Assessing Contributions To Group Assignments

Assessment & Evaluation in Higher Education
Vol. 29, No. 6, December 2004
Assessing contributions to group
assignments
Lucy Johnston* and Lynden Miles
University of Canterbury, New Zealand
We report the use of a combination of self- and peer-assessment in an undergraduate social
psychology laboratory course. Students worked in small groups on a self-directed empirical project
that they each wrote up independently as a laboratory report. Marks for the written assignment
were moderated by a contribution index measure based on the self- and peer-assessment
measures. Our analyses indicated that: (i) students took the peer-assessment process seriously,
clearly differentiating between group members on the contributions questionnaires; (ii) students
show a self-bias, rating their own contribution to the group task higher than that of other group
members; (iii) for a large majority of students the contribution index resulted in very little
moderation of the final assignment marks; (iv) there was a strong correlation between the
contribution index and the overall assignment score. Implications for the assessment of group
work are considered.
Introduction
Group work is frequently included in tertiary education courses, across a wide
variety of subject domains, including biology (Garvin et al., 1995), business studies
(Freeman, 1995), civil engineering (Rafiq & Fullerton, 1996) and computing and
information systems (Lejk et al., 1996). Group projects have a number of learning
benefits (Jacques, 1984; Michaelsen, 1992; Freeman, 1995), including exposing
students to other points of view from which they can learn and permitting the
development of more comprehensive assignments than is possible for individual-
based projects (Mello, 1993). Group projects also provide an opportunity for the
development of inter-personal and teamwork skills, such as communication, leader-
ship, planning and time management, skills that are highly sought by employers
taking on graduates (Harvey & Green, 1994). In addition, students themselves
respond positively to the opportunity that group-based projects provide for team-
work and for active learning (Garvin et al., 1995).
Despite the potential benefits of group projects, they do raise problems for student
evaluation and assessment. Converting the effectiveness of a given student’s contri-
bution to a group into a numeric grade is a complicated task. In encouraging
teamwork it is essential that students feel confident that they will be rewarded fairly
*Corresponding author. Department of Psychology, University of Canterbury, Private Bag 4800,
Christchurch, New Zealand. Email: l.johnston@psyc.canterbury.ac.nz
ISSN 0260-2938 (print)/ISSN 1469-297X (online)/04/060751-18
 2004 Taylor & Francis Ltd
DOI: 10.1080/0260293042000227272

752 L. Johnston and L. Miles
for their contributions and that any ‘free-riders’ will not benefit unduly from the
efforts of others. The extent to which students should be rewarded for the process
of the project versus the outcome of the project is, however, a contentious issue.
Debate continues as to whether it is appropriate, in the tertiary sector, for a student
to obtain higher marks than other group members simply as a result of greater than
average effort to the group endeavour (Lejk et al., 1996). The usual practice in
higher education is for students to be graded solely on the quality of a submitted
piece of work without consideration of the effort or input into that product. Indeed,
in most cases the grader will be unaware of the amount of time and effort that a
student has devoted to an assignment. For individual-based projects, however, both
the input and output are determined by the individual. In this case, rewarding the
student only for the output of their endeavours is, then, a reasonable means of
assessment. For group assignments, however, the link between individual inputs and
the output is not so clear; a ‘free-rider’, for example, might receive a high grade
despite having very little input into an otherwise good group. Accordingly, in group
assignments, the allocation of some marks to students based on their contribution to
the group project rather than just on the final outcome of that project is not unusual.
How to allocate such marks has, however, been a topic of considerable debate and
analysis.
It is often difficult for a lecturer or tutor to adequately assess the contribution of
individuals to a group project since much of the work occurs outside formal teaching
sessions. Accordingly, grading of the contribution of individuals to a group
assignment has increasingly been handed over to those with the most relevant
information, i.e. the co-workers. Such peer-assessment has taken many forms (see
Lejk et al., 1996, for a survey of such methods), but the essential components are the
same; group members evaluate each other and this evaluation is incorporated into
the individual student’s assessment, either as an addition to the marks awarded for
the submitted assignment (Earl, 1986; Falchikov, 1988, 1991; Gibbs, 1992;
Habeshaw et al., 1993: Hindle, 1993) or as a moderator of those marks (Goldfinch
& Raeside, 1990; Conway et al., 1993; Goldfinch, 1994). In some instances student
peer-assessment has included evaluation of the quality of the work produced as well
as the contribution of students to that work. Although there is some evidence that
students reliably reproduce staff marks (Orpen, 1992), some concern has been
raised about the ability of students to assess the quality of work (Burke, 1969; Beard
& Hartley, 1984; Rowntree, 1987), with student scores tending to cluster around
the mean score with relatively little variance (Hughes & Large, 1993; Freeman,
1995). In our study the quality of the written assignment was assessed by the
laboratory tutors and lecturers, with the peer-assessment related only to the group
processes.
Peer-assessment has largely been applied to situations where the work group
submits a single project (Lejk et al., 1996). The project is awarded a mark, which is
then moderated for each group member as a function of the peer-assessment. Group
work does not always, however, culminate in the production of a single piece of
work. In some cases each individual group member may submit a piece of work
based on their group project and hence each student is assessed individually after

