solve problems; the current study revealed that A Solution to Item 9 Using the Guess-and-Check Strategy
they might not be helpful for solving algebraic problems.
The strategies used for solving arithmetic problems could
not help the algebra novice students in Class 2 to achieve
a high success rate when compared to the algebraic
strategies used by the algebra proficient students in Class
1. Although model-drawing, unitary, and guess-and-check
strategies are valid and helpful methods for students to
gain an understanding of mathematical concepts, they are
not as efficient as algebraic strategies in solving algebraic
word problems. This finding supports the arguments of
Fong (1994) and Jiang (2005, 2016): it is important for
students in the transition period to gradually move from
model drawing, unitary, and guess-and-check to algebraic
strategies. Fong (1994) used several examples to show how In the unit “Equations” of secondary school mathematics,
to transition from the model drawing to algebraic strategies; the first lesson is “From Arithmetic to Equations.” What
the solutions in Figure 1 indicated that similar ways could relationship to use as the basis for setting up an equation
be used to build the links between unitary and algebraic is emphasized through a comparison between arithmetic
strategies. An example below shows how to help students and algebraic methods for solving word problems (Cai,
to move from the guess-and-check to algebraic strategies. 1998). The guess-and-check strategies could also be
In the solution shown in Figure 2, the student checks compared with algebraic strategies by emphasizing that
the constraint whether the total distance of the journey is the relationship used to form the equation in the algebraic
150 km, which is actually the basis on which most students method is usually the last constraint checked by the students
using the algebraic method set up equation(s) (15x + 75(6 − in the process of using the guess-and-check method. This
x) = 150, where x is set as the time she cycled). Researchers might be another way to help students to develop a
generally acknowledge that students have difficulties using better understanding of the connections between arithmetic
unknowns to express other unknowns when moving from strategies and algebraic strategies. Such an idea needs to be
arithmetic to algebraic thinking (Bednarz & Janvier, 1996; further investigated.
Stacey & MacGregor, 1999). However, we argue that Discrepancies existed in the item difficulty measures of
students might have greater difficulty choosing which three items. The strategy analysis revealed that the use
relationship to base on for setting up an equation. For of formula as a special kind of arithmetic strategies could
example, Problem 9 involves five relationships (T D = make items 10 and 11 easy. On the other hand, item
D1 + D2 , T T = T1 + T2 , D1 = S1 × T1 , D2 = S2 × T2 , 10 was relatively more difficult than item 11. This is
and T D = AS × T T ). In the process of setting up the probably because the complexity of computations involved
above equation, four of them except T D = AS × T T have in item 10 made the use of the guess-and-check strategy less
 CEJEME 32

effective, although this strategy is often taken as an effective 5(4), 225–229.

problem-solving strategy for such problems. Cai, J., & Nie, B. (2007). Problem solving in Chinese
Although the current study presents important new mathematics education: Research and practice. ZDM
information explaining item misfit and DIF between the Mathematics Education, 39, 459–475.
Singaporean and Chinese students found in the previous Cheung, K. C., Jiang, C., & Fan, L. (under review).
study (Jiang et al., 2018), limitations exist. First, only Research and research culture in mathematics
14 word problems on speed were included in the current education: The case in Macao, China. In A. Bill,
study. Further studies could include more problems, L. Fan, & V. Catherine (Eds.), Building research
for example, the problems with the distance-time and/or cultures in mathematics education: Learning from
speed-time graphs, in the test. Second, the current study asian countries.
collected data from Singaporean and Chinese students; Curriculum Planning and Development Division [CPDD].
further studies could be conducted with samples from other (2012). Mathematics syllabus: Primary one to six.
countries, in particular students from western countries. Singapore: CPDD, Ministry of Education.
Third, more evidence from the teaching of mathematical Curriculum Planning and Development Division [CPDD].
problem solving in classrooms is needed for explaining the (2019). Mathematics syllabuses: Secondary one
results obtained in the current study. Fourth, in addition to four, express course, normal (academic) course.
to the traditional IRT model used in our previous study Singapore: CPDD, Ministry of Education.
(Jiang et al., 2018) and the mixed IRT model used in Curriculum Planning and Development Division [CPDD].
the present study, future research can also adopt cognitive (2020). Mathematics syllabus: Primary one to six.
diagnostic models for analysis (Wang et al., 2016) to better Singapore: CPDD, Ministry of Education.
understand students’ cognitive processes so as to develop Fan, L., & Zhu, Y. (2007). Representation of
instruction tailored to students’ needs and to improve problem-solving procedures: A comparative look at
student achievement. Despite these limitations, this study China, Singapore, and US mathematics textbooks.
is the first that specifically focuses on word problems on Educational Studies in Mathematics, 66, 61–75.
speed and examines their item difficulty levels with students Fong, H. K. (1994). Bridging the gap between secondary
from Singapore and China, two high-performing countries and primary mathematics. Teaching and Learning,
in international mathematics study. We hope this study will 14(2), 73–84.
