JOURNAL OF APPLIED MEASUREMENT, 17(3),

Copyright© 2016

The Self-assessment Practices

of Hong Kong Secondary Students:
Findings with a New Instrument

Zi Yan
The Hong Kong Institute of Education

Self-assessment is a core skill that enables students to engage in self-regulated learning. The purpose of this
study was to examine the psychometric properties of a Self-assessment Practice Scale and to depict the charac-
teristics of self-assessment practices of Hong Kong secondary students using this newly developed instrument.
A total of 6,125 students from 10 Hong Kong secondary schools completed the survey. Both Rasch and factor
analyses revealed a two-dimension scale structure (i.e., Self-directed Feedback Seeking and Self-reflection).
The two subscales demonstrated acceptable psychometric properties and suggestions for further improvement
were proposed. The findings regarding self-assessment practices of secondary students indicated that, in general,
students were quite used to engaging in self-reflection based on available feedback, but they were less disposed
to taking the initiative and seek feedback on their own performance. Key demographic variables, e.g., gender and
age, played important roles in students’ self-assessment practices. Girls had significantly higher self-assessment
measures on both scales than did boys. Junior students had higher measures on both scales than did their senior
counterparts. Implications and directions for future research were discussed.

Requests for reprints should be sent to Zi Yan, Department of Curriculum and Instruction, The Hong Kong
Institute of Education, 10 Lo Ping Road, Tai Po, Hong Kong, e-mail: zyan@[Link].
2 Yan

Introduction feedback seeking and self-reflection, rather than

requiring accurate self-judgment.
The current international education reform
Despite the important role self-assessment
moving towards a constructivist paradigm re-
practices can play in students’ learning, a detailed
quires students to be self-regulated learners who
description of its characteristics is missing in lit-
build their own knowledge. Consequentially, an
erature. For example, which kind(s) of activities
overall aim of the school curriculum in Hong
do students undertake in their self-assessment?
Kong is to help students to learn how to learn
What is the prevalence of self-assessment prac-
and equip students with the ability and disposi-
tices? Are there any differences in self-assessment
tion to pursue life-long learning (CDC, 2001).
practices for different groups of students (e.g., by
One core skill that enables students’ engagement
gender, or age)? This study aimed to fill this gap.
in self-regulated learning is self-assessment, by
which students take responsibility for reflecting The conceptualizations and roles of self-
on and evaluating their own learning progress assessment
and/or products (Kirby and Downs, 2007; Lan-
gendyk, 2006). Compared to other assessments, In the self-assessment literature, it seems
in particular tests and examinations, which often that there are quite different conceptualizations
have negative consequences for students’ leaning and interpretations of what actually constitutes
attitudes and behaviours, self-assessment has the self-assessment. These conceptualizations could
potential to improve students’ commitment to and be summarized into three major groups. First
autonomy in learning (Paris and Paris, 2001). is that self-assessment is conceptualized as the
personal ability/skill of carefully considering the
Perhaps largely due to its name, self-
quality of one’s own work and making a summa-
assessment, the majority of research on the
tive evaluation of one’s own knowledge, skills, or
topic has focused on its assessment function.
performance (e.g., Eva and Regehr, 2008; Noonan
Unfortunately, many studies have revealed that
and Duncan, 2005). The second group views
self-assessment is inaccurate, and an unreliable
self-assessment as a type of assessment satisfying
indicator of student performance (Brown and
summative purposes (e.g., Boud, 1999; Bosma,
Harris, 2013; Ross, 2006); and that inaccuracy
Laszakovits, and Hattery, 2007). The third group
of self-assessment is somewhat a by-product of
conceptualizes self-assessment as a learning strat-
a human disposition of being unrealistic about
egy or process which promotes students’ learning
one’s own abilities (Dunning, Heath, and Suls,
(e.g., Panadero and Alonso-Tapia, 2013; Paris
2004). On the other hand, Boud (1999) argued
and Paris, 2001; Puustinen and Pulkkinen, 2001;
that self-assessment can contribute more to learn-
Zimmerman and Moylan, 2009).
ing by acting as a learning process rather than as
a substitute for other-conducted assessments. In The implications of viewing self-assessment
other words, engagement in self-assessment itself, as a personal ability/skill are limited as Eva and
rather than providing accurate self evaluations, Regehr (2008) asserted that self-assessment is
is more important because engaging in self- unlikely to be developed as a personal skill;
assessment results in learning gains by enhancing inaccuracy of self-assessment being a natural
students’ self-regulation and metacognitive skills consequence of unrealistic human nature (Dun-
(Kostons van Gog, and Paas, 2012; Panadero, ning, Heath, and Suls, 2004). This position has
Alonso-Tapia, and Huertas, 2012). Therefore, it been supported in a large number of studies (e.g.,
is self-assessment practice, rather than the scores Harris and Brown, 2010; Kuncel, Credé, and
or ratings generated by self-assessment, that Thomas, 2005; Powel and Gray, 1995).
really matters in terms of improving students’ From the perspective of assessment, Ross
performance. A realistic approach in developing (2006) concluded, in a review study, that self-
students to be self-regulated learners is to enhance assessment was a less reliable indicator of stu-
teachable self-assessment practices, such as dent performance than were other assessments
 Self-assessment Practices of Hong Kong Secondary Students 3