Assessing contributions to group assignments 753
learning collaboratively (Hufford, 1991). This is the case for our laboratory assign-
ment. Students worked together on a group project but each wrote a research report
that was graded by either a laboratory tutor or lecturer prior to adjustment by peer
evaluation factors. One aim of the present study was to investigate the use of
peer-assessment in such a context. Individual assessment after group work may
provide some protection for students against ‘free-riders’ but does not negate the
utility of peer-assessment, since the basis for the individual reports is in the group
work and so may be enhanced or impeded by high or low contributing group
members, respectively.
A second aim of this project was to consider the question of whether or not
self-assessment should be included in peer-assessment. Methods of peer-assessment
differ according to whether they include self-assessment (see for example Goldfinch,
1994) or not (see for example Conway et al., 1993). Some researchers have argued
for the inclusion of self-assessment (for example Goldfinch, 1994) and others for its
omission (Falchinov, 1991; Lejk & Wyvill, 2001). However, as Lejk and Wyvill
(2001) noted, useful analysis of the self-assessment component of peer-assessment
is infrequent. We did include self-evaluations, but our analysis also considered the
consequences of omitting self-assessments on the grades obtained by the students.
Goldfinch (1994) argued for the inclusion of self-assessment, noting that in some
cases, where only peer but not self-assessment was allowed, one group member may
be more generous in his or her evaluations than the rest of the group and accordingly
penalize himself or herself. By including self-assessment, overly generous students
would presumably inflate their own grade as much as they do that of their peers.
However, where self- and peer-assessments have differed it appears as if students
actually tended to assess themselves lower than did their peers (Krause & Popovich,
1996), especially the more able students (Lejk & Wyvill, 2001). Inclusion of
self-assessment may then disadvantage more able students, which may be a reason
not to include self-assessment (Lejk & Wyvill, 2001). Falchikov (1991) also argued
against the inclusion of self-assessments, simply because she found a much lower
agreement between self- and peer-ratings than between peer-ratings, although her
very low sample size (n 4) must be noted. Analysis of the impact of the inclusion
and exclusion of self-assessment in the present study will add further to this
literature.
The present paper presents details of the employment of a peer-assessment
method in an advanced undergraduate social psychology laboratory class. Students
worked in small groups on an empirical project which each group member then had
to write up individually. After submitting their assignment each student was re-
quired to complete a contributions questionnaire for themselves and for each of the
other members of their group, i.e. self- and peer-assessment. The contributions
questionnaire was completed secretly, to facilitate greater discrimination by students
in their peer-assessments (Lejk Wyvill, 2001). A ratio of mean individual contri-
bution ratings to mean rating for the work group was used to moderate the grade
given to the individual’s written assignment. Accordingly, the mark awarded to the
written assignment could be either increased or decreased as a result of an individ-
ual’s relative contribution to a group project.

Method
Participants
Sixty-one (48 female, 13 male) undergraduate students completed the group-based
assignment and contributions questionnaires. Groups consisted of 3 (n 2), 4
(n 10) or 5 (n 3) students. Five of the groups comprised female students only
and 7 both male and female students. The groups were self-selected by the students.
The project ran in the second semester of a full-year course and so it is likely that
group selection was based on friendship and on previous interactions in the first
semester laboratory sessions.
Nature of the assignment
The aim of the assignment was to give students experience in conducting social
psychological research. Students worked in small groups designing and conducting
a piece of social psychological research. Three topic areas, and key references, were
introduced by the course lecturers and students were to select one of these areas in
which to develop their own project, within the parameters provided by the lecturers
concerned. Prior approval for these research projects had been obtained from the
University of Canterbury Human Ethics Committee by the course coordinator. The
topics available were: (i) investigating the relationship between empathy and proso-
cial behaviour; (ii) testing the Ideal Standards Model (Fletcher et al., 1999; Simpson
et al., 2001) in mate selection; (iii) investigating the nature of impression formation
from faces.
Groups designed their own study, collected data and analysed their results. Each
student was then required to individually write up their group’s study. Each group
worked with a laboratory tutor (a graduate student in social psychology), meeting
formally at 2 week intervals throughout the duration of the project (10 weeks). In
the final week of the semester each group was required to give a brief (15 min)
presentation to the class on their project. The entire assignment contributed 20% of
the marks for the full year undergraduate paper in social psychology.
Contributions questionnaire
Students were informed, before they divided into work groups and began the
project, that the marks assigned to their written assignment would be moderated by
self- and peer-assessment of the contribution of each member to the group project.
Full details of the moderation procedures were given to students both orally and in
a written handout. After the reports have been handed in, each student rated the
contribution of each member of their work group, including themselves, on a
number of task functions and group management skills (Johnson Johnson, 1975;
Falchikov, 1988, 1991): generation of research ideas; contribution to background
literature searches and reviews; contribution to research design and methodology;
contribution to data collection; contribution to data analysis and interpretation;
contribution to feedback presentation; contribution to project management (e.g.