bring more research to similar mathematics topics. Fong, H. K. (1998). Solving challenging mathematical
In summary, the analysis using the mixed IRT model problems: The heuristic approach for primary
provided more insights into cross-culture differences in schools. Kingsford Educational Services.
mathematics achievement, which stem from curricular Greer, B. (1992). Multiplication and division as models
differences. The findings of this study have important of situations. In D. A. Grouws (Ed.), Handbook
implications for the validity and understanding of a of research on mathematics teaching and learning:
potential source of measurement error in mathematical A project of the National Council of Teachers of
problem solving. The findings also have clear instructional Mathematics (pp. 276–295). Macmillan Publishing.
implications for teachers with respect to instructional Horng, W. S. (2012). Narrative, discourse and mathematics
design. Lastly, the study provides insights and implications education: An historian’s perspective. In T. Y. Tso
for designing and conducting cross-culture research. (Ed.), Proceedings of the 36th Conference of
the International Group for the Psychology of
References Mathematics Education (Vol. 1, pp. 5–21). PME.
Bednarz, N., & Janvier, B. (1996). Emergence and Jiang, C. (2005). Strategies and errors in solving speed
development of algebra as a problem-solving tool: problems: A comparative study between Chinese
Continuities and discontinuities with arithmetic. In and Singapore students [Unpublished doctoral
N. Bednarz, C. Kieran, & L. Lee (Eds.), Approaches dissertation]. National Institute of Education,
to algebra: Perspectives for research and teaching Nanyang Technological University.
(pp. 115–136). Kluwer Academic Publishers. Jiang, C. (2009). Word problems on speed: Students’
Cai, J. (1998). Developing algebraic reasoning in the strategies and errors. In K. Y. Wong, P. Y. Lee,
elementary grades. Teaching Children Mathematics, P. Y. F. B. Kaur, & S. F. Ng (Eds.), Mathematics
33 Jiang et al.

education: The Singapore journey (pp. 244–262). Mullis, I. V. S., Martin, M. O., Foy, P., & Arora, A. (2012).
World Scientific. TIMSS 2011 international results in mathematics.
Jiang, C. (2016). Mathematical problem solving: Chinese TIMSS & PIRLS International Study Center, Boston
and Singaporean students’ strategies and errors in College.
solving speed word problems (in Chinese). Science National Council of Teachers of Mathematics [NCTM].
Press. (1989). Curriculum and evaluation standards for
Jiang, C., & Chua, B. L. (2010). Strategies for solving school mathematics. NCTM.
three fraction-related word problems on speed: A National Institute for Curriculum and Textbook Research
comparative study between Chinese and Singapore (Ed.). (1999). A compilation of curriculum standards
students. International Journal of Science and and syllabi for primary and secondary schools in
Mathematics Education, 8, 73–96. the 20th century: Mathematics. Beijing: People’s
Jiang, C., Hwang, S., & Cai, J. (2014). Chinese Education Press.
and Singaporean sixth-grade students’ strategies for Organization for Economic Co-operation and Development
solving problems about speed. Educational Studies [OECD]. (2014). PISA 2012 results: What
in Mathematics, 87, 27–50. students know and can do – student performance in
Jiang, C., Kim, D., & Wang, C. (2018). Psychometric mathematics, reading and science (Vol. 1). OECD
properties of an instrument to measure word problem Publishing.
solving skills in mathematics. Journal of Applied People’s Education Press (Ed.). (2013). The compulsory
Measurement, 19(3), 283–302. education textbook series: Mathematics 2b. People’s
Kieran, C. (1996). The changing face of school algebra. Education Press.
In C. Alsina, J. M. Alvarez, B. Hodgson, C. Laborde, Rost, J. (1990). Rasch models in latent classes: An
& A. Pérez (Eds.), Eighth International Conference integration of two approaches to item analysis.
on Mathematics Education: Selected lectures (pp. Applied Psychological Measurement, 14(3),
271–290). S.A.E.M. Thales. 271–282.
Lamon, S. J. (2007). Rational numbers and proportional Rost, J., & von Davier, M. (1994). A conditional item-fit
reasoning: Towards a theoretical framework for index for Rasch models. Applied Psychological
research. In F. Lester (Ed.), Second handbook of Measurement, 18(2), 171–182.
research on mathematics teaching and learning (pp. Stacey, K., & MacGregor, M. (1999). Learning the
629–668). Information Age Publishing. algebraic method of solving problems. The Journal
Lee, N. H., Koay, P. L., Charlotte, C., Ong, B. L., & Tan, of Mathematical Behavior, 18(2), 149–167.