(e.g., test scores, teacher/parent ratings). A teachers’ written comments on the assignments,
more recent review of self-assessment (Brown marks on exam papers, etc.—are available, the
and Harris, 2013) echoed that, in general, self- self-assessment practices students conduct are
assessment was not robustly accurate and that related to self-reflection. Reflection refers to
caution is warranted in using self-assessment mental endeavor aiming at exploring and elabo-
results for accountability purposes due to its rating one’s understanding of problems encounter
inaccuracy, especially for younger or less able during learning (Mann, Gordon, and McLeod,
students. Blackwood (2010) further reported that 2009). Reflective thinking is crucial component
overconfidence in self-assessment is particularly of self-assessment process (McMillan and Hearn,
salient for Chinese students. The inaccuracy of 2008).This activity helps students to understand
self-assessment, along with its detachment from better the learning process and to identify their
reality, attenuated the status of self-assessment as strengths and weaknesses. By reflection students
a useful type of assessment (Brown and Harris, can obtain information on what they learned,
2014) since assessment practices must be valid what they need to work on, and how they can
and reliable in order to inform decision-making achieve the goal.
(Messick, 1989). However, if the information regarding stu-
Although self-assessment, as a personal skill dents’ performances are missing or the meaning
or a type of assessment, seems inherently flawed of such information is blurred, students need to
and the associated assessment value is therefore take responsibility for seeking feedback about the
diminished, that should not dismiss the possible quality of their performances (Boud, 1999; Kirby
merits of self-assessment as a key learning strat- and Downs, 2007; Langendyk, 2006). Although
egy. From a pedagogical perspective, self-assess- the most accessible feedback source is from
ment encourages students to reflect on their own students themselves, e.g., self-testing, self-rating
performances and enhances their self-regulation one’s own performance, or self-recording learning
which, in turn, leads to improved learning. In progress, Boud (1999, p. 122) argued that “the
this sense, the importance of the accuracy of self- practice of self-assessment does not imply that
assessment seems reduced as Panadero, Brown, this engagement is an isolated or individualistic
and Strijbos (2014, p. 11) argued that “While it activity. It commonly involves peers, makes use
seems appropriate that an assessment be accurate, of teachers and other practitioners and draws upon
it might well be that there are benefits from en- appropriate literature.” In his conceptualization,
gaging in self-assessment, even if it is unrealistic self-assessment practice involves a process by
or inaccurate as long as students overcome any which one takes initiative for seeking feedback
unrealism or inaccuracy.” Following this line of from various sources including teachers, peers,
reasoning, the self-assessment practices students and parents, though the role and the degree of
engage in, rather than making a final evaluation, involvement of external parties may vary. Thus,
appear crucial for learning improvement. self-directed feedback seeking is another indis-
pensable component of self-assessment practices.
The components of self-assessment practices
In this study, the focus is on self-assessment
When the need for self-assessment arises, the practices of Hong Kong secondary students.
first thing students have to do is to gather informa- There are two purposes in this study. One is to
tion that gives them evidence as to the quality of develop a scale to assess student self-assessment
their performances. According to the availability practices and to provide a preliminary examina-
of such information/feedback, the current study tion of the psychometric properties of that scale.
differentiated students’ self-assessment practices The second is to detail the characteristics of
into two classes, namely, self-directed feedback Hong Kong secondary students’ self-assessment
seeking and self-reflection. When the feedback practices. The findings should help researchers
regarding students’ performances—such as and teachers to better understand and promote
4 Yan

students’ self-assessment practices which lead to final scale (see Appendix 1). The 10-item scale,
improved learning outcomes. instead of the 16-item scale, was used because
of two reasons. First, this scale had to maintain
Methodology a minimum number of items since it was part of
The methodology section of this paper con- a survey administered in a large project. Second,
sists of two parts: the procedure for scale develop- the 6 items being removed from the final scale
ment including the construction of an item pool, received significantly less endorsement from the
and the subsequent administration of the scale. pilot group of students than the 10 items being
The research sample and data analysis methods kept. The frequencies for the 10 items in the final
used are described in the latter section. scale ranged from 48% to 88%, while the frequen-
cies for the 6 items being removed ranged only
Scale development from 8% to 12%. Among the 10 items, 7 items
are about self-directed feedback seeking and 3
Although self-assessment could be un-
items are related to self-reflection. Respondents
derstood in a subject-based context, the self-
to the scale were asked to indicate their level of
assessment practice in this study refers to those
agreement with each statement on a six-point
activities in which students tend to engage across
Likert-type scale, ranging from Strongly Disagree
different subject areas. The initial item pool was
(1), Disagree (2), Slightly Disagree (3), Slightly
built from descriptions about self-assessment
Agree (4), Agree (5) to Strongly Agree (6). Given
practices which came from two major sources: (1)
that the questionnaires were to be completed by
available instruments relevant to self-assessment
Hong Kong local students, items were developed
or self-regulated learning (e.g., Motivated Strate-
in Cantonese.
gies for Learning Questionnaire developed by
Pintrich, Smith, Garcia, and Mckeachie, 1991; the Sample
Self-directed Learning Scale developed by Mok,
Cheng, Moore, and Kennedy, 2006); as well as The survey was administered in 10 Hong
(2) consultative discussions with teachers (N = 6) Kong secondary schools that had participated in
and secondary students (N = 25). The identified a Quality Education Fund project supported by
self-assessment practices were formulated as 18 the Hong Kong government. The10 participating
statements. These statements were then subjected schools were carefully selected with a stratified
to an intensive review by a panel of three experts sampling method using school bands as the
in the field of educational assessment to examine stratum to ensure that the participating students
the face and content validity of the statements. represented a wide range of academic ability. All
Experts rated the appropriateness of each state- students in the participating schools were invited
ment in a scale (yes/no/unsure) independently. to participate in the survey. Among the 6,125
The statement was retained only if the majority respondents who completed the questionnaire,
of panel (2 or 3 experts) agreed that the item was there were 2,971 (48.5%) males, 3,074 (50.2%)
appropriate to be used to gauge self-assessment females, and 80 (1.3%) who omitted that infor-
practice. This process resulted in a 16-item scale. mation. The sample consists of students across
The expert panel was also asked to review the all secondary school year levels from S1 (983,
clarity, possible ambiguity, and potential bias of 16.0%), S2 (1092, 17.8%), S3 (1045, 17.1%),
each of the remaining items. Necessary modifi- S4 (1089, 17.8%), S5 (1206, 19.7%), to S6 (710,
cations were made. The revised 16 items were 11.6%).
then presented to another group of secondary Data analysis
students (N = 25) who were asked to select their
most frequently-used self-assessment practices. In fulfilling the first research purpose, both
The frequencies were tallied and the 10 most Rasch analysis (Rasch, 1960) and factor analy-
frequently endorsed items were included in the sis were used to investigate the psychometric
properties of the scale. For the second research
 Self-assessment Practices of Hong Kong Secondary Students 5