communication with the laboratory assistant; organizing group meetings; preparing
materials). For each of the tasks a score of between 3 and –1 was given (3, major
contribution; 2, some contribution; 1, minor contribution; 0, no contribution; –1, a
hindrance to the group). As previous research has demonstrated greater agreement
in peer assessment on general impressions of an individual than on specific dimen-
sions (Thornton Byham, 1982; Klimoski Brickman, 1987; Schectman, 1992),
a total rating across the seven dimensions was calculated for each target. Accord-
ingly, the maximum possible score for any individual was 21 and the minimum –7.
For each group member their average rating out of 21 was calculated and then the
average rating for each work group calculated. For each group member a ratio of
their rating to the group average was calculated (contribution index) and used as a
multiplier on the mark for the written report. If an individual contributed more than
an average amount for their group, the ratio would exceed 1.0 and the mark for their
written report increase. If an individual contributed less than an average amount for
their group, the ratio would be 1.0 and the mark for their written report decrease.
It was acknowledged that it would be possible, using this ratio measure, for a student
to score 100% for this assignment and, if this occurred, the student would be
credited with 100% for this component of the course (i.e. they would receive
more than the ‘maximum’ number of marks assigned to this component of course
evaluation).
Results and discussion
Our analysis will be presented in four parts: (i) descriptive analysis of the contribu-
tions ratings; (ii) consideration of the relationship between the contributions ratings
and performance on the written assignment; (iii) strategies; (iv) consideration of
alternative ways of using the contributions ratings to calculate marks for the
assignment.
Contributions ratings
Each student rated themselves and the other members of their work group. From
these ratings a number of measures were calculated: ratings of their own contribu-
tion to the group (rating of self); the mean rating given by other members of the
group to each individual (others ratings of self); the mean rating given by each
individual to the other group members (ratings of others in group); the mean rating
for each individual from themselves and other group members (own contribution);
the mean rating for all members of the group (mean group contribution); the
contribution index (own contribution/mean group contribution). The mean, median
and range for each of these measures in shown in Table 1. The distribution of the
contribution index measure is also shown in Figure 1.
A repeated measures ANOVA (rating of self; others rating of self; rating of others
in group) revealed a significant effect [F(2,120)3.49, P0.05]. Ratings of the self
were significantly higher (Fisher LSH, P0.05) than others rating of the self and
rating of others in group, the latter two not differing from one another

Table 1. Contribution ratings
Minimum Maximum
Mean Median
Rating of self 8.00
18.47 21.00
19.00
Others ratings of self 17.09 18.25 3.5 21.00
Ratings of others in group 21.00
17.00 7.33
18.00
Own contribution 8.50
17.36 21.00
18.00
Mean group contribution 17.32 17.85 13.00 20.75
Contribution index 1.256
1.002 0.562
1.008
(Means 18.47 versus 17.09 and 17.00). Individuals saw themselves as contribut-
ing more to the group project than other members of their work groups. There was,
however, no difference between own contribution (mean of self- and peer-ratings of
self) and the mean group contribution (Ms 17.36 versus 17.32), suggesting that
once self and other perceptions of contribution were combined there was little
difference in the scores of group members within a given workgroup. This is
reflected in the mean and median values for the contribution index being very close
to 1.0 (mean 1.002, median1.008).
Additional analyses considering the impact of sex of student, size of group and
type of group (mixed sex or all female) on the contribution measures revealed few
significant effects. There were no significant effects of sex of student. There was an
Figure 1. Distribution of Contribution Index scores

effect of group size on the ratings of self only, with participants rating their own
contribution lower if they were in a group of 5 rather than a group of either 3 or 4
students [F(2,58) 6.07, P0.01, Ms18.33, 19.30 and 16.47). There were no
significant differences between the mean scores of those in mixed sex work groups
and those in female only work groups. Considering only the female students, there
were significant differences of group type (all female versus mixed sex) in the scores
on the others ratings of the self [F(1,46) 4.60, P0.05, Ms 15.94 versus 18.21]
and the ratings of others in the group [F(1,46)4.34, P0.05, Ms 15.62 versus
18.02] with females both being rated higher by others and rating other group
members higher when in mixed sex groups. Considering only students in the mixed
groups there were no significant differences between male and female students on
any of the contribution measures.
Pearson correlation coefficients were calculated between the ratings of the individ-
ual by themselves and by other group members. There were no significant correla-
tions between ratings of the self and ratings of the self by others, i.e. there was no
correlation between the ratings given to the self and given to the self by other group
members [r(61) 0.159, ns], indicating a difference in the perception of an individ-
ual by themselves and by other group members. This result parallels the results from
the between-groups analysis reported above, which demonstrated a difference be-
tween self-ratings and others’ ratings of the self and is consistent with the findings
of Falchikov (1991). Similarly, there was no correlation between ratings of one’s
own contribution and that of other group members [r(61) 0.155, ns], again
suggesting a differentiation of own contribution from that of other group members.
This again parallels the effect seen in the between-groups analysis, which showed
individuals to rate their contribution higher than that of other members of their work
group. Together these findings suggest that individuals are somewhat distorted in
their view of their own contribution to the group project. This is also reflected in the
absence of a significant correlation between ratings of the self and the self contribu-
tion index [r(61) 0.097, ns].
Within each work group we calculated Kendall’s measure of concordance, a
measure of agreement in the ranking of items across judges and ranges (0, no
agreement; 1, perfect agreement). Within each group we converted the contribution
ratings provided by each group member to ranks, with higher ranks given to those
who were rated higher on the contributions questionnaire. For each group we
calculated Kendall’s coefficient of concordance, once including self-assessment and
once excluding self-assessment. Across the 15 groups the mean Kendall’s coefficient
of concordance was W0.423 (range 0.042–0.933) when self-assessment was
included and W0.584 (range 0.204–1.00) when self-assessment was excluded
[this difference was significant, F(1,14)17.95, P0.001]. There was relatively
strong agreement within groups as to who were the greater and lesser contributors
to the group project. Taking consensus as a reasonable indicator of accuracy, this
finding would also suggest that students were indeed able to accurately evaluate the
contribution of group members. Consistent with the previous findings of self-bias,
agreement was greater when self-assessments were excluded. There was no effect of
the sex of the group (all female versus mixed) on Kendall’s coefficient of concord-

