Student and feedback: Which type of feedback is preferable?

Abstract

This research: 1) made a comparison of genetic problem-solving ability among participants with different ability levels and different types of feedback and 2) studied the interaction of participant ability level and type of feedback with genetic problem-solving ability. Participants were 786 twelfth-grade students in the first semester of the 2017 academic year (May 2017–September 2017) from 7 schools in the Bangkok educational service area. The results revealed that: 1) the excellent group had the highest ability level (M = 1.006, SD = 0.411); 2) in the moderate group (M = 0.497, SD = 0.452) and poor group (M = −0.595, SD = 0.735) 2), participant ability level and type of feedback interacted with genetic problem-solving ability (F = 9.200, p = .000); and 3) simplified directive feedback was appropriate for the poor group because of their limited basic knowledge while the moderate and excellent groups who were equipped with better basic knowledge and comprehensive skills did well with worked example feedback.

Full text

Kasetsart Journal of Social Sciences 41 (2020) 269–274
Forecasting equilibrium quantity and price on the world
natural rubber market
Suratwadee Arunwarakorn
a
,
*
, Kamonchanok Suthiwartnarueput
b
,
Pongsa Pornchaiwiseskul
c
,
1
a
Logistics and Supply Chain Management, Graduate School, Chulalongkorn University, Bangkok 10330, Thailand
b
Department of Commerce, Chulalongkorn Business School, Chulalongkorn University, Bangkok 10330, Thailand
c
Department of Economics, Faculty of Economics, Chulalongkorn University, Bangkok 10330, Thailand
Article Info
Article history:
Received 27 March 2017
Revised 2 July 2017
Accepted 25 July 2017
Available online
1
September
2017
Keywords:
equilibrium price,
equilibrium quantity,
natural rubber,
simultaneous equation
Abstract
Natural rubber on the world market has had small increases in demand and big increases in
supply. Therefore, demand and supply are imbalanced and this impacts the natural rubber
price of the world market causing a decline. This study aimed: (1) to develop de-mand and
supply models to predict the world natural rubber quantity using simultaneous equations;
(2) to predict all explanatory variables in the demand and supply models using the simple
moving average technique; and (3) to estimate the equilibrium quantity and price for world
natural rubber during 2017e2026. First, in the demand model, there was a positive
relationship of the explanatory variables of world natural rubber production quantity,
synthetic rubber price, percentage year of year (%YOY) of gross domestic product (GDP),
and the exchange rate, while the negative relationship variable was natural rubber price. In
the supply model, the positive relationship variables were natural rubber price, mature
area, rainfall, and crude oil price, while the negative relationship variables were world
natural rubber stock and urea price. Second, the predicted variables indicated that
production, %YOY of GDP, exchange rate, amount of stock, and the mature area tended to
gradually increase, while the synthetic rubber price, urea price, rainfall, and crude oil price
tended to slowly decrease from 2017 to 2026. Finally, the equilibrium quantity forecast
tended to gradually increase from 953.75 to 95 7.15 thousand tonnes, and the equilibrium
price tended to fluctuate and decrease from 169.78 to 162.05 thousand yen from 2017 to
2026. Consequently, this study may be helpful to the governments of the world's impor-
tant natural rubber producing countries to plan policies to reduce natural rubber pro-
duction costs and stabilize the natural rubber price in the future, such as by setting suitable
areas of world natural rubber plantation in each country, and defining appropriate and
sustainable alternative crop areas in each country.
© 2017 Kasetsart University. Publishing services by Elsevier B.V.
Introduction
The natural rubber market of the world is primarily
concentrated in China, Europe, India, USA, and Japan,
respectively, which were the top five countries of natural
rubber consumption in 2015 (International Rubber Study
*Corresponding author.
E-mail addresses: sarunwarakorn@gmail.com, fagrsda@ku.ac.th
(S. Arunwarakorn).
Peer review under responsibility of Kasetsart University.
1
Co-first authors.
Kasetsart Journal of Social Sciences
journal homepage: http://kjss.kasetsart.org
http://dx.doi.org/10.1016/j.kjss.2017.07.013
2452-3151/© 2017 Kasetsart University. Publishing services by Elsevier B.V.
Kasetsart Journal of Social Sciences 40 (2019) 1e9
Student and feedback: Which type of feedback is preferable?
Chutaphon Masantiah, Shotiga Pasiphol
*
, Kamonwan Tangdhanakanond
Faculty of Education, Chulalongkorn University, Bangkok 10330, Thailand
article info
Article history:
Received 22 April 2018
Revised 20 June 2018
Accepted 31 July 2018
Available online 25 September 2018
Keywords:
assessment as learning,
computer-based testing,
genetic problem-solving ability,
immediate feedback,
student ability level
ABSTRACT
This research: 1) made a comparison of genetic problem-solving ability among participants
with different ability levels and different types of feedback and 2) studied the interaction
of participant ability level and type of feedback with genetic problem-solving ability.