purpose, descriptive analyses and t-test/ANOVA item fit statistics. The standardized form of the
were applied to students’ Rasch-calibrated self- outfit mean square, i.e., Outfit ZSTD, was used
assessment measures to reveal (a) the range and as the item fit statistic in this study. In many
the distribution of the students’ responses to those studies, items were regarded as misfitting to the
items, and (b) the differences in self-assessment Rasch model if the Outfit ZSTD was larger than
practices by gender and year level. +2.0. However, a relatively lenient critical value,
With the principle that an objective measure- i.e., +3.0, could be a more meaningful criterion
ment should target only one attribute or dimension for large samples (Linacre, 2002). Given that
at one time (Bond and Fox, 2015) as a tenet, Rasch the sample used in this study was over 6,000,
analysis is used to check the extent to which items +3.0 was adopted as the critical value of Outfit
in any scale belong to a single unidimensional ZSTD for detecting misfit items. Differential Item
construct. Once the unidimensional scale is built, Functioning (DIF) across gender and year level
person ability and item difficulty can be calibrated was also checked since such key demographic
onto a single latent trait scale. Persons/items are variables were found to have substantial impact
placed on an ordered trait continuum according on self-regulated learning activities or readiness
to their ability/difficulty estimated by the Rasch in previous studies (e.g., Bidjerano, 2005; Pokay
model. Therefore the meaning of person measures and Blumenfeld, 1990; Reio, 2004; Reio and Da-
and item difficulties can be interpreted under the vis, 2005). DIF analysis aims to check construct
same framework and direct comparisons between equivalences across groups. DIF exists when an
person measures and item difficulties can be item has different levels of difficulty for certain
conducted. According to the Rasch model, the groups of subjects after controlling their levels
outcome of a person’s answer to a given item is of latent trait. As suggested by previous studies
determined by the comparison of person ability (e.g., Wang, Yao, Tsai, Wang, and Hsieh, 2006),
and item difficulty. The higher a person’s ability a difference equal to or larger than 0.5 logits was
compared to the difficulty of the item, the higher considered as substantial DIF in this study.
the probability that s/he will have a right answer Factor analysis, on the other hand, usually
to the item. Mathematically, the person ability aims to identify the underlying factor structure
measure and item difficulty measure are estimated and determine which items belong to each factor.
in the same way in the Rasch model. First, the Both exploratory factor analysis and confirmatory
ratio of each person’s percentage of correct an- factor analysis were used in this study. Explor-
swers over the percentage of incorrect answers is atory factor analysis aims to identify the number
calculated, and then that ratio is transformed into and the content of constructs underlying a set of
odds of a successful response. Finally, the natural items; while confirmatory factor analysis intends
logarithm of those odds is calculated as the person to test the fit between the empirical data and a
ability measure (or, similarly, the item difficulty pre-specified factor structure (Thompson, 2004,
measure). Through several rounds of iteration, the p. 6). In this exploratory factor analysis, a factor
raw data are transformed into interval measures loading higher than .4 was used as the criterion
(in logits) of person ability and item difficulty to select items (Nunnally, 1978). In confirmatory
(Bond and Fox, 2015). factor analysis, multiple fit indices including
With the Rasch model, a “data fit the model” chi-square (c2), Goodness of Fit Index (GFI),
approach is employed (Andrich, 2004). This ap- Adjusted Goodness of Fit Index (AGFI), com-
proach requires that the empirical data must meet parative fit index (CFI), the Tucker-Lewis-Index
specific a prior requirements for the purpose of (TLI), and the standardized root-mean-square
fundamental measurement. In current study, the error of approximation (RMSEA) were used to
criteria used to examine the data-model fit include examine the model data fit.
response category functioning, dimensionality Although these two analytical methods are
investigations, Rasch separability reliability, and built on different assumptions, using both to-
6 Yan

gether brings advantage in refining instruments ability curves (Figure 1) that the threshold calibra-
as well as providing a holistic interpretation of tions for the rating scale satisfied this requirement.
research results (Richardson, 2005). For example, In other words, a higher measure on the items
confirmatory factor analysis provides numerous represents a higher level of latent trait under
overall model fit indices, while Rasch analysis measurement. Each category had a distinct peak in
emphasizes more on the fit of each item to the the graph, indicating each category was the most
model. Some empirical studies, including a recent probable performance level for given groups of
investigation of learners in Hong Kong (Deneen, persons with a specific level of latent trait.
Brown, Bond, and Shroff, 2013) obtained slightly Putting all the 10 items together, the one-
different but complementary results from these dimension assumption was first checked. One
two analytical methods. This study attempted to self-reflection item (1: When the teachers ask the
examine whether these two analytical methods, others, I will think about the answers) showed
based on the same set of data, would generated substantial misfit to the Rasch model. The Outfit
different or similar results. ZSTD of this item (+9.9) was higher than +3.0.
Results After removing item 1, another self-reflection
item (7: I will try to figure out the reasons for the
Rasch analysis mistakes I made after receiving the marked exam
paper) appeared misfitting. The Outfit ZSTD was
Since all items in the scale shared the same +8.9. After removing items 1 and 7, the results
six-point rating scale, rating scale Rasch model of Rasch analysis showed that all remaining 8
using Winsteps 3.73 (Linacre, 2011) was items in the scale fit the unidimensional Rasch
applied to conduct the Rasch analysis. The cat- model quite well.
egory function of the six-point rating scale was The dimensionality investigations are based
first checked. If the categories function well, the on a map generated by Winsteps plotting the
threshold calibrations (the intersection points results of principal components analysis (PCA)
of adjacent category curves) of the rating scale of residuals. The results in Figure 2 revealed that
should increase monotonically (Linacre, 2006). these two misfitting items (1 and 7), together with
It can be seen from the response category prob- item 8 (I will identify the areas for improvement
Figures

6
Category Probability

4 5

3
2

Measure relative to item difficulty

Figure 1. The response category probability curves for the 6-point rating scale.
Figure 1. The response category probability curves for the 6-point
rating scale.
 Self-assessment Practices of Hong Kong Secondary Students 7

according to the teacher’s comment on my as- The Rasch analyses were then undertaken
signment), which is also a self-reflection item, separately for the two scales. The Rasch person
constituted a second dimension other than the reliabilities for these two scales were 0.89 and
Rasch dimension representing the other self- 0.78 respectively. Over half (57% and 60% re-
assessment items. The dimensionality map for this spectively) of variance in the data was explained
scale showed that the loadings of three items (1, by the Rasch measures in each of these two scales.
7, and 8) on the first contrast (.50, .76., and .64 The items measures, Outfit ZSTDs, DIF across
respectively) were substantially different from the gender and year level are presented in Table 1.
loadings of other items (e.g., the loadings of items It can be seen from Table 1 that the Outfit
2, 3, and 10 are −.34, −.45, and −.33 respectively). ZSTDs for all items in the two scales fall into the
It indicated that there were contrasting patterns in acceptable range (lower than +3.0) except that
the residuals. In other words, while other items the Outfit ZSTD for item 1 was marginal (3.2).
represent the main Rasch dimension, these three Although there were some DIF effects with sta-
items come from a separate dimension. tistical significance (p < .05) across year level, no
The results of this dimensionality investiga- DIF value exceeded 0.5 logits which is suggested
tion conformed to the 2-dimension conceptual as a cut-off value of substantive DIF according to
framework of self-assessment practices discussed Linacre (2006). Given that large sample size (e.g.,
earlier. Therefore, the 10 items were divided into N = 6,125 in this study) might flag trivial differ-
two scales. One was a Self-reflection scale of 3 ences as statistically significant, these DIF results
items (1, 7, and 8), gauging students’ self-initiated were considered as not substantively meaningful.
activities to reflect on their performances based on In sum, all the criteria including Rasch separabil-
available feedback. The other was a Self-directed ity reliability, variance explained by measures,
Feedback Seeking scale, which was comprised of item fit statistics, and DIF gave credence to the
7 items (2, 3, 4, 5, 6, 9, and 10), assessing students’ appropriateness of using this instrument as two
self-initiated activities to search for feedback on unidimensional scales with the current sample.
their performances from a variety of sources.

-1 0 1
-+---------------------------------+---------------------------------+- COUNT
.8 + | +
| A | | 1
.7 + | +
| B | | 1
C .6 + | +
O | | |
N .5 + C | + 1
T | | |
R .4 + | +
A | | |
S .3 + | +
T | | |
.2 + | +
1 | | |
.1 + | +
L | | |
O .0 +---------------------------------|---------------------------------+
A | | |
D -.1 + | D + 1
I | | |
N -.2 + | E + 1
G | | |
-.3 + | de + 2
| | bc | 2
-.4 + | +
| | a | 1
-+---------------------------------+---------------------------------+-
-1 0 1
ITEM MEASURE

Figure 2. The standardized residual contrast 1 plot.