Table 2. Assignment marks before and after adjustment by the contribution index
Mean Minimum
Median Maximum
Pre-adjustment
Of 20 13.09 13.65 5.7 16.68
As a percentage 83.39
65.47 28.5
68.25
Post-adjustment
Of 20 13.22 13.41 3.2 19.79
As a percentage 98.93
66.21 16.02
67.04
Change
Of 20 4.04
0.132 4.42
0.138
As a percentage 0.741 0.689 20.18
22.12
ance. There was, however, an effect of group size with higher concordance for
smaller groups. This effect was significant when self-assessment was excluded
[F(2,12)5.77, P0.05, Ms1.0 for the groups with 3 members, 0.56 for the
groups with 4 members and 0.43 for the groups with 5 members) and marginally
significant (P0.07) when self-assessments were included (Ms 0.8, 0.374 and
0.365). This finding may simply reflect the fact that the smaller a group is the more
salient the contribution of each group member is or that with fewer judges there is
less potential for variance of ratings.
Contribution ratings and performance
The marks awarded for the written assignment, before and after adjustment by the
contribution index, are shown in Table 2. There were no effects of group type
(mixed or all female) or group size on assignment marks. There was, however, a
significant effect of sex on the assignment marks. Female students received higher
marks before adjustment [F(1,59)11.90, P0.01, Ms 67.73 versus 57.14%], a
difference that was enhanced after adjustment for group contribution
[F(1,59) 9.17, P0.01, Ms 69.07 versus 55.66%]. However, given the low
number of male students (n 13), this finding should be interpreted with caution.
There was no significant difference between the assignment marks before and
after adjustment by the contribution index. Consistent with a mean contribution
index very close to 1.0, there was very little mean change in the marks awarded (a
median increase of 0.689%). There was, however, a wide range of changes to marks,
from a decrease of 22.12% to an increase of 20.18%, as shown in Figure 2. Where
there was a large discrepancy between the contribution of an individual and that of
the work group this was reflected in a large impact on the assigned marks; the
correlation between the contribution index and the change in percentage points was
highly significant [r(61) 0.959, P0.001].
Correlations between the contribution ratings and assignment marks are shown in
Table 3. There was no relationship between ratings of self and assignment marks,
either before or after adjustment [r(61) 0.132, ns; r(61) 0.059, ns]. Consistent
with the previous findings, this suggests that individuals do not accurately evaluate

Figure 2. Distribution of change in assignment marks (as percentage points)
their own contributions. There was, however, a significant relationship between
other’s ratings of the individual and marks [r(61) 0.382, P0.01; r(61) 0.534,
P0.001] and between the contribution index and marks [r(61) 0.484, P0.001;
r(61) 0.797, P0.001]. These correlations indicate that those who contributed
most to the group project also received higher marks for the project. It is also of
interest that there is a significant difference (P 0.003) in the correlation
coefficients between the contribution index and marks before and after adjustment
for contribution. The correlation is significantly stronger after adjustment of the
marks. A strong contribution to the group effort does enhance overall performance
on the assignment.
It has been suggested that self-assessment varies as a function of ability level, with
higher ability individuals under-rating themselves and lower ability individuals
Table 3. Correlations between contribution measures and assignment marks
Mark before adjustment Mark after adjustment
0.059
Rating of self 0.132
Others rating of self 0.382a
0.534b
0.484b
0.797b
Contribution index
a
P0.01.
b
P0.001.

