Examination of Witness (Questions 220
- 239)
MONDAY 7 APRIL 2003
PROFESSOR CAROL
TAYLOR FITZ-GIBBON
Jeff Ennis
220. A particular hobbyhorse of mine, Professor
Fitz-Gibbon—and I think you are alluding to the fact that
this is something we ought to focus on—I always believe
that whatever the educational institution, you need to have a
good social mix in that educational institution for it to be successful
or for it to have the potential to be more successful. Would you
agree with that?
(Professor Fitz-Gibbon) I would want
to see evidence, and I was very interested in reading the evidence
presented by Sally Tomlinson who referred to an experiment done
in Stockholm. I did not have time to go to the library unfortunately,
and I want to get Torsten Husén's biography and read about
that, because I do not know why I had not heard about it before.
That was a randomised control trial, as I understood it, of having
schools that were segregated, or selective schools, versus comprehensive
schools. They came to the conclusion in Stockholm that the comprehensive
model worked best. I would believe it on those grounds, and I
would like to read about it.
221. Going back to the chart, you seem to be
advocating more vertical grouping in the secondary school setting.
Obviously, it happens more in primary schools at the moment, where
you get vertically grouped classes.
(Professor Fitz-Gibbon) Yes.
222. Do you think that that would make a difference
in terms of improving achievement at secondary level?
(Professor Fitz-Gibbon) Well, I think we should find
out. I think also that a very good reason for using cross-age
tutoring is that children enjoy it. Students enjoy it. We do need
joy in work: this is their childhood; this is not just a matter
of meeting government targets. This is their childhood, and they
should enjoy it. The fact that they like learning by teaching
should encourage us to use it. Also, it makes them nice to each
other and very helpful.
223. I guess that same philosophy supports the
need to integrate more special educational needs pupils into that
setting.
(Professor Fitz-Gibbon) I think inclusion is basically
a good idea, but if we want to be experimental about it, then—for
example, let us look at gifted students, at the other end. Is
it justifiable to identify some students as gifted and give them
a special programme? You might argue that it is cost-effective
because they need to go fast and they need to be together. So
let us draw a borderline; but within the error of measurement
around that borderline, randomly, put some into the gifted programme
and others not—this is totally ethical because it is all
errors and measurement in the borderline—and then follow
up the progress and see if actually some children hate being labelled
"gifted". "No, I am not", they say; they do
not want to be labelled "gifted". Others may relish
it. We have to do thousands of experiments to find out what works.
We cannot just argue the results. We need to find out with good
evidence, and then have more experiments to design interventions
that do work. But we need to measure a lot of outcomes, not just
achievement. Childhood is not just about achievement. What do
you remember when you leave school?
Chairman
224. One of the things that one remembers is
actually the school not as an individual but as part of a social
system in the school. Does your methodology not tend to pinpoint
just the individual achievement rather than—one of the things
we found when we looked at a range of schools, in a number of
cities, was that it does matter that if you have a school which
is a social system of kids coming from relatively low-income families,
with a background of not just low achievement but something more
complex—no tradition of educational achievement from their
parents or the people they know who live in their street and in
their community—there is a sense in which one never gets
the lift in those educational aspirations and achievement. There
is a social system. Are you not always looking at the individual,
and you cannot actually evaluate the social system of the school?
(Professor Fitz-Gibbon) Well, we need
to look at a lot of aspects. We do evaluate social systems. We
get measures of racism and of alienation—and those vary
from school to school quite substantially.
225. Levels of alienation are not the same as
saying that this is a school in an old mining community or an
old ship-building community, where none of their parents have
ever gone to higher education, where no-one ever had any aspirations
for education at all; there are no books in the homes, or very
few, and no parents reading to a child at an early age. Is that
not a social system that your analysis is missing?
(Professor Fitz-Gibbon) It is a question of what you
can alter. As a teacher, the students come to you. You cannot
go back into their homes and change their homes; it is what you
do with them in the time they come to school. They should not
be doing homework in primary schools. It does not matter; it does
not make any difference; it does not improve their achievement.
It is what we do in school, and that must be aimed at making them
happy, kind, nice individuals, and good parents subsequently by
learning social skills as well as cognitive skills.
226. So the evidence we took when we did the
Early Years inquiry about the importance of early years stimulation,
reading to children and playing games, creative play—you
think before they get to school that does not matter.
(Professor Fitz-Gibbon) I think a lot of these early
childhood interventions need replicating. The one that has been
most strongly promoted . . .
227. SureStart?
(Professor Fitz-Gibbon) No, that is a new one, but
the one from the States that SureStart was probably modelled on.
The name escapes me just now.
228. HeadStart.
(Professor Fitz-Gibbon) HeadStart was one, yes, but
that did not produce such strong evaluations. There was the—
Valerie Davey
229. HighScope.
(Professor Fitz-Gibbon) Yes. As someone
said at one of our evidence-based conferences: "Why has that
not been replicated?" The initial data that was so well promoted—and
the students were followed up—were 64 students. That was
hardly a large sample. But at every stage of schooling, the schooling
should be as good and as much fun as possible, not just pre-school—or
we will lose them later.
Chairman: This is all very interesting,
but we must move on. I will ask David to look at league tables.
Mr Chaytor