8 Yan

Table 1
Item Measures, Outfit ZSTDs, DIF across Gender and Year level for the Two Scales.
DIF
Scale/Item Measure Outfit ZSTD Gender 1
Level2
Self-directed feedback seeking
Item #2 0.03 2.9 −0.08 (−0.16, 0.19)
Item #3 0.22 −8.9 0.00 (−0.03, 0.04)
Item #4 −0.27 −5.2 0.12 (−0.24, 0.15)
Item #5 0.10 2.5 0.10 (−0.06, 0.10)
Item #6 −0.19 −6.6 −0.10 (−0.05, 0.07)
Item #9 0.10 2.8 0.00 (−0.23, 0.19)
Item #10 0.02 −0.8 0.00 (−0.09, 0.10)

Self-reflection
Item #1 0.11 3.2 −0.02 (−0.19, 0.13)
Item #7 −0.30 −6.2 0.00 (−0.03, 0.02)
Item #8 0.19 −9.0 0.06 (−0.08, 0.16)
Note. 1
The figures in this column are the item difficulty differences, in logits, between males and
females.
2
The figures in parentheses in this column are the range of item difficulty differences, in
logits, among different year levels. For example, (−0.16, 0.19) for item 1 means that the item
difficulty differences between each year level and its average difficulty (baseline) range from
−0.16 to 0.19 logits.

Factor analysis Confirmatory factor analysis was conducted

subsequently on the second half of the sample us-
The percentage of missing data for the 10 ing AMOS 21.0 with a purpose of further assess-
items ranged from 0.1% to 12.4%. According ing how well the data fit the two-factor structure
to Allison’s (2003) recommendation, maximum that was identified from the exploratory factor
likelihood methods with the expectation maxi- analysis. The results showed that the GFI (0.930),
misation algorithm was used to impute missing CFI (0.927), and TLI (0.903) indices were all
data. The dataset was randomly split into two half: higher than 0.9 except AGFI (0.887). The chi-
one for exploratory factor analysis, the other for square (c2 = 1082.976, df = 34) was significant at
confirmatory factor analysis. Exploratory factor the 0.01 level and RMSEA was 0.100. Inspection
analysis, using maximum likelihood estimation of modification indices revealed a strong correla-
with direct oblique rotation, was first conducted tion between the residuals of item 9 and 10. In
on the first half of the sample to explore the factor a revised model, the residuals of item 9 and 10
structure of the data set. Using eigen values > 1 were allowed to correlate and re-analysis was
as the criterion, the results of analysis identified conducted (see Figure 3). The chi-square statistics
two factors. The scree test (Cattell, 1966) also decreased to 594.037, although it was still signifi-
supported that factor structure. A factor loading cant. This result is not surprising given that chi-
higher than .4 was used as the criterion to select square statistics are also quite sensitive to large
items (Nunnally, 1978). Items 2, 3, 4, 5, 6, 9, and sample size. RMSEA was improved to 0.074. All
10 loaded on factor 1 with factor loadings rang- other fit statistics were also improved including
ing from .66 to .80. Items 1, 7, and 8 loaded on GFI (0.961), AGFI (0.935), CFI (0.961) and TLI
factor 2 with factor loadings ranging from .41 to (0.947). These results showed acceptable fit be-
.93. No item demonstrated cross-loadings on both tween the empirical data and the proposed factor
factors. The factor structure was consistent with structure. The correlation between the residuals
the proposed 2-dimension conceptual framework of item 9 and 10 (r = .41) could be explained by
of self-assessment practices. the strongly-related item contents. Both item 9
 Self-assessment Practices of Hong Kong Secondary Students 9

(I will ask my group members to comment on my scale and Self-reflection scale. The Rasch model
work in group activities) and item 10 (I will invite calibrates person measures and item difficulties
others to test how well I have mastered the course along the same interval-level latent trait scale. The
contents) are about students’ willingness to invite left-hand side shows the distribution of students
feedback from others on their own performance. along the latent trait scale; while the item difficul-
Furthermore, as expected, a strong correlation ties are located by item number on the right-hand
(r = .72) was found between the two factors, i.e., side of the scale. The students with the highest
self-directed feedback seeking and self-reflection. self-assessment levels and the items with highest
difficulty levels are located at the top of the map,
Students’ self-assessment practices while the students with the lowest levels and the
The first purpose of this study was to examine easiest items are located at the bottom. The item
the psychometric properties of the scales. The response category thresholds are also presented in
positive results presented above give credence to the map by the notation of x.y (e.g., “2.5” stands
use of the scales with the current sample. There- for the 5th threshold of the item 2).
fore, the second purpose of this study—to investi- For the Self-directed feedback seeking scale
gate the characteristics of students’ responses—is (see Figure 4), the mean of students’ abilities
reported so as to contribute to understanding of (M = −0.10, SD = 1.65) was virtually identical
Hong Kong secondary students’ self-assessment to the mean difficulty of the items (M = .00,
practices. This section provides a descriptive SD = 0.16). Student ability ranged from −5.88 to
analysis of the range and the distribution of the +6.47 logits; while item difficulty ranged from
students’ self-assessment abilities; and note any −0.27 to +0.22 logits. The range of item difficulty
similarities and differences of self-assessment was much smaller than the range of student abil-
practices across gender and year level. ity. However, the ranges of difficulty levels of the
Figures 3 and 4 present the item-person vari- response categories for each item, as presented on
able maps for the Self-directed feedback seeking the right-hand side of the map, covered a wide

#2 e2

.72 #3 e3

.79
#4 e4
.79

.73
#5 e5
SDFS .76

.64 #6 e6

.67
#9 e9
.72
.41
9
#10 e10

.53 #1 e1

.73
SR #7 e7
.80

#8 e8

[Link]
Figure Resultsofofthetheconfirmatory
confirmatory factor
factor analysis
analysis onon
thethe proposed two-factor model.
proposed
two-factor model.
10 Yan

range of difficulty along the latent trait scale, logits. Similar to the Self-directed feedback seek-
and provided quite good coverage to the range ing scale, the items in the Self-reflection scale,
of student abilities. together with their response categories, covered
For the Self-reflection scale (see Figure 5), the range of student abilities.
the mean of students’ abilities (M = 0.99, SD Comparatively speaking, the items in the
= 2.06) was substantially higher than the mean Self-reflection scale are easier for these secondary
difficulty of the items (M = .00, SD = 0.22). Stu- school students to endorse: they are more likely to
dent ability ranged from −4.73 to +6.56 logits; engage in self-reflection based on available feed-
while item difficulty ranged from −0.30 to +0.19 back than to take initiative for seeking feedback