Participants were 786 twelfth-grade students in the first semester of the 2017 academic
year (May 2017eSeptember 2017) from 7 schools in the Bangkok educational service
area. The results revealed that: 1) the excellent group had the highest ability level
(M ¼1.006, SD ¼0.411); 2) in the moderate group (M ¼0.497, SD ¼0.452) and poor group
(M ¼�0.595, SD ¼0.735) 2), participant ability level and type of feedback interacted with
genetic problem-solving ability (F ¼9.200, p¼.00 0); and 3) simplified directive feedback
was appropriate for the poor group because of their limited basic knowledge while the
moderate and excellent groups who were equipped with better basic knowledge and
comprehensive skills did well with worked example feedback.
©2018 Kasetsart University. Publishing services by Elsevier B.V.
Introduction
Assessment as learning (AAL) in the educational context
is now a popular trend. AAL is advantageous over both
assessment of learning (AOL) and assessment for learning
(AFL). Both AOL and AFL only provide knowledge of results
to students while AAL informs the student's own strength
and weakness via feedback for further development
(Srichot, 2013).
Based on the AAL definition, students play an important
role in self-evaluation to find out their own strengths and
weaknesses, so feedback is the way to enhance the students’
self-evaluated accuracy. Feedback can be provided by
various sources, for example, instructor, classmate, parents,
and even the individual. Certainly, feedback can be provided
by various methods (oral presentation, and paper-based
instruction). Thus, feedback influences students differently,
based on the source of feedback (Klaimanee, 2015; Yastibas
&Yastibas, 2015; van der Kleij, Eggen, Timmers, &
Veldkamp, 2012). Technology has made substantial leaps
forward so that a computer-based system with immediate
feedback is now one of the most effective ways to provide
feedback. A computer-based system has many advantages,
for example, it is easier to manage, gains more attention
from students, and test result can be informed instantly once
the test has been completed (Attali, 2011). Zhang, Zhang,
Luo, and Geng (2016) studied the effectiveness of immedi-
ate feedback and its relationship to feedback and memory
strategies. Their results revealed that the memories of both
adults and adolescents were significantly better when
equipped with immediate feedback. In contrast, adults were
more vulnerable to false memories when there was no
immediate feedback and showed less learning effectiveness
compared to adolescents. As a result, it can be concluded
that immediate feedback is helpful for the learning and
memory strategies of everyone.
Iron (2008, as cited in Lumthong, 2010) stated that
delayed feedback was definitely ineffective; thus, delayed
feedback was equivalent tono feedback. Furthermore, most
of the feedback in classrooms was delayed feedback as well
*Corresponding author.
E-mail address: pallusathenaz@gmail.com (S. Pasiphol).
Peer review under responsibility of Kasetsart University.
https://doi.org/10.1016/j.kjss.2018.07.020
2452-3151/©2018 Kasetsart University. Publishing services by Elsevier B.V.
Abstract
Keywords:
Article Info

C. Masantiah et al. / Kasetsart Journal of Social Sciences 41 (2020) 269–274
270
as providing only knowledge of results to the students.
The instructor may need to put in a lot of effort to employ
immediate feedback in a classroom environment, so testing
a system with immediate feedback seems to be worthwhile
for best practice of immediate feedback to determine
effectiveness of the type of feedback among students with
various ability levels.
Literature Review
Most academic studies related to testing systems with
immediate feedback are in the field of student's mathe-
matical ability assessment and only few are available in
other fields. However, a testing system with immediate
feedback should be applied to other fields of scientific study
employing mathematical approaches, such as: rectilinear
motion in Physics, solution concentration in Chemistry, and
genetic problems in Biology. Based on the literature review,
many students were unable to solve genetic problems
(Arunpruksakul, 2016). Additionally, Viraphotchananan
(2014) found that 44.12 percent of students were unable to
calculate the genotype in a genetic problem while 8.82
percent of students were able to do the calculation but did
it incorrectly. Genotype calculation is crucial in genetic
problems, especially for the Mendelian laws of inheritance
and beyond (incomplete dominance, and co-dominance).
The current research could not identify any academic
study regarding a testing system with immediate feedback
for Biology (Genetics) so it would be advantageous to begin
with a testing system with immediate feedback for genetic
probability problems to determine its effectiveness.
Relevant Theory
Bloom's Taxonomy is a cognitive process dimension
representing a hierarchy of cognitive complexity. The
cognitive domain of Bloom's taxonomy comprises 6 tiers:
1) remember, 2) understand, 3) apply, 4) analyze, 5) eval-
uate, and 6) create. Bloom's taxonomy is used as a tradi-
tional model in educational assessment. On the other hand,
the RISE model of Wray (2013) is a relatively new cognitive
model of meaningful feedback that aligns to the cognitive
domain of Bloom's taxonomy. The RISE model also has a
hierarchical structure with 4 tiers: 1) reflect (R), 2) inquire
(I), 3) Suggest (S), and 4) Evaluate (E). Information on each
tier of the RISE model can be seen in Table 1.