over-rating themselves (Lejk Wyvill, 2001). To investigate this possibility amongst
our participants we used a quartile split on the unadjusted marks to categorize
participants according to ability level. A discrepancy score was calculated for each
participant between their self-assessment and their assessment by peers; positive
scores represented higher self- than peer-ratings and negative scores the converse. A
single factor ANOVA (quartiles lowest/low medium/high medium/highest) on the
discrepancy scores revealed a significant effect [F(3,59)4.25, P0.01]. Post hoc
tests (Tukey HSD, P0.05) revealed a significant difference between those in the
highest and lowest quartiles, but contrary to predictions, those in the highest quartile
over-rated themselves whilst those in the lowest quartile under-rated themselves
(Ms 1.04 versus –0.76). Discrepancy scores for those in the middle quartiles feel
in between the highest and lowest quartiles (Ms 0.58 and –0.44).
Strategies
Somewhat surprisingly there appeared to be no instances of group collaboration over
the ratings given. For no group were all group members given the same ratings by
all members. Indeed, only 16 (26%) students failed to differentiate between group
members and gave the same ratings to all members of their group, including
themselves. A further 2 (3%) students rated all the other group members the same,
but themselves differently. We interpret these data as indicative of the students
taking the contributions rating task seriously. This is consistent with the relatively
high mean values on the Kendall’s concordance measure reported above. In dis-
cussing possible group strategies we are also mindful of the earlier analyses, which
indicated a differentiation between the ratings given to the self and those given to the
self by other group members; there was a tendency for participants to rate their own
contribution as greater than that of other group members. This self bias may in itself
have been sufficient to prevent group strategies developing. The contributions
questionnaires were also filled in individually and were confidential such that it is
possible that even if groups had discussed group strategies these were not adhered
to in the actual ratings. There were indeed two work groups in which there may have
been a conflict between prior discussions that all group members were to be given
maximum marks and recognition of differential contributions—in these groups there
was almost blanket ratings of all group members (e.g. in one group of 4, 2 group
members gave all 4 members the maximum rating of 21 while 2 gave 2 members
ratings of 21 and 2 ratings of 20).
Individuals could also have adopted strategies in their ratings in order to increase
the likelihood of obtaining a high personal contribution index. This may be particu-
larly pertinent to individuals who have contributed more or less than the average
group member. Somebody who has made above average contributions to the group
may attempt to emphasize this by rating their own contribution very highly but that
of others lowly, rating themselves higher than others rate them and rating others
lower than others do. Such a rating strategy would benefit the high contributing
individual if the other group members similarly recognized their high level of
contribution. Ironically, somebody who has contributed little to the group may show

a similar pattern of ratings in order to mask their lower contribution. Do our data
show any evidence of such patterns? We looked at the students with the three
highest (1.26, 1.25 and 1.23) and the three lowest (0.562, 0.646 and 0.796)
contribution indices. The three with the highest indices did indeed rate themselves
as the highest contributing group member, but in each case at least one other group
member also did so. For the three lowest contribution indices, one individual also
rated themselves as the lowest contributing group member. The other two individu-
als gave all the group members, including themselves, the highest contribution rating
(the maximum ratings of 21), suggesting that they were indeed attempting to reduce
the impact of their low contribution to the group (rated as 5 and 2 in one instance
and 5, 4 and 4 in the other), but did not employ the optimal strategy to do so, only
over-rating their own contribution and not denigrating the contribution of other
group members. It is possible that students would feel more comfortable in inflating
their own contribution than devaluing that of others. It is noteworthy that two of the
work groups contained both one of the highest and one of the lowest scoring
students (1.26 and 0.562 and 1.25 and 0.646) on the contributions index, suggest-
ing that in each of these groups one group member may have compensated for a
‘free-rider’.
Alternative approaches
The analyses above report detail regarding how the peer- and self-assessments were
calculated and used in the grading of our laboratory assignment. Our data allow us
to consider the possible impact of using the contributions ratings in different ways.
We consider two such alternatives; removing the self-assessments and using the
contribution index to allocate additional marks rather than as a moderator of the
marks awarded for the written assignment.
There was no correlation between self- and peer-assessments, with self-assess-
ments being higher, on average, than peer-assessments. Similarly, there was no
correlation between self-assessments and marks awarded for the assignment, but a
significant correlation between peer-assessments and marks awarded. This raises the
question of whether inclusion of self-assessment leads to a less valid mark than if
only peer-assessment were used (Falchikov, 1991; Lejk et al., 1996). We calculated
the contribution index for each student omitting the self-assessment ratings (contri-
bution index mean peer-assessment/mean group contribution). The mean peer-as-
sessment ratings were the ‘other ratings of self’ scores considered above. The mean
group score had to be recalculated, however, using only the peer-assessments.
Differences between this revised index and that including self-assessments were
small for the mean (1.00 versus 1.002) and median (1.02 versus 1.008) scores but
somewhat larger for the minimum (0.27 versus 0.56) and maximum (1.42 versus
1.26) scores. There was, however, a very strong correlation between the two
contribution index measures [r(61) 0.945, P0.05]. The correlations between
each of the indices and the assignment scores are also very similar, both pre-adjust-
ment [index 1, r(61) 0.484, P0.05; index 2, r(61) 0.431, P0.05] and
post-adjustment [index 1, r(61) 0.792, P0.05; index 2, r(61) 0.727, P0.05].