Student Item Item threshold

-------------------- ----------------- -----------------------------------
6 . +
|
|
|
. |
5 . +
|
|
. |
|
4 +
. |
|
. |
. |
3 + 3.5 5.5 9.5
.# | 2.5 10.5
. | 4.5 6.5
.#### |
.## |
2 .### +
. |
.#### | 3.4 5.4 9.4
.##### | 2.4 10.4
. | 4.4 6.4
1 .###### +
.###### |
.############ |
.####### |
. | #3 #5 #9
0 .####### M+M #2 #10 3.3 5.3 9.3
.######## | #4 #6 2.3 10.3
.####### | 4.3 6.3
.####### |
.###### |
-1 .####### + 3.2 5.2 9.2
.#### | 2.2 10.2
.### | 4.2 6.2
.### |
.### | 3.1 5.1 9.1
-2 .## + 2.1 10.1
.## | 4.1 6.1
.## |
.# |
.# |
-3 .# +
. |
. |
|
. |
-4 +
|
|
. |
|
-5 .# +
-----------------------------------------------------------------------------
EACH “#” IS 46 STUDENTS. EACH “.” IS 1 TO 45 STUDENTS

Figure 4. Item-person map for the Self-directed Feedback Seeking scale.

 Self-assessment Practices of Hong Kong Secondary Students 11

from various sources. The most endorsed item through assessing myself using reference books
was 7 (I will try to figure out the reasons for the or notes), 5 (I will keep track of my progress
mistakes I made after receiving the marked exam through the performance record), and 9 (I will
paper). The most difficult items were on the Self- ask my group members to comment on my work
directed feedback seeking scale: 3 (I often check in group activities).
whether I have fully mastered the course contents

Student Item Item threshold

-------------------- --------------- ----------------------------
6 .## +
|
|
|
|
|
5 .## +
|
|
|
. |
|
4 +
.#### |
|
|
|
|
3 . + 1.5 8.5
.############ |
. |
. | 7.5
|
|
2 +
.########### |
|
. |
.# | 1.4 8.4
|
1 +
.########### M| 7.4
|
|
. |
.############ | #1 #8
0 .## +M
|
| #7
.####### | 1.3 8.3
. |
.# |
-1 .##### + 7.3
. |
.### | 1.2 8.2
. |
.## |
| 7.2
-2 .# + 1.1 8.1
|
|
.# | 7.1
. |
. |
-3 . +
. |
|
. |
|
|
-4 . +
----------------------------------------------------------------------
EACH “#” IS 69 STUDENTS. EACH “.” IS 1 TO 68 STUDENTS

Figure 5. Item-person map for the Self-reflection scale.

12 Yan

Table 2 below presents comparisons of needs, formulating learning goals, identifying hu-
Rasch-calibrated mean student measures by man and material resources for learning, choosing
gender and year level. and implementing appropriate learning strategies,
A t-test and an ANOVA were undertaken and evaluating learning outcomes.” This study
to investigate the differences of self-assessment adopted this line of argument and further revealed
practices by gender and year level respectively. that such “initiative” in self-assessment practices
Table 2 reports a significant (p < .01) gender effect could be understood on two related, but different
on students’ responses to both scales. Females, psychometric dimensions: those being, seeking
on average, had significantly higher self-directed feedback and self-reflection.
feedback seeking and self-reflection abilities than The first purpose of this study was to examine
did males. The results of ANOVA showed that the psychometric properties of a newly developed
year level was also an important factor with sig- self-assessment practice scale for secondary stu-
nificant (p < .01) effect on students’ self-directed dents. Both Rasch and factor analyses resulted in
feedback seeking and self-reflection abilities: two scales, namely, the Self-directed Feedback
junior students (S1 and S2) reported higher self- Seeking scale containing 7 items and the Self-
assessment abilities than did senior students (S3 reflection scale consisted of 3 items. The resultant
to S6) on both scales. The results also showed two unidimensional scales satisfied the measure-
that there were no significant interaction effects ment requirements of the Rasch model, i.e., all
between gender and year level for both scales. items in a scale measure a single latent trait, and
demonstrated sufficiently good psychometric
Discussion properties. Both exploratory and confirmatory
The key feature distinguishing self-assess- factor analyses echoed this two-factor structure
ment from other kinds of assessment is that self- in students’ responses to the items.
assessment is initiated by the student her/himself. This result indicated that students’ self-as-
Boud (1999) emphasized that self-assessment is sessment practices should be understood from two
a process by which one takes the initiative for perspectives. One dimension concerns students’
seeking feedback on one’s own performance. This self-directed activities of seeking feedback on
is in line with the requirement of self-directed their own performances from different sources.
learning. According to Knowles (1975, p. 18), That feedback might come from themselves,
self-directed learning is “a process by which in- including asking themselves questions (2 and 6)
dividuals take the initiative, with or without the and record-keeping (5); from external materials
assistance of others, for diagnosing their learning including reference books or notes (3) and past

Table 2
Comparison of Student Measures across Gender and Year Level.
Self-directed feedback seeking Self-reflection
Gender
Male (N = 2971) −0.16 0.83
Female (N = 3074) −0.05 1.14
t-test results t = −2.593, df = 6043, p = .010 t = −5.771, df = 6043, p = .000

Year level
Secondary 1 (S1) (N = 983) 0.02 1.41
Secondary 2 (S2) (N = 1092) −0.04 1.03
Secondary 3 (S3) (N = 1045) −0.15 0.86
Secondary 4 (S4) (N = 1089) −0.03 0.91
Secondary 5 (S5) (N = 1206) −0.26 0.92
Secondary 6 (S6) (N = 710) −0.15 0.77
ANOVA results F(5, 6119) = 4.444, p = .000 F(5, 6119) = 11.446, p = .000
 Self-assessment Practices of Hong Kong Secondary Students 13