Type of Feedback and Motivation
Positive feedback is narration or description of positive
points of view on personal behavior to encourage students
to exhibit a desired behavior. Certainly, positive feedback
needs to be consistent, especially in the first phase of
behavior modification. Otherwise, the behavior modifica-
tion may fail.
Negative feedback is the opposite of positive feedback,
so negative feedback is attacking criticism without
providing any solution or clarification. Occasionally, nega-
tive feedback may be considered as humiliation because it
leads students to feel ashamed or results in a decline in
self-confidence. Undesired behavior will be consistent if
there is no information but only criticism provided to
students (Musikthong &Lekdamrongkul, 2013).
Type of Feedback Based on Response Time
Delayed feedback is feedback provided after awhole test
or a whole set of behaviors has finished (Sinhaa &Glassa,
2015 ). Iron (2008, as cited in Lumthong, 2010) stated that
delayed feedback was definitely ineffective; thus, delayed
feedback was considered the equivalent to no feedback.
In contrast, immediate feedback is feedback provided
instantlyafter the desired behavior, for example, testing, oral
presentation (Sinhaa &Glassa, 2015). Immediate feedback
provides information to students once the desired behavior
has finished so they can know their own strengths and
weaknesses. As a result, student can make proper progress.
Type of Feedback Based on Sources
Feedback can be divided into 4 sources: 1) instructor, 2)
friend, 3) parents, and 4) computer-based system. Yastibas
and Yastibas (2015) reported that friend's feedback for a
writing session greatly reinforced self-confidence and
diminished anxiety. Diab (2015) also reported that stu-
dent's self-evaluation feedback was noticeably inaccurate
compared to the instructor's evaluation feedback.
Computer-based feedback was frequently found in the
testing system. Most testing systems used the multiple-try
feedback/answer-until correct condition (Attali, 2015).
Student Ability Level
Student ability level can be measured by many methods,
including scoring-based systems, and the analysis-via-item
response theory model. Measurement of student ability
using the item response theory model is advantageous over
scoring-based systems as student ability does not depend
only on the test item attribute (Kanjanawasee, 2014). A
1-parameter model (Rasch model) was employed for this
research because it best fitted approximately 200 samples
(Chang, 2001; Foley, 2010). The logistic function of a
1-parameter model (Rasch model) can be defined using the
following equation:
Pið
q
Þ¼ eð
q
n�bjÞ
1þeð
q
jn�bjÞ
P
i
(
q
)¼probability of student with ability (
q
) will
respond to test item “i”correctly
Table 1
Four hierarchical tiers of RISE model
4 Elevate Raise to a higher degree or
purpose in future iteration
3 Suggest Introduce ideas for improvement
of current iteration
2 Inquire Seek information and/or provide
ideas through questioning
1 Reflect Recall, ponder, and articulate
Source:Wray (2013)
e

C. Masantiah et al. / Kasetsart Journal of Social Sciences 41 (2020) 269–274 271
as providing only knowledge of results to the students.
The instructor may need to put in a lot of effort to employ
immediate feedback in a classroom environment, so testing
a system with immediate feedback seems to be worthwhile
for best practice of immediate feedback to determine
effectiveness of the type of feedback among students with
various ability levels.
Literature Review
Most academic studies related to testing systems with
immediate feedback are in the field of student's mathe-
matical ability assessment and only few are available in
other fields. However, a testing system with immediate
feedback should be applied to other fields of scientific study
employing mathematical approaches, such as: rectilinear
motion in Physics, solution concentration in Chemistry, and
genetic problems in Biology. Based on the literature review,
many students were unable to solve genetic problems
(Arunpruksakul, 2016). Additionally, Viraphotchananan
(2014) found that 44.12 percent of students were unable to
calculate the genotype in a genetic problem while 8.82
percent of students were able to do the calculation but did
it incorrectly. Genotype calculation is crucial in genetic
problems, especially for the Mendelian laws of inheritance
and beyond (incomplete dominance, and co-dominance).
The current research could not identify any academic
study regarding a testing system with immediate feedback
for Biology (Genetics) so it would be advantageous to begin
with a testing system with immediate feedback for genetic
probability problems to determine its effectiveness.
Relevant Theory
Bloom's Taxonomy is a cognitive process dimension
representing a hierarchy of cognitive complexity. The
cognitive domain of Bloom's taxonomy comprises 6 tiers:
1) remember, 2) understand, 3) apply, 4) analyze, 5) eval-
uate, and 6) create. Bloom's taxonomy is used as a tradi-
tional model in educational assessment. On the other hand,
the RISE model of Wray (2013) is a relatively new cognitive
model of meaningful feedback that aligns to the cognitive
domain of Bloom's taxonomy. The RISE model also has a
hierarchical structure with 4 tiers: 1) reflect (R), 2) inquire
(I), 3) Suggest (S), and 4) Evaluate (E). Information on each
tier of the RISE model can be seen in Table 1.
