Home' Teacher : August 2011 Contents curriculum & assessm ent 29
All items had already been ‘m arked’ against
the distribution described above and shown
to be from the same absolute distribution, yet
when final grades were calculated, the dis-
play of letter g rades had a statistical impact.
I’m the first to raise the caveat that these
are the results of a pair of teacher’s marks.
It would be great to repeat the experiment
with a larger sa mple, or cross -subjects, to
gain empirical evidence about the efficacy
of this framework of assessment.
So what does it all mean? In short, nu mer-
ical marking and grading in all likelihood is
completely fine, since this seems to be what
teachers do whether or not they’re permit-
ted to use numbers. The anecdotal evidence
from students and parents alike is that they
understood percentages and nu mber g rades
but they can’t make head nor tail of matrices
of several hundred written words, despite
attempts to render them i n plain E nglish.
My concern is that experienced teach-
ers using nu mbers to identify and calculate
achievement were able to judge student per-
formance, with very low variance between
teachers in a subject area. My discussions
with scores of teachers using matrices reveal
that marking variance between teachers is
now quite a bit higher. Indeed, 10 colleagues
and I with a mean teaching experience of 11
years, looking at one student profile, showed
variance from C- to B, all of us arguing with
passion and reason. This variance of 7 to
10 on a 15-point scale is an error of exactly
+/-10 per cent. I n contrast, similar exercises
under previous assessment regimes which
used percentage thresholds and nu merical
ma rks for assessment were able to achieve
precision within a third of a band from A+
to E- or +/- 3 per cent error.
It seems we have teachers out there, in
panel-based monitoring and verification
systems, u nwilling to blink on their meth-
ods of generating g rades that will eventually
calculate student university entrance indi-
ces, which may be disadvantaging students.
Importantly, it ought to be quite easy to
arra nge for quantitative analyses of the pre-
cision of various models of assessment using
real teachers and real students. If it’s not bet-
ter for learni ng and assessment , there’s a very
real question: what are we doing it for? T
Mick Wilkinson is a Queensland teacher.
Links Archive June-July 2011 Navigation Previous Page Next Page