exam papers (4); and from other persons (9 and scale’s reliability and, more importantly, allow
10). The other dimension consists of students’ more thorough investigation into students’ self-
reflective activities including self-checking reflection activities based on different feedback
based on teachers’ oral feedback (1), figuring out provided
reasons for mistakes on marked exam papers (7) The satisfactory psychometric properties of
and identifying areas for improvement based on the two scales paved the way for the accomplish-
teachers’ comments (8). In other words, the major ment of the second purpose of this study, that is, to
activities students need to undertake to complete generate a general picture of the self-assessment
self-assessment include gathering feedback as practices of Hong Kong secondary students.
to the quality of their performance from various
sources and self-reflection. Thus, self-assessment Although it is not appropriate to compare
is not only about evaluating one’s own perfor- directly the difficulty levels of items in these two
mance according to pre-established criteria, scales since the items were calibrated in different
but also about encouraging students to draw on measurement metrics, the comparison between
their diverse personal resources to facilitate such students’ responses did provide information about
evaluation. the relative ease of endorsement of the items in
these two scales. The mean of students’ abilities
The analysis results identified directions of was similar to the mean difficulty of the items in
further improvement for the scales. Boud (1999) the Self-directed feedback seeking scale; while
argued that students’ self-assessment practices substantially higher than the mean difficulty of the
may involve feedback from various stakeholders. items in the Self-reflection scale. In other words,
From this perspective, the items included in the students found that the items in the Self-reflection
Self-directed Feedback Seeking scale appear to scale were relatively easier to endorse, and items
be not sufficiently comprehensive. For example, in the Self-directed feedback seeking scale were
parents and teachers are not specifically men- relatively difficult to endorse. This finding indi-
tioned in those items. This is partially due to the cated that students were more used to conducting
item selection procedure adopted in this study, self-reflection based on available feedback (e.g.,
i.e. the most frequently used 10 self-assessment teachers’ oral feedback and written comments) in
practices were included into the final scale. The order to identify the reasons for mistakes or the
exclusion of parents or teachers as feedback areas for improvement.
sources indicated that seeking feedback from
parents or teachers might not be popular practice Previous studies (Panadero, Alonso-Tapia,
for Hong Kong secondary students. A speculative and Huertas, 2012; Panadero, Alonso-Tapia, and
but possible explanation is that students might not Reche, 2013) revealed that use of self-assessment
see teachers to be effective in solving their prob- had positive effect on self-regulation and the
lems (Mok, Kennedy, Moore, Shan, and Leung, magnitude of the effect varied according to dif-
2008). Therefore, students may lack motivation ferent self-assessment tools. In particular, scripts
to seek feedback from teachers. However, the enhanced self-regulation more than rubrics.
inclusion of these feedback sources to provide Panadero, Alonso-Tapia, and Reche (2013) fur-
a more comprehensive picture of students’ self- ther pointed out that this is because scripts—spe-
assessment practices still remains open in further cific questions designed in steps to analyze the
development of this scale, as well as in subsequent learning process throughout a task—increased
empirical investigation. metacognitive awareness and, in turn, activated
more learning strategies. Students’ engagement
The results also indicated that more items in self-reflection has similar functions to that
need to be added into the Self-reflection scale. of scripts. By self-reflection, students analyze
The current version of the scale contains only 3 the learning process which leads to the learning
items with lower Rasch person reliability of 0.79. outcome and ask themselves specific questions in
More items would increase the Self-reflection order to identify their strengths and weaknesses.
14 Yan

Even though students’ self-assessments might not but not direct-inquiry behavior since it might
be accurate, such reflection is still a pedagogically bring too much undesired individual attention to
useful activity. Exploring their answers to “why” one-self. Mok and her colleagues (2008) further
questions might enhance metacognitive aware- pointed out that threat to self-esteem, and main-
ness and result in better understanding of both taining face probably inhibited Chinese students
the knowledge itself and the adequacy of one’s from seeking help or feedback.
own personal constructions (Eva and Regehr, This study found that key demographic
2008). In particular, these Hong Kong students variables, e.g., gender and age, played important
were willing to figure out the reasons for the roles in students’ self-assessment practices. Girls
mistakes they made after receiving their marked had significantly higher self-assessment measures
exam papers (item 7). This is a good sign of stu- on both scale than did boys, meaning that girls
dent self-regulated learning since, as Carless and generally used more cognitive and metacognitive
Lam (2014) argued, formative use of information learning strategies such as seeking feedback and
provided by testing is a potentially productive self-reflection. This finding is consistent with
way of enhancing students’ performance. Such those of previous studies (e.g., Bidjerano, 2005;
formative use is especially crucial for students Pokay and Blumenfeld, 1990). Gender differences
in an examination-dominant context, such as that in self-assessment are possibly due to differences
in Hong Kong. in self-consciousness favoring females reported
In contrast, taking the initiative to seek in literature (Alanazi, 2001; Csank and Conway,
feedback on their own performances appeared 2004). Furthermore, given that the similar nature
to be relatively more difficult for these students. of feedback seeking and help seeking, this finding
Students were particularly reluctant to check echoed the gender difference in help seeking re-
whether they had fully mastered the course vealed in previous studies. For example, Cheong,
contents through self-assessing using reference Pajares, and Oberman (2004) reported that girls
books or notes (item 3), keep track of progress perceived greater benefits from help-seeking and
through their performance records (item 5), or were more willing to seek help than did boys.
asking their group members to comment on their Nadler (1997) argued that this was probably
work in group activities (item 9). With regard to because help-seeking had less psychological cost
students’ resistance to items 3 and 5, a specula- for girls than for boys.
tive explanation might be related to the difficulty The results showed that year level (including
levels of self-assessment tasks. It seems that age) is another important factor related to student
students are used to assessing themselves with self-assessment practices. A general pattern is that
available testing questions (e.g., items 2 and 6) or junior students (S1 and S2) had higher measures
past exam paper (item 4). In contrast, conducting on both scales than did their senior counterparts
self-assessment based on reference books/notes/ (S3 to S6). This is inconsistent with some previ-
performance records appears more challenging ous studies (e.g., Reio, 2004; Reio and Davis,
since it requires specific skills and more cognitive 2005) that reported that self-regulated learning
workload to design self-assessment tasks/ques- ability and readiness increased with age. A closer
tions. As for item 9, the possible reasons could be investigation revealed that students’ scores on the
traced into the social context since socio-cultural Self-directed feedback seeking scale were high-
factors have substantial influences on students’ est at S1 and lowest at S3, S5, and S6; students’
perception and experience of assessment (Carless scores on the Self-reflection scale were highest
and Lam, 2014) and, in particular, on feedback at S1 and lowest at S3 and S6. The most likely
seeking behaviour (Morrison, Chen, and Salgado, similarity between S3 and S6 students that might
2004). As De Luque and Sommer (2000) found, account for this finding is the pressure from high-
in seeking feedback, collectivist cultures tended stakes assessment. In Hong Kong, S6 students
to prefer indirect inquiry and self-monitoring, take the most important public examination, the
 Self-assessment Practices of Hong Kong Secondary Students 15