Type of Feedback and Motivation
Positive feedback is narration or description of positive
points of view on personal behavior to encourage students
to exhibit a desired behavior. Certainly, positive feedback
needs to be consistent, especially in the first phase of
behavior modification. Otherwise, the behavior modifica-
tion may fail.
Negative feedback is the opposite of positive feedback,
so negative feedback is attacking criticism without
providing any solution or clarification. Occasionally, nega-
tive feedback may be considered as humiliation because it
leads students to feel ashamed or results in a decline in
self-confidence. Undesired behavior will be consistent if
there is no information but only criticism provided to
students (Musikthong &Lekdamrongkul, 2013).
Type of Feedback Based on Response Time
Delayed feedback is feedback provided after awhole test
or a whole set of behaviors has finished (Sinhaa &Glassa,
2015 ). Iron (2008, as cited in Lumthong, 2010) stated that
delayed feedback was definitely ineffective; thus, delayed
feedback was considered the equivalent to no feedback.
In contrast, immediate feedback is feedback provided
instantlyafter the desired behavior, for example, testing, oral
presentation (Sinhaa &Glassa, 2015). Immediate feedback
provides information to students once the desired behavior
has finished so they can know their own strengths and
weaknesses. As a result, student can make proper progress.
Type of Feedback Based on Sources
Feedback can be divided into 4 sources: 1) instructor, 2)
friend, 3) parents, and 4) computer-based system. Yastibas
and Yastibas (2015) reported that friend's feedback for a
writing session greatly reinforced self-confidence and
diminished anxiety. Diab (2015) also reported that stu-
dent's self-evaluation feedback was noticeably inaccurate
compared to the instructor's evaluation feedback.
Computer-based feedback was frequently found in the
testing system. Most testing systems used the multiple-try
feedback/answer-until correct condition (Attali, 2015).
Student Ability Level
Student ability level can be measured by many methods,
including scoring-based systems, and the analysis-via-item
response theory model. Measurement of student ability
using the item response theory model is advantageous over
scoring-based systems as student ability does not depend
only on the test item attribute (Kanjanawasee, 2014). A
1-parameter model (Rasch model) was employed for this
research because it best fitted approximately 200 samples
(Chang, 2001; Foley, 2010). The logistic function of a
1-parameter model (Rasch model) can be defined using the
following equation:
Pið
q
Þ¼ eð
q
n�bjÞ
1þeð
q
jn�bjÞ
P
i
(
q
)¼probability of student with ability (
q
) will
respond to test item “i”correctly
Table 1
Four hierarchical tiers of RISE model
4 Elevate Raise to a higher degree or
purpose in future iteration
3 Suggest Introduce ideas for improvement
of current iteration
2 Inquire Seek information and/or provide
ideas through questioning
1 Reflect Recall, ponder, and articulate
Source:Wray (2013)
e
b
j
¼difficulty parameter of test item “i”presenting ICC
point is at
q
point where probability of correct response is
at 0.50
e¼2.718
b
j
¼parameter that changes according to each test item
attribute
a
i
¼fixed parameter
c
i
¼0
Theoretical Model
Based on the literature review, there are many types
of feedback and the type of feedback can affect students
differently (Klaimanee, 2015; Yastibas &Yastibas, 2015;
van der Kleij et al., 2012). Not only the type of feed-
back, but the student ability level is a crucial factor. Shute
(2008) stated that knowledge of result feedback and
directive feedback were sufficient for high ability-level
students but there was no related study for moderate
and low ability-level students. As a result, this research
aimed to make a comparative study of student ability
level, genetic problem-solving ability, and the type of
feedback (see Figure 1).
This research aimed: 1) to compare genetic problem-
solving ability among participants with different ability
levels and different types of feedback and 2) to study the
interaction of participant ability level and type of feedback
with genetic problem-solving ability.
Methods
Selection of School and Sample
The population was 35,708 twelfth-grade students in
the 2017 academic year, from 119 schools in the Bangkok
educational service area.
Initially, samples were acquired from four different
types of school categorized by the number of students: 1)
small, a school with less than 500 students; 2) medium, a
school with 500e1,499 students, 3) large, a school with
1,500e2,499 students, and 4) extra-large, a school with
greater than 2,500 students. Then, samples were acquired
from mathematics-science programme students. This point
was crucial to ensure that students possessed some basic
knowledge of Biology so the students could deal with the
testing system properly.