762
L.
Johnston
and
L.
Miles
Table 4. Alternative calculations of the impact of the contribution index
Mean difference between this
Maximum measure (out of 20 marks) and
Measure (contribution index of that based on the contribution
Median (out (out of 5
Minimum (out
Mean (out
of 5 marks) marks)
of 5 marks)
of 5 marks)
1.0, equivalent to x marks/5) index ratio
3.26 3.28 1.83
3.25/5 4.08 0.148
4.01 4.03 2.25
4/5 5.02 0.603
5.01 5.04 2.81
5/5 6.28 1.61

The change to the assignment scores using this revised contribution index was small
(mean–0.858%, median –0.912%) but with a wide range (from –30.57 to
45.67%). There was no significant difference between the change to the assign-
ment scores (Ms 0.689 and –0.858) or in the final assignment scores using the two
different contribution index measures. Further, the impact on the assignment scores
using the two measures of contribution index correlated highly [r(61) 0.932,
P0.05], as did the two final assignment scores [r(61) 0.964, P0.05].
In our assessment we used the ratings on the contributions questionnaire as a
multiplier of the students’ assignment marks and in doing so integrated the marks
allocated to the process and the product of the group assignment. An alternative
approach is to assign completely separate marks to these two aspects of the group
project. For example, fewer possible marks might be allocated to the written
assignment and additional marks attributed solely based on the contribution ratings.
This would keep the total potential marks for the assignment constant but totally
separate the marks awarded for process and for output. We considered one possible
such approach, calculating grades on the basis of the marks awarded for the written
assignment contributing 15% of the total course marks and those awarded based on
the contributions rating contributing an additional 5% of the course marks. In using
this method a decision must be made as to the number of marks out of 5 to allocate
to an average contribution to the group (a contribution index of 1.0). We computed
three measures, equating a contribution index of 1.0 to a mark of 3.25/5 which was
equivalent to the mean percentage (65%) for the marked assignments prior to
adjustment; equating a contribution index of 1.0 to a mark of 5/5, which is
equivalent to assuming that average contribution was the maximum possible; equat-
ing a contribution index of 1.0 to a mark of 4/5 (80%). The impact of these different
measures can be seen in Table 4.
Assuming that the average contribution was equivalent to a maximal contribution
(contribution index of 1.0 5/5 marks) resulted in a mean allocation of 5 marks
for this component of the assessment and an average increase in marks of 1.61/20
(8.05%) for the assignment. Having the mean contribution being awarded maxi-
mum marks is somewhat unrealistic. Using the mean mark (65%) for the written
assignments as the benchmark for the contributions marks (i.e. a contribution index
of 1.065% or 3.35/5) is somewhat more realistic. This resulted in a mean mark
(3.26/5) very close to the mean mark for the written assignments (9.75/15) and a
total mark (out of 20) very close to that calculated above using the contribution
index ratio. The mean difference in total marks between these two methods of
calculation was –0.148/20 (0.74%). Equating a contribution index of 1.0 to a mark
of 4/5 produced a slight increase in overall marks (0.603/20, 3.02%) compared with
the index measure, but a mark of 4/5 (or 80%) is also unrealistic as a mean mark for
an assignment at this level and therefore results in a greater relative weighting of the
contributions component of the mark than the written assignment.
General discussion
We have discussed the nature and consequences of employing a self- and peer-as

sessment component to the awarding of marks for a group-based project in an
advanced undergraduate social psychology laboratory course. We now highlight a
number of key points involved in this method of assessment and consider the general
usefulness of peer- and self-assessment.
On average, the contribution index measure employed in our laboratory had little
impact on the marks awarded to students for this assignment, a mean increase of
0.7%. This lack of impact does not, in our view, however, indicate that inclusion of
the peer evaluation was not worthwhile. Our aim in using peer-assessment was
primarily to engage students in their work groups, to have them take responsibility
for their learning (Michaelsen, 1992; Rafiq Fullerton, 1996) and to minimize the
likelihood of ‘free-riders’. An average contribution index very close to 1.0 is consist-
ent with achieving these aims as it indicates that most students were perceived to be
‘pulling their weight’ in their work groups. This interpretation is supported by
evidence that students were taking the task seriously and differentiating between
group members rather than simply giving ‘blanket’ ratings. There was a relatively
high agreement on the rankings of the contributions of group members within each
work group, suggesting that students did find it possible to evaluate relative contri-
butions to the project by different group members. Further, the range of contribu-
tion index scores and the impact on assignment marks suggests that ‘free-riders’
were being penalized for their lack of contribution to the group effort and those who
made above-average contributions to the group were being rewarded for so doing.
There were some noteworthy differences in our results and those of previous
studies. Students in our sample rated their own contribution more highly than did
their peers, in contrast to previous evidence that individuals tend to under-rate
themselves (Krause Popovich, 1996). Further, in our sample there was no effect
of ability level on the tendency to over- or under-rate oneself (Lejk Wyvill, 2001).
The nature of the assignment in which we employed peer-assessment was somewhat
different from that of most previous studies (Lejk et al., 1996) with each student
within each work group submitting a unique written assignment based on the group
project which was marked by a laboratory tutor or lecturer and then moderated as
a function of the peer-assessment regarding matters of process. It is possible that
including an independently submitted written assignment and/or having the peer-as-
sessment component confined to matters of process (Burke, 1969; Beard Hartley,
1984; Rowntree, 1987) may explain the discrepancies between the present research
and past studies, but further research is required before any firm conclusions can be
drawn.
However, one concern raised by our analysis is the possibility of students receiving
double rewards, or punishments, for their assignment through the grading and the
contribution index measure. There was a strong correlation between the pre-adjust-
ment marks and the contribution index measure, suggesting that those who con-
tributed most were already receiving high marks whilst those who were not
contributing were receiving low marks. It would appear that, despite it being a group
project, it was difficult to achieve high marks on the written assignment if one did
not also contribute to the group endeavours. Low marks and low contribution index
scores could also, of course, indicate low motivation and engagement in the course.