C. S. (2018a). Shaping mathematics coursebook: 6A. Verschaffel, L., & De Corte, E. (1997). Word problems:
Marshall Cavendish Education. A vehicle for promoting authentic mathematical
Lee, N. H., Koay, P. L., Charlotte, C., Ong, B. L., & Tan, understanding and problem solving in the primary
C. S. (2018b). Shaping mathematics coursebook: 6B. school? In T. Nunes & P. Bryant (Eds.), Learning and
Marshall Cavendish Education. teaching mathematics: An international perspective
Mayer, R. E. (1981). Frequency norms and structural (pp. 69–97). Psychology Press/Erlbaum (UK) Taylor
analysis of algebra word problems into families, & Francis.
categories, and templates. Instructional Science, 10, von Davier, M. (2001). WINMIRA [Computer software].
135–175. Groningen, the Netherlands: ASC-Assessment
Ministry of Education. (1990a). Mathematics Systems Corporation.
syllabus: Primary 1 to 3, Primary 4 to 6 Wang, L., Guo, D., & Meng, M. (2016). Application of
(normal course), Primary 4 to 8 (extended course). cognitive diagnose model in mathematics educational
Singapore: Curriculum Planning Division, Ministry assessment (in Chinese). Journal of Mathematics
of Education. Education, 25(6), 15–19.
Ministry of Education. (1990b). Mathematics syllabus:
Secondary 1 to 2 (special/express course), Secondary
1 to 2 (normal course). Singapore: Curriculum
Planning Division, Ministry of Education.
 CEJEME 34

Appendix: The Mathematical Problem-Solving 9. On Sunday, Judy went to see her grandma who lived
Test on Speed 150 km away. After cycling at an average speed of 15
A. Short-answer questions km/h for a few hours, she got tired and took a lift from
the passing truck. The truck’s average traveling speed
1. A man drove at 72 km/h for 2 hours, then the distance was 75 km/h. When she got to her grandma’s house,
he traveled was km. she checked the time and knew that the trip took her 6
hours. Find the time she spent cycling.
2. It takes a motorist hours to travel 136 km at a
speed of 68 km/h. 10. Two places R and S were 300 km apart. Mike left R
3. 3. Judy cycled 45 km in 3 hours, then her speed was and drove at 84 km/h towards S. At the same time, Bill
km/h. left S at 60 km/h and drove towards R. How long did
they take to meet?
B. Open-response questions
11. Two motorists, Jack and John, are having a race in the
4. A man went to visit his friend who lived 27 km away.
athletic track. John starts first. Half an hour later, Jack
After walking at a speed of 6 km/h for 1 21 hours, he
begins to chase after him. John’s driving speed is 84
finished the remaining journey at a jogging speed of
km/h, and Jack’s driving speed is 90 km/h. How long
12 km/h and got to his friend’s house at the appointed
does Jack take to catch up with John?
time. Find his average speed for the whole journey.
12. On Sunday, the students of one class went out for a
5. Mike made a journey from City P to City Q. In the first
picnic. At first, they traveled in the crowded city for 40
hour, he covered 13 of it. In the second hour, he covered
1 minutes at a speed of 27 km/h. Then they covered 15 of
5 of the whole journey. Finally, he took 2 hours to the remaining journey at a speed of 54 km/h. Finally,
finish the remaining journey at a speed of 42 km/h.
they traveled 32 of the whole journey in one hour on the
Calculate his average speed for the whole journey.
highway. Calculate the average speed for the whole
6. Town A and Town B are 20 km apart. Mike left Town journey.
A at 8.00 a.m. and cycled at 16 km/h towards Town B.
10 minutes after Mike started, Bill cycled from Town 13. Sunday morning, Rebecca and her parents went out
B towards Town A at 15 km/h. (a) At 8.30 a.m., how to enjoy the natural scenery. On the way to the
far was Mike away from Town A and how far was Bill destination, they traveled at a slow speed of 40 km/h.
away from Town B? (b) Did they meet up at 8.30 a.m.? On the way back, they drove at a faster speed of 120
If not, how far were they apart from each other at 8.30 km/h. When they came back home, they found that
a.m.? they had been out for 2 hours. Find the average speed
for this round trip (ignoring time at the destination).
7. Mike made a journey from City P to City Q. In the first
half an hour, he covered 17 of it. In the second half an 14. On the first day of this new term, Teacher Lee went
hour he covered 13 of the remaining journey. Finally, to the bookshop to pick up the ordered textbooks. On
he took another half an hour to finish the journey at a the way to the bookshop, his speed was as slow as 24
speed of 72 km/h. Calculate his average speed for the km/h because of the heavy traffic. On the way back,
whole journey. the traffic was light, so he took only one hour. If the
average speed for the round trip was 36 km/h, find
8. Yesterday afternoon, Mike drove to the kindergarten the speed on the way back (ignoring time spent in the
to take back his son. It was raining on the way to bookshop).
the kindergarten, so he had to drive at a slow speed
of 39 km/h, and it took him 1 13 hours to get there. He
returned soon at a faster speed along the same way.
When he came back home, his wife told him that his
average speed for the round trip was 52 km/h (ignoring
the time spent in the kindergarten). Find his speed on
the way back.