Groupings and Sample Size
Based on a 1-parameter model, samples were 786 twelfth-
grade students from seven schools in the Bangkok educational
service area. Subsequently, the 786 students were distributed
into 242 students for the excellent group, 309 students for
the moderate group, and 235 students for the poor group.
Instruments
1) Pretest of genetic problem-solving ability
The pretest was a mixed-format test (constructed-
response item and multiple-choice item). The 16 items were
divided into 8 constructed-response items and 8 multiple-
choice items. The scoring system was dichotomous (0, 1)
with a total score of 16 points. The test comprised two
contents (basic knowledge of genetics and Mendelian laws
of inheritance). The test reliability was analyzed using
the IRT 1-parameter and reliability was 0.757. Finally, the
participants were divided into three groups (excellent,
moderate, and poor) based on norm-referenced criteria.
2) Testing system with immediate feedback
The testing system was a mixed-format test with 20 test
items distributed into 10 multiple-choice items and 10
constructed-response items. The testing system also shared
the same contents as the pretest (basic knowledge of ge-
netics and Mendelian laws of inheritance). Participants
were allowed to answer until correct under the specific
conditions following. Participants were given 4 points if
they achieved the correct response at the first attempt then
3 points at the second attempt and 2 points at the third
attempt and 1 point at the fourth attempt, with 0 points at
the fifth attempt even if they achieved the correct response.
A total score of the testing system was 80 points. Testing
reliability was analyzed using the Graded Response Model
(GRM) and testing reliability was 0.739. Five types of
feedback were employed in the testing system, with all of
them designed by application of the RISE model of Wray
(2013). Information on the types of feedback employed in
the testing system can be seen in Table 2.
3) Post-test of genetic problem-solving ability
The post-test was also a mixed-format test and shared the
same contents which were Mendelian's laws of inheritance
and basic knowledge of genetics. The post-test comprised 8
items distributed into 4 for multiple-choice items (5 choices)
and 4 for constructed-response items. The scoring system
was dichotomous (0, 1) with a total score of 8 points.
The reliability of the post-test was analyzed using the IRT
1-parameter model and resulted in a score of 0.756.
Design
1) Five types of feedback were employed in testing system.
All of them were designed by application of the RISE
model (Wray, 2013).
Student ability level
- Excellent
- Moderate
- Poor
Type of feedback
ApplicaƟon of RISE model
- Full direcƟve feedback
- ParƟal direcƟve feedback
- Full worked example feedback
- ParƟal worked example feedback
- Knowledge of results feedback
GeneƟc problem-
solving ability
Figure 1 Theoretical model

C. Masantiah et al. / Kasetsart Journal of Social Sciences 41 (2020) 269–274
272
2) The testing system was developed using Adobe Flash
(Adobe Inc., San Jose, CA, USA). The testing system was
online based with an administrator.
3) Participants were distributed into three groups based on
their performance in the pretest. The excellent group
consisted of 242 participants, the moderate group con-
sisted of 309 participants, and finally, the poor group
consisted of 235 participants. Importantly, each partici-
pant was provided with only one type of feedback for a
whole test. The distribution of each type of feedback for
all participants was: 1) FWF for 67 participants, 2) PWF
for 105 participants, 3) FDF for 95 participants, 4) PDFfor
68 participants, and 5) KORF for 68 participants.
4) Testing was held in a computer room of each school. Each
participant was provided with an instruction manual, a
registration code, a username and password, and note-
paper. Once the instruction session had finished, each
participant eventually logged into the testing system.
The testing period was 90 min.
Data Analysis
1) Pretest and post-test results of genetic problem-solving
ability were analyzed using descriptive statistics (mean
and SD).
2) Reliability of the test was conducted by application of
the IRL 1-parameter model via MULTILOG (SSI Inc.,
Skokie, IL, USA). The quality of the test was analyzed
using difficulty (b) and a parameter of each participant's
ability level (
q
).
3) Two-way ANOVA was used to compare the means of ge-
netic problem-solving ability and types of feedback using
the SPSS software package (SPSS Inc., Chicago, IL, USA).
Results
1) Genetic problem-solving ability of participants and type
of feedback
The poor group's post-test score mean was noticeably
low (M ¼2.35, SD ¼2.033) and the poor group's ability
level (
q
) mean was the lowest (M(
q
)¼�0.595, SD ¼0.735).
The moderate group's post-test score mean was mediocre
(M ¼5.43, SD ¼1.258) which conformed to their ability
level (
q
) mean (M(
q
)¼0.497, SD ¼0.452). Finally, the
excellent group's post-test score mean was the highest
(M ¼6.80, SD ¼1.054) and their ability level mean (
q
) was
also considerably high (M(
q
)¼1.006, SD ¼0.411).
2) Interaction of Ability Level and Type of Feedback with
Genetic Problem-Solving Ability
The results of the two-way ANOVA revealed that both
the ability level and type of feedback interacted with
genetic problem-solving ability (F ¼9.200, p¼.000).