Given the importance of both process and product in effective group work, the
possibility of being twice rewarded is perhaps not of concern. Our analyses also
considered the possibility of using the contributions index to award separate ‘contri-
bution marks’ for the assignment rather than being used to moderate the awarded
marks. Although one could argue that this separation of product and process marks
is appropriate, it is noteworthy that the two measures have very similar impacts on
students’ marks for the assignment.
Whether self-assessment should be included within peer-assessment has been a
controversial issue, with some researchers arguing that self-assessment should not be
included (see for example Lejk Wyvill, 2001) and others that it should (see for
example Goldfinch, 1994). In our laboratory we used a contribution index calcu-
lated from both self- and peer-assessments of contribution to the assignment but,
given previous questioning of whether self-assessment was a valid measure
(Falchikov, 1991; Lejk et al., 1996; Lejk Wyvill, 2001), we also included analysis
of the relative impact of a contribution index calculated solely on peer assessment.
As expected (Falchikov, 1991; Lejk et al., 1996), there was little relationship
between self- and peer-assessments of the individual. Further, only peer-assess-
ments, and not self-assessments, correlated significantly with both the contribution
index and marks awarded. However, the impact on the assignment marks of the two
contribution indices (including and omitting self-assessment) was very similar. The
main impact of the exclusion of the self-ratings from the contribution index was to
increase the spread of the distribution of that index, with the minimum index lower
and the maximum index higher when self-ratings were omitted compared with when
they were included. In turn, this resulted in a wider range of adjustments to the
assignment marks, although the mean and median adjustments remained very
similar.
Clearly the level of agreement between self- and peer-ratings is lower than that
amongst peer-ratings (Falchikov, 1991; Goldfinch, 1994; Krause Popovich, 1996;
Lejk Wyvill, 2001). Whether these findings should be taken as indicating that it
would be better to omit than include self-ratings in peer-assessment is unclear.
Self-assessment is an important skill to learn and omission of self-ratings from peer
assessment may reduce the extent of such self-reflection. The key question is
whether the learning benefits of self-reflection on one’s own contributions toward
group-based learning outweigh any small differences in the impact on the assign-
ment grades. For most of the students in our research, omitting the self-ratings from
the contribution index would have had a minimal impact on their final assignment
grade, although doing so would have been advantageous to the top students and
disadvantageous to the bottom students. In our opinion, the benefits of self-assess-
ment in terms of the promotion of critical, reflexive thinking (Somervell, 1993) are
sufficiently strong such that we will include self-assessments in future evaluations.
Future research might consider investigating whether self- and peer-ratings within
this assessment process might best be used in different ways. Peer-ratings might be
best used to moderate grades, as described in this paper, whilst self-ratings might
best be used in a developmental manner, teaching self-reflection and critical evalu-
ation skills.

Conclusion
Peer-assessment allows for the contribution of individuals to a group project to be
evaluated, and enhances the learning experience of students. Peer-assessment has
been shown to promote independent, reflective, critical learners (Somervell, 1993),
to enhance motivation for participation amongst students (Michaelsen, 1992) and to
encourage students to take responsibility for their learning (Rafiq Fullerton,
1996). Feedback from peer-assessment also allows students to learn about their own
effectiveness in a group setting. In sum, our project supports the positive contribu-
tion of peer-assessment within student group-based assignments. While, for the
majority of our students their final result for the assignment was not substantially
moderated by the contribution index, the range of index scores indicated that this
method of assessment was sensitive to both below- and above-average contributors.
In addition, knowledge that individual contributions to the group project were to be
assessed likely resulted in a low incidence of free-riding and, consequently, more
involvement in group-based learning. We would endorse the inclusion of such a
method of peer-assessment in estimating individual contributions to group produc-
tivity when assessing group-based projects at the tertiary level.
Acknowledgements
The contributions of Alice Boyes, Myron Friesen, Geraldine Henry and Nickola
Overall are gratefully acknowledged.
Notes on contributors
Lucy Johnston is a Senior Lecturer in Psychology at the University of Canterbury,
New Zealand. Her main research and teaching interests are in stereotyping and
stigmatization.
Lynden Miles is a Ph.D. student in the Department of Psychology at the University
of Canterbury. His research focuses on implicit processes in the perception of
facial expressions of emotion.
References
Beard, R. Hartley, J. (1984) Teaching and learning in higher education (4th edn) (London,
Paul Chapman).
Burke, R. (1969) Self-assessments and peer ratings, Journal of Educational Research, 61, 444–448.
Conway, R., Kember, D., Sivan, A. Wu, M. (1993) Peer assessment of an individual’s
contribution to a group project, Assessment Evaluation in Higher Education, 19, 45–56.
Earl, S. E. (1986) Staff and peer assessment—measuring an individual’s contribution to group
performance, Assessment Evaluation in Higher Education, 11, 60–69.
Falchikov, N. (1988) Self and peer assessment of a group project designed to promote the skills
of capability, Programmed Learning and Educational Technology, 25, 327–339.
Falchikov, N. (1991) Group process analysis: self and peer assessment of working together in a
group, in: S. Brown P. Dove (Eds) Self and peer assessment (Birmingham, SCED
Publications).