Hong Kong Diploma in Secondary Education measurement precision by taking into account the
(HKDSE). For S3 students, although the public attenuation of correlations between dimensions
examination (Hong Kong Certificate of Education when those dimensions are calibrated separately
Examination, HKCEE) was discontinued in 2012, (Wang, et al., 2006). Second, this study relied
those students still face pressure of high-stakes solely on data from secondary students. Further
internal assessment. S3 students’ performances on research could expand the usage these two scales
internal assessments will determine the elective to primary students and/or university students to
subjects they are allowed to choose at S4. This investigate those students’ self-assessment prac-
is a decision with highly important consequence tices. Such investigations will not only provide
as students will sit for those elective subjects in further empirical evidences regarding the quality
HKDSE in S6. In this sense, it seems that students and appropriateness for use with primary students
are more likely to engage in self-assessment when or university students, but also depict the charac-
they have less pressure from high-stakes assess- teristics of self-assessment practices of students
ment. Previous studies (e.g., Yan, 2014a, 2014b; of various ages so as to better the understanding
Yan and Cheng, 2015) reported that Hong Kong of developmental factors and learning environ-
teachers are facing challenges regarding how to ment that are related to students’ self-assessment
reconcile the roles of summative and formative practices. Third, given that students’ perceptions
assessments. The present study seems to indicate and experiences of assessment are easily affected
that students have to deal with a similar dilemma. by socio-cultural factors (Carless and Lam, 2014),
The awareness of gender and year level dif- it would be worthwhile to apply the scales, with
ferences in terms of self-assessment practices has necessary modifications or additional items, on
implications to instructional design. Given that a sample from a different socio-cultural context
boys appear to be less likely to use metacogni- to see whether a self-assessment practice pattern
tive learning strategies, including self-directed different from Hong Kong students will arise. It
feedback seeking and self-reflection, in their will shed light on contextual factors relevant to
learning, teachers might need to provide more self-assessment practices from a wider socio-
encouragement and design gender-specific cultural perspective.
supportive measures so as to enhance boys’ en- Acknowledgements
gagement in self-assessment. Furthermore, the
lowest level of engagement in self-assessment This research was partially supported by
demonstrated by S3 and S6 students reminds us the Internal Research Grant (grant no. RG
of the potential conflict between self-assessment 72/2013-2014R) from the Hong Kong Institute
and high-stakes assessment. A learning-oriented of Education.
and pressure-free environment seems important
in promoting student self-assessment. References
There are three main lines of future research Alanazi, F. M. (2001). The revised self-conscious-
that could be explored. First, further development ness scale: An assessment of factor structure,
of the instrument is warranted, especially for the reliability, and gender differences in Saudi
Self-reflection scale that has only 3 items in the Arabia. Social Behavior and Personality, 29,
current version, in order to facilitate a more thor- 763-776.
ough investigation into students’ self-assessment Allison, P. D. (2003). Missing data techniques
practices. Given that students’ self-assessment for structural equation modeling. Journal of
practices comprise different but correlated Abnormal Psychology, 112, 545-557.
dimensions, as demonstrated in this study, a Andrich, D. (2004). Controversy and the Rasch
multidimensional Rasch model should be con- model: A characteristic of incompatible para-
sidered for future investigations so as to calibrate digms? Medical Care, 42, 1-16.
simultaneously both dimensions and increase the
16 Yan

Bidjerano, T. (2005, October). Gender differences Csank, P. A., and Conway, M. (2004). Engaging
in self-regulated learning. Paper presented at in self-reflection changes self-concept clarity:
the annual meeting of the Northeastern Educa- On differences between women and men, and
tional Research Association, Kerhonkson, NY. low-and high-clarity individuals. Sex Roles,
Blackwood, T. (2010). Metaknowledge in higher 50, 469-480.
education: Self-assessment accuracy and Curriculum Development Council. (2001). Learn-
its association with academic achievement. ing to Learn—The Way Forward in Curricu-
Doctoral dissertation. Available at http:// lum. Hong Kong: Author. Available at http://
[Link]/2233/1/blackwood. [Link]/en/curriculum-development/
tony_dba.pdf cs-curriculum-doc-report/wf-in-cur/index.
Bond, T. G., and Fox, C. M. (2015). Applying html
the Rasch model: Fundamental measurement De Luque, M. F. S., and Sommer, S. M. (2000).
in the human sciences (3rd ed.). New York, The impact of culture on feedback-seeking be-
NY: Routledge. havior: An integrated model and propositions.
Bosma, J., Laszakovits, D., and Hattery, R. Academy of Management Review, 25, 829-849.
(2007). Self-assessment for maintenance of Deneen, C., Brown, G. T. L., Bond, T. G., and
certification. Journal of the American College Shroff, R. (2013). Understanding outcome-
of Radiology, 4, 45-52. based education changes in teacher education:
Boud, D. (1999). Avoiding the traps: Seeking Evaluation of a new instrument with prelimi-
good practice in the use of self assessment and nary findings. Asia-Pacific Journal of Teacher
reflection in professional courses. Social Work Education, 41, 441-456.
Education, 18, 121-132. Dunning, D., Heath, C., and Suls, J. M. (2004).
Brown, G. T. L., and Harris, L. R. (2013). Student Flawed self-assessment: Implications for
self-assessment. In J. H. McMillan (Eds.), The health, education, and the workplace. Psy-
SAGE handbook of research on classroom chological Science in the Public Interest, 5,
assessment (pp. 367-393). Thousand Oaks, 69-106.
CA: Sage. Eva, K. W., and Regehr, G. (2008). “I’ll never
Brown, G. T. L., and Harris, L. R. (2014). The fu- play professional football” and other fallacies
ture of self-assessment in classroom practice: of self-assessment. Journal of Continuing Edu-
Reframing self-assessment as a core compe- cation in the Health Professions, 28, 14-19.
tency. Frontline Learning Research, 3, 22-30. Harris, L. R., and Brown, G. T. L. (2010, May).
Carless, D., and Lam, R. (2014). Develop- “My teacher’s judgment matters more than
ing assessment for productive learning in mine:” Comparing teacher and student
Confucian-influenced settings: Potentials and perspectives on self-assessment practices in
challenges. In C. Wyatt-Smith, V. Klenowski, the classroom. Paper presented at the An-
and P. Colbert, (Eds.), Designing assessment nual Meeting of the American Educational
for quality learning (pp. 167-182). Dordrecht, Research Association, Denver, CO.
Netherlands: Springer. Kirby, N. F., and Downs, C. T. (2007). Self-
Cattell, R. B. (1966). The scree test for the number assessment and the disadvantaged student:
of factors. Multivariate Behavioral Research, Potential for encouraging self-regulated
1, 245-276. learning? Assessment & Evaluation in Higher
Education, 32, 475-494.
Cheong, Y. F., Pajares, F., and Oberman, P. S.
(2004). Motivation and academic help-seeking Knowles, M. (1975). Self-directed learning: A
in high school computer science. Computer guide for learners and teachers. San Fran-
Science Education, 14, 3-19. cisco, CA: Jossey-Bass.
 Self-assessment Practices of Hong Kong Secondary Students 17