The summary information can be seen in Table 3.
The result of the simple effect analysis of the mean of
genetic problem-solving ability of each group after treat-
ment (immediate feedback) was as follows.
The poor group exhibited higher genetic problem-
solving ability at a statistically significant level of 0.05
(F ¼3.456, p¼.009) while the mean of the overall poor
group ability was relatively low (M ¼�0.595, SD ¼0.730).
The moderate group also exhibited higher genetic
problem-solving ability at a statistically significant level of
0.05 (F ¼26.904, p¼.000) while the mean of the overall
moderate group ability was mediocre (M ¼0.497,
SD ¼0.450).
The excellent group exhibited higher genetic problem-
solving ability at statistically significant level of 0.05 as well
(F ¼14.554, p¼.000) while the mean of the overall excellent
group ability was impressive (M ¼1.011, SD ¼0.407).
Table 3
Interaction of ability level and type of feedback with genetic problem-solving ability
Sources of variance Type III Sum of squares df Mean square F p
GROUP 319.601 2 159.801 727.607 .000
FEEDBACK 7.396 4 1.849 8.418 .000
GROUP * FEEDBACK 16.164 8 2.020 9.200 .000*
Error 169.111 770 0.220
Total 614.073 786
Corrected Total 589.092 785
Table 2
Type of feedback based on application of RISE model
RISE model Type of feedback Condition Type of feedback employed in testing system
Correct response Incorrect response
Inquire Worked example feedback 1. Full worked example feedback (FWF)
2. Partial worked example feedback (PWF)
Directive feedback 3. Full directive feedback (FDF)
4. Partial directive feedback (PDF)
Reflect Knowledge of results feedback 5. Knowledge of results feedback (KORF)
Source:Wray (2013)

C. Masantiah et al. / Kasetsart Journal of Social Sciences 41 (2020) 269–274 273
2) The testing system was developed using Adobe Flash
(Adobe Inc., San Jose, CA, USA). The testing system was
online based with an administrator.
3) Participants were distributed into three groups based on
their performance in the pretest. The excellent group
consisted of 242 participants, the moderate group con-
sisted of 309 participants, and finally, the poor group
consisted of 235 participants. Importantly, each partici-
pant was provided with only one type of feedback for a
whole test. The distribution of each type of feedback for
all participants was: 1) FWF for 67 participants, 2) PWF
for 105 participants, 3) FDF for 95 participants, 4) PDFfor
68 participants, and 5) KORF for 68 participants.
4) Testing was held in a computer room of each school. Each
participant was provided with an instruction manual, a
registration code, a username and password, and note-
paper. Once the instruction session had finished, each
participant eventually logged into the testing system.
The testing period was 90 min.
Data Analysis
1) Pretest and post-test results of genetic problem-solving
ability were analyzed using descriptive statistics (mean
and SD).
2) Reliability of the test was conducted by application of
the IRL 1-parameter model via MULTILOG (SSI Inc.,
Skokie, IL, USA). The quality of the test was analyzed
using difficulty (b) and a parameter of each participant's
ability level (
q
).
3) Two-way ANOVA was used to compare the means of ge-
netic problem-solving ability and types of feedback using
the SPSS software package (SPSS Inc., Chicago, IL, USA).
Results
1) Genetic problem-solving ability of participants and type
of feedback
The poor group's post-test score mean was noticeably
low (M ¼2.35, SD ¼2.033) and the poor group's ability
level (
q
) mean was the lowest (M(
q
)¼�0.595, SD ¼0.735).
The moderate group's post-test score mean was mediocre
(M ¼5.43, SD ¼1.258) which conformed to their ability
level (
q
) mean (M(
q
)¼0.497, SD ¼0.452). Finally, the
excellent group's post-test score mean was the highest
(M ¼6.80, SD ¼1.054) and their ability level mean (
q
) was
also considerably high (M(
q
)¼1.006, SD ¼0.411).
2) Interaction of Ability Level and Type of Feedback with
Genetic Problem-Solving Ability
The results of the two-way ANOVA revealed that both
the ability level and type of feedback interacted with
genetic problem-solving ability (F ¼9.200, p¼.000).
The summary information can be seen in Table 3.
The result of the simple effect analysis of the mean of
genetic problem-solving ability of each group after treat-
ment (immediate feedback) was as follows.
The poor group exhibited higher genetic problem-
solving ability at a statistically significant level of 0.05
(F ¼3.456, p¼.009) while the mean of the overall poor
group ability was relatively low (M ¼�0.595, SD ¼0.730).
The moderate group also exhibited higher genetic
problem-solving ability at a statistically significant level of
0.05 (F ¼26.904, p¼.000) while the mean of the overall
moderate group ability was mediocre (M ¼0.497,
SD ¼0.450).
