A t

Autograding
di E
Excel
l spreadsheets:
d h t An A alternative
lt ti method
th d off assessmentt
Moira
M i Sarsfield
S fi ld andd Steve
St Cook
C k
Department of Life Sciences
Sciences, Faculty of Natural Sciences,
Sciences Imperial College London
London.

ABSTRACT THE AUTOGRADING PROCESS

In the first year of the biology degree stream at Imperial we now use three different The autograding process was developed by Dr Steve Cook
Cook, and it works as follows:
forms of assessment for practical classes: • Students
St d t download
d l d a spreadsheet
d h t template
t l t from
f Bl kb d (A).
Blackboard (A) The
Th
• online q quizzes in Blackboard;; template
p can include calculations,, graphs,
g p , MCQs,, short answer q questions,, and
• autograding using Excel spreadsheets; and more.

• traditionally marked write-ups

write ups, usually prepared by the student in Word
Word. • They complete the spreadsheet, following the instructions given in the
practical schedule ((B)) and on the spreadsheet
p p itself and upload
p the completed
p
version into the Blackboard assignment dropbox.
dropbox
The drivers for the introduction of the new assessment methods were:
• to deal with increasing numbers – from 110 to 220 students per class – • The tutor constructs a model answer file ((C),
), which includes information on the
meaning many more papers to mark; correct answers. These may be simple comparisons, Excel calculations based
on data in the spreadsheet or regexes.
regexes General feedback and scores for
• to improve the quality of feedback,
feedback making it more extensive
extensive, more consistent correct answers
ans ers are also incl
included
ded in this file
file.
and more quickly delivered
delivered.
• All of the completed spreadsheets are downloaded from Blackboard in ZIP
format
format.
This poster describes the autograding method which is particularly useful for
assessing the ability of first year students to analyse and present data from practical • A Perl script grades all the spreadsheets in the ZIP file by comparing the data
classes
classes. with the model answer
answer.
• Marks and feedback are inserted automatically
y into a copy
py of each student's
The autograding
g g process
p is described,, and examples
p are p
presented to highlight
g g spreadsheet (D),
(D) which is renamed with their username
username.
educational and technical features.
features Extensions and alternative uses of autograding
• Ag
gradebook is p
prepared
p ((E)) showing
g overall results from all students.
are discussed
discussed, and details are provided on the reactions of staff and students to this
new assessmentt method.
th d • Results are returned via Blackboard: general feedback (F),
(F) individual grade
and link to individual marked spreadsheet including feedback.

The
h spreadsheet
p d h is p
protected
d so that
h the
h students
d The
h model
d l answer checks
h k the
h student’s
d ’
can only
y enter data into the appropriate
pp p cells. answer, using
g the student’s own data. Numerical
Full comparisons,
p
instructions regexes
g and
are provided
p ffunctions can be
and each used
s within the
spreadsheet
sp s iss model answer.
colour coded
in the same
way so that
students
become
Boolean logic
g can
accustomed to
be used in the
the
model answer – e.g.g
requirements
requirements.
ANDD (&&), ORR (|).
(|)

The program can

Students must be run once
calculate
l l t without marking
th i answers
their to ascertain class
using
i E
Excell Electronic assessment data for use
Follow the instructions in the Excel spreadsheet. For Biuret and Bradford, use your data to perform linear regressions of
f
formulae,
l BSA mass (x) vs. absorbance (y) and calculate the slope (m), y-intercept (c) and the correlation coefficient (r). Use these
within the model
which
h h is a parameters to calculate the mass of ovalbumin in the assay tube, and from this, the ovalbumin concentration in the answer.
original
i i l UNKNOWN
UNKNOWN. The Th fi finall result
lt should
h ld be
b th
the protein
t i concentration
t ti iin th
the original
i i l UNKNOWN iin mg mL L−11.
very useful
f l
transfer-able
skill.
Statistics for
f
eachh question
are p
produced in
the gradebook
g
allowing g a check
to be made on
the overall
results. Common
errors can also be
noted and
general feedback
provided on
these.
these

Marks and general

g feedback
f are automatically y inserted onto a copy
py off The gradebook
Th d b k collates
ll details
d l off
the sp
spreadsheet
s which iss returned to the sstudent and further
f feedback
f iss the
h marksk and
d answers for
f eachh
provided for
p f all students
s s via the discussion
sc ss forum.
f student.

EXTENSIONS AND ALTERNATIVE USES REACTIONS OF STUDENTS

Paragraph
g p answers The autograding
g g system
y can also be used to assess A survey
y of first year
y students showed that 90% p preferred autograding
g g
paragraph answers which are manually marked.
marked In this case
case, the marker gives a spreadsheets to paper write
write-ups
ups. The reasons given for this preference were:
mark and feedback for each paragraph answer within the gradebook and then this • Lots of feedback – 60%
i f
information
ti is
i copied
i d back
b k into
i t each
h student’s
t d t’ spreadsheet.
d h t M Markers
k commentt th
thatt
this improves the consistency of their marking, as they can cross-check marks, and • Quick turnaround – 60%
that they can easily provide more extensive feedback for each student, as text fields • Transferable skills – 60%
are automatically completed within Excel.
Excel
Peer review Spreadsheets were used on the Biochemistry degree stream to gather The only negative comment made was of ‘harshness’ (20% of respondents).
peer review grades and comments on poster presentations
presentations. The data was then
collated into a gradebook using the autograding program,
program and individual feedback REACTIONS OF STAFF
was provided.
Staff
St ff appreciate
i t that
th t the
th autograding
t di process makesk markingki quicker,
i k simpler
i l and d
Collection
C ll ti off d
data
t with
ith validation
lid ti Th autograding
The t di process was used d to
t collect
ll t more objective.
j Well-written general
g feedback usuallyy covers most cases,, and
student choices for final yyear options
p on the Biomedical Science degreeg stream. individual clarification can be provided where students have further questions.
questions
The options available are very complex, and we wanted to ensure that students
provided all required information and didn’t
didn t select impossible combinations
combinations. This The turnaround time can be rapid,
p , which is useful when results of one p
practical feed
required data validation which was achieved within a data-collection
data collection spreadsheet
spreadsheet. into a subsequent practical
practical.
Ch i
Choices were collated
ll t d iinto
t a single
i l spreadsheet
d h t usingi th
the autograding
t di program. It is important that this is not the only assessment method used. Writing up a
F ili it with
Familiarity ith autograding
t di makes
k these
th methods
th d very easy for
f students
t d t to
t use. traditional lab report is still a required skill which must be practised and assessed
assessed.