Kostons, D., van Gog, T., and Paas, F. (2012). Evaluation and Research in Education, 21,
Training self-assessment and task-selection 188-213.
skills: A cognitive approach to improving self- Morrison, E. W., Chen, Y. R., and Salgado, S.
regulated learning. Learning and Instruction, R. (2004). Cultural differences in newcomer
22, 121-132. feedback seeking: A comparison of the United
Kuncel, N. R., Credé, M., and Thomas, L. L. States and Hong Kong. Applied Psychology,
(2005). The validity of self-reported grade 53, 1-22.
point averages, class ranks, and test scores: Nadler, A. (1997). Autonomous and dependent
A meta-analysis and review of the literature. help seeking: Personality characteristics and
Review of Educational Research, 75, 63-82. the seeking of help. In B. Sarason, I. Sarason,
Langendyk, V. (2006). Not knowing that they do and R. G. Pierce, (Eds.), Handbook of person-
not know: Self-assessment accuracy of third- ality and social support (pp. 258-302). New
year medical students. Medical Education, York, NY: Plenum Press.
40, 173-179. Noonan, B., and Duncan, C. R. (2005). Peer
Linacre, J. M. (2002). What do infit and outfit, and self-assessment in high schools. Practi-
mean-square and standardized mean? Rasch cal Assessment, Research and Evaluation,
Measurement Transactions, 16, 878. 10(17), 1-8.
Linacre, J. M. (2006). A user‘s guide to WIN- Nunnally, J. (1978). Psychometric theory (2nd
STEPS/MINISTEP: Rasch-model computer ed.). New York, NY: MacGraw-Hill.
programs. Chicago, IL: [Link]. Panadero, E., and Alonso-Tapia, J. (2013). Self-
Linacre, J. M. (2011). Winsteps (Version assessment: Theoretical and practical conno-
3.73) [Computer software]. Beaverton, OR: tations. When it happens, how is it acquired
[Link]. and what to do to develop it in our students.
Mann, K., Gordon, J., and McLeod, A. (2009). Electronic Journal of Research in Educational
Reflection and reflective practice in health Psychology, 11, 551-576.
professions education: A systematic review. Panadero, E., Alonso-Tapia, J., and Huertas, J.
Advances in Health Sciences Education, 14, A. (2012). Rubrics and self-assessment scripts
595-621. effects on self-regulation, learning and self-
McMillan, J. H., and Hearn, J. (2008). Student efficacy in secondary education. Learning and
self-assessment: The key to stronger student Individual Differences, 22, 806-813.
motivation and higher achievement. Educa- Panadero, E., Alonso-Tapia, J., and Reche, E.
tional Horizons, 87, 40-49. (2013). Rubrics vs. self-assessment scripts
Messick, S. (1989). Validity. In R. L. Linn (Eds.), effect on self-regulation, performance and
Educational Measurement (pp. 13-103). New self-efficacy in pre-service teachers. Studies in
York, NY: MacMillan. Educational Evaluation, 39, 125-132.
Mok, M. M. C., Cheng, Y. C., Moore, P. J., and Panadero, E., Brown, G. T. L., and Strijbos, J-W.
Kennedy, K. J. (2006). The development and (2014). The future of student self-assessment:
validation of the self-directed learning scale Known unknowns and probable directions.
(SLS). Journal of Applied Measurement, 7, Paper presented at the biennial conference of
418-449. the Assessment and Evaluation SIG, Madrid,
Spain.
Mok, M. M. C., Kennedy, K., Moore, P., Shan, P.,
and Leung, S. (2008). The use of help-seeking Paris, S. G., and Paris, A. H. (2001). Classroom
by Chinese secondary school students: Chal- applications of research on self-regulated
lenging the myth of “the Chinese learner.” learning. Educational Psychologist, 36, 89-
101.
18 Yan

Pintrich, P. R., Smith, D. A., Garcia, T., and Mck- Ross, J. A. (2006). The reliability, validity, and
eachie, W. J. (1991). A manual for the use of utility of self-assessment. Practical Assess-
the motivated strategies for learning question- ment Research & Evaluation, 11, 1-13.
naire (MSLQ). Ann Arbor, MI: National Center Thompson, B. (2004). Exploratory and confirma-
for Research to Improve Postsecondary teach- tory factor analysis: Understanding concepts
ing and Learning, University of Michigan. and applications. Washington, DC: American
Pokay, P., and Blumenfeld, P. C. (1990). Predict- Psychological Association.
ing achievement early and late in the semester: Wang, W. C., Yao, G., Tsai, Y. J., Wang, J. D.,
The role of motivation and use of learning and Hsieh, C. L. (2006). Validating, improving
strategies. Journal of Educational Psychol- reliability, and estimating correlation of the
ogy, 82, 41-50. four subscales in the WHOQOL-BREF using
Powel, W. D., and Gray, R. (1995). Improving multidimensional Rasch analysis. Quality of
performance predictions by collaboration with Life Research, 15, 607-620.
peers and rewarding accuracy. Child Study Wilson, M. (2005). Constructing measures: An
Journal, 25, 141-154. item response modeling approach. Mahwah,
Puustinen, M., and Pulkkinen, L. (2001). Models NJ: Lawrence Erlbaum.
of self-regulated learning: A review. Scandi- Yan, Z. (2014a). Predicting teachers’ intentions
navian Journal of Educational Research, 45, to implement school-based assessment using
269-286. the Theory of Planned Behaviour. Educational
Rasch, G. (1960). Probabilistic models for some Research and Evaluation, 20, 83-97.
intelligence and attainment tests. Copenhagen, Yan, Z. (2014b). School-based assessment in
Denmark: Danish Institute for Educational secondary schools. In C. Marsh, and J. C-K
Research. (Expanded edition, 1980. Chicago, Lee, (Eds.), Asia’s high performing education
IL: University of Chicago Press.) systems: The case of Hong Kong (pp. 274-
Reio, T. G., Jr. (2004). Prior knowledge, self- 288). New York, NY: Routledge.
directed learning, and curiosity: Antecedents to Yan, Z., and Cheng, E. C. K. (2015). Primary
classroom learning performance. International teachers’ attitudes, intentions and practices
Journal of Self-Directed Learning, 1, 18-25. regarding formative assessment. Teaching and
Reio, T. G., and Davis, W. (2005). Age and gender Teacher Education, 45, 128-136.
differences in self-directed learning readiness: Zimmerman, B. J., and Moylan, A. R. (2009).
A developmental perspective. International Self-regulation: Where metacognition and
Journal of Self-Directed Learning, 2, 40-49. motivation intersect. In D. J. Hacker, J. Dun-
Richardson, J. T. E. (2005). Instruments for losky, and A. C. Graesser, (Eds.), Handbook
obtaining student feedback: A review of the of Metacognition in Education (pp. 299-315).
literature. Assessment & Evaluation in Higher New York, NY: Routledge.
Education, 30, 387-415.
 Self-assessment Practices of Hong Kong Secondary Students 19

Appendix 1. Self-assessment Practice Scale

Self-directed Feedback Seeking Scale

#2 I will find some questions to test my understanding after each lesson.
#3 I often check whether I have fully mastered the course contents through assessing myself using
reference books or notes.
#4 I will check whether I have fully understood course contents by doing past exam papers.
#5 I will keep track of my progress through the performance records.
#6 I will ask myself questions to check whether I have understood the course contents.
#9 I will ask my group members to comment on my work in group activities.
#10 I will invite others to test how well I have mastered the course contents.
Self-reflection Scale
#1 When the teachers ask the others, I will think about the answers.
#7 I will try to figure out the reasons for the mistakes I made after receiving the marked exam
papers.
#8 I will identify areas for improvement according to the teacher’s comments on my assign-
ment.
Note: These statements are a translation, following the translation-back translation process, of the
items which were administered in Cantonese.