The excellent group exhibited higher genetic problem-
solving ability at statistically significant level of 0.05 as well
(F ¼14.554, p¼.000) while the mean of the overall excellent
group ability was impressive (M ¼1.011, SD ¼0.407).
Table 3
Interaction of ability level and type of feedback with genetic problem-solving ability
Sources of variance Type III Sum of squares df Mean square F p
GROUP 319.601 2 159.801 727.607 .000
FEEDBACK 7.396 4 1.849 8.418 .000
GROUP * FEEDBACK 16.164 8 2.020 9.200 .000*
Error 169.111 770 0.220
Total 614.073 786
Corrected Total 589.092 785
Table 2
Type of feedback based on application of RISE model
RISE model Type of feedback Condition Type of feedback employed in testing system
Correct response Incorrect response
Inquire Worked example feedback 1. Full worked example feedback (FWF)
2. Partial worked example feedback (PWF)
Directive feedback 3. Full directive feedback (FDF)
4. Partial directive feedback (PDF)
Reflect Knowledge of results feedback 5. Knowledge of results feedback (KORF)
Source:Wray (2013)
Based on Simple effect analysis, the poor group, mod-
erate group, and excellent group exhibited significantly
higher genetic problem-solving ability after treatment
(immediate feedback), so multiple comparison was used to
specify the type of feedback that was actually effective for
each group as seen in Table 4.
The poor group exhibited higher genetic problem-
solving ability with PDF compared to FWF at a statisti-
cally significant level of .05.
The moderate group exhibited higher genetic problem-
solving ability with FWF, PWF, PDF, and KORF compared to
FDF at a statistically significant level of .05. The moderate
group with PWF also exhibited higher genetic problem-
solving ability compared to PDF at a statistically signifi-
cant level of .05.
The excellent group exhibited higher genetic problem-
solving ability with FWF, PWF, and KORF compared to
FDF, and PDF at a statistically significant level of .05.
Discussion
The results indicated feedback affected students
equipped with various level of comprehensive skill and
basic knowledge differently. Nonetheless, the feedback was
helpful for everyone. Attali (2011, 2015) also found a huge
improvement in student's learning effectiveness after
feedback was provided.
The poor group exhibited higher genetic problem-
solving ability with directive feedback (both FDF and
PDF) compared to worked example feedback as the poor
group seemed to use less time to comprehend feedback
compared to other groups. Consequently, the poor group
was likely to neglect complicated feedback so feedback
provided to the poor group needed to be concise and
simplified.
On the other hand, the moderate and excellent groups
exhibited higher genetic problem-solving ability with
worked example feedback (both FWF and PWF) compared
to directive feedback, as both groups were likely to use a
considerable amount of time to comprehend feedback.
Worked example feedback was slightly more complicated
than directive feedback so it required some comprehensive
skill yet provided more detail through its case study
approach. The excellent group also tended to use more time
to comprehend feedback compared to the poor and mod-
erate groups, which reflected that students with various
ability levels used different amounts of time with feedback
or testing (Gouli, Gogoulou, &Grigoriadou, 2008; van der
Kleij et al., 2012; Yastibas &Yastibas, 2015).
Feedback is helpful for learning and studying but each
student requires a different type of feedback. According to
the research findings, ability level plays an important role
in indicating the best type of feedback for particular stu-
dents. The poor group did well with simplified directive
feedback while the moderate and excellent groups did well
with worked example feedback. The instructor should pay
attention to student ability level and design a proper
teaching plan to maximize the benefit of each type of
feedback.
Conclusion and Recommendation
The results explained the relationship of student ability
level and type of feedback with genetic problem-solving
ability. The type of feedback noticeably influenced stu-
dents with different ability levels. Thus, the instructor
needs to pay attention to student ability level and provide
the proper type of feedback to a specific group of students
based on their ability level.
The poor group required special attention because of
their limited basic knowledge and comprehensive skills.
Instructor should consistently provide feedback to ensure
that the poor group knows its own strengths and weak-
nesses, so the poor group can make further progress prop-
erly. Simplified directive feedback did well with the poor
group and thus, the instructor should consider using direc-
tive feedback that is not too complicated for the poor group.
The moderate and excellent groups also needed
consistent feedback as did the poor group. In contrast, both
of moderate and excellent groups did well with worked
example feedback instead of directive feedback so the
instructor should consider using worked example feedback
for both groups.
Based on the research findings, the type of feedback
makes a big difference for each student group. In a class-
room environment, the instructor should pay attention to
the majority of students to prepare the proper type of
feedback; for example, directive feedback should be
employed for classrooms with a majority of low ability-
level students. When the proper type of feedback is
applied, such feedback may help to improve the learning
effectiveness of students.
Conflict of Interest
There is no conflict of interest.
Acknowledgments
The research was funded by the 90th anniversary of
Chulalongkorn University Fund (Ratchadaphiseksomphot
Endowment Fund).