Fletcher, G. J. O., Simpson, J. A., Thomas, G. Giles, T. (1999) Ideals in intimate relationships,
Journal of Personality and Social Psychology, 76, 72–89.
Freeman, M. (1995) Peer assessment by groups of group work, Assessment Evaluation in Higher
Education, 20, 289–299.
Garvin, J. W., Butcher, A. C., Stefani, L. A. J., Tariq, V. N., Lewis, M. H. R., Blumson, N. L.,
Govier, R. N. Hill, J. A. (1995) Group projects for first-year university students: an
evaluation, Assessment Evaluation in Higher Education, 20, 273–288.
Gibbs, G. (1992) Improving the quality of student learning (Bristol, Technical and Educational
Services Ltd).
Goldfinch, J. (1994) Further developments in peer assessment of group projects, Assessment
Evaluation in Higher Education, 19, 29–35.
Goldfinch, J. Raeside, R. (1990) Development of peer assessment technique for obtaining
individual marks on a group project, Assessment Evaluation in Higher Education, 15,
210–225.
Hughes, I. E. Large, B. J. (1993) Staff and peer-group assessment of oral communication skills,
Studies in Higher Education, 18, 379–385.
Habeshaw, S., Gibbs, G. Habeshaw, T. (1993) 53 interesting ways to assess your students (3rd
edn) (Melksham, Cromwell Press).
Harvey, L. Green, D. (1994) Employer satisfaction (Birmingham, University of Central Eng-
land).
Hindle, B. P. (1993) The ‘Project’: putting student-controlled, small-group work and transferable
skills at the core of a geography course, Journal of Geography in Higher Education, 17, 11–20.
Hufford, T. L. (1991) Increasing academic performance in an introductory biology course,
Bioscience, 41, 107–108.
Hughes, I. E. Large, B. J. (1993) Staff and peer-group assessment of oral communication skills,
Studies in Higher Education, 18, 379–385.
Jacques, D. (1984) Learning in groups (London, Croom Helm).
Johnson, D. W. Johnson, R. T. (1975) Joining together. Group theory and group skills (Englewood
Cliffs, NJ, Prentice Hall).
Klimoski, R. Brickman, M. (1987) Why do assessment centers work? The puzzle of assessment
center validity, Personnel Psychology, 30, 353–361.
Krause, J. E. Popovich, N. G. (1996) A group interaction peer/self assessment process in a
pharmacy practice course, American Journal of Pharmaceutical Education, 60, 136–145.
Lejk, M., Wyvill, M. Farrow, S. (1996) A survey of methods of deriving individual grades from
group assessments, Assessment Evaluation in Higher Education, 21, 267–280.
Lejk, M. Wyvill, M. (2001) The effect of the inclusion of self-assessment with peer assessment
of contributions to a group project: a quantitative study of secret and agreed assessments,
Assessment Evaluation in Higher Education, 26, 551–561.
Mello, J. A. (1993) Improving individual member accountability in small group settings, Journal
of Management Education, 17, 253–259.
Michaelsen, L. K. (1992) Team learning: a comprehensive approach for harnessing the power of
small groups in higher education, To Improve the Academy, 11, 107–122.
Orpen, C. (1982) Student vs. lecturer assessment of learning: a research note, Higher Education,
11, 567–572.
Rafiq, Y. Fullerton, H. (1996) Peer assessment of group projects in civil engineering, Assessment
Evaluation in Higher Education, 21, 69–81.
Rowntree, D. (1987) Assessing students: how shall we know them (revised edition) (London,
Kogan Page).
Schechtman, Z. (1992) Interrater reliability of a single group assessment procedure administered
in several educational settings. Journal of Personnel Evaluation in Education, 6, 31–39.
Simpson, J. A., Fletcher, G. J. O. Campbell, L. (2001) The structure and function of ideal
standards in close relationships, in: G. J. O. Fletcher M. Clark (Eds) Blackwell handbook
of social psychology: interpersonal processes (Oxford, Blackwell), 86–106.

Somervell, H. (1993) Issues in assessment, enterprise and higher education: the case for self-, peer
and collaborative assessment, Assessment Evaluation in Higher Education, 18, 221–233.
Thornton, S. C. Byham, W. C. (1982) Assessment center and managerial performance (New York,
Academic Press).

Assessing Contributions To Group Assignments

More Related Content

Similar to Assessing Contributions To Group Assignments (20)

More from Cassie Romero (20)

Recently uploaded (20)

Assessing Contributions To Group Assignments