Table 4
Multiple comparison of participant's self-evaluation accuracy based on
Tamhane's T2 technique
Group Feedback Mean
Difference (I-J)
SD p
(I) (J)
Poor PDF FWF 0.43 0.140 .027*
Moderate FWF FDF 0.50 0.095 .000*
PWF FDF 0.54 0.057 .000*
PDF FDF 0.28 0.055 .000*
KORF FDF 0.40 0.066 .000*
PWF PDF 0.27 0.063 .001*
Excellent FWF FDF 0.52 0.081 .000*
FWF PDF 0.47 0.076 .000*
PWF FDF 0.49 0.060 .000*
PWF PDF 0.43 0.053 .000*
KORF FDF 0.53 0.067 .000*
KORF PDF 0.47 0.061 .000*
p<.05 was taken to be significant

C. Masantiah et al. / Kasetsart Journal of Social Sciences 41 (2020) 269–274
274
References
Arunpruksakul, N. (2016, July). A comparison of learning outcome and
problem-solving ability of mathayomsuksa IV students from
problem-based learning. In Paper presented at 1st National Education
Conference: Educational Management for Local Development towards
ASEAN Community: New Direction in the 21st century, Kalasin, Thailand.
[in Thai]
Attali, Y. (2011). Immediate feedback and opportunity to revise answers:
Application of a graded response IRT model. Applied Psychological
Measurement, 35(6), 472e479.
Attali, Y. (2015). Effects of multiple-try feedback and question type during
mathematics problem solving on performance in similar problems.
Computers &Education, 88, 260e267.
Chang, A. (2001). Comparison of actigraphic, polysomnographic, and
subjective assessment of sleep parameters in sleep-disordered pa-
tients. Sleep Medicine, 2(5), 389e396.
Diab, N. M. (2015). Effectiveness of written corrective feedback: Does type
of error and type of Correction matter? Assessing Writing, 24, 16e34.
Foley, B. (2010). Improving IRT parameter estimates with small sizes: Eval-
uating the efficacy of a new data augmentation technique (Unpublished
doctoral dissertation). University of Nebraska, Lincoln.
Gouli, E., Gogoulou, A., & Grigoriadou, M. (2008). Supporting self-, peer-,
and collaborative- assessment in e-learning: The case of the PEer and
Collaborative ASSessment Environment (PECASSE). Journal of Inter-
active Learning Research, 19(4), 615e647.
Kanjanawasee, S. (2014). Modern test theories. Bangkok, Thailand: Chu-
lalongkorn University Press. [in Thai]
Klaimanee, S. (2015). The effect of feedback information provided on the
implementation of 5S at student dormitory Boromarajonani College
of Nursing, Nakhon Ratchasima. The Journal of Boromarajonani College
of Nursing, 20, 67e79. [in Thai]
Lumthong, D. (2010). The effects of feedback styles on visual art develop-
ment: An application of feedback and feedforward approaches (Un-
published Master’s Thesis). Chulalongkorn University, Bangkok. [in
Thai]
Musikthong, J., & Lekdamrongkul, P. (2013). Clinical teaching: Feedback.
Retrieved from http://www.ns.mahidol.ac.th/english/th/departments/
MN/th/km/56/km_feedback.html. [in Thai]
Shute, V. J. (2008). Focus on formative feedback. Review of Educational
Research, 78(1), 153e189.
Sinhaa, N., & Glassa, A. L. (2015). Delayed, but not immediate, feedback
after multiple-choice questions increases performance on a subse-
quent short-answer, but not multiple-choice, exam: Evidence for the
dual-process theory of memory. The Journal of General Psychology,
142 (2), 118e13 4.
Srichot, P. (2013). Development toward Thailand, new direction of educa-
tional assessment. Retrieved from https://www.facebook.com/notes/
453779464670556/. [in Thai]
van der Kleij, F. M., Eggen, T. J. H. M., Timmers, C. F., & Veldkamp, B. P.
(2012). Effects of feedback in a computer-based assessment for
learning. Computers &Education, 58, 263e272.
Viraphotchananan, S. (2014). The improvement of learning effectiveness:
Genetic problem solving of Mendelian laws of inheritance by Punnett
square: Mathayomsuksa, 6/1. Maewongpittayakhom School. Retrieved
from www.sns.ac.th/vichakarn/word/114.doc. [in Thai]
Wray, E. (2013). RISE model. Retrieved from http://www.emilywray.com/
rise-model.
Yastibas, G. C., & Yastibas, A. E. (2015). The effect of peer feedback on
writing anxiety in Turkish EFL (English as a foreign language) stu-
dents. Procedia Social and Behavioral Sciences, 199, 530e538.
Zhang, F., Zhang, X., Luo, M., & Geng, H. (2016). The effects of feedback on
memory strategies of younger and older adults. PLoS One, 11(12).
https://doi.org/10.1371/journal. pone.0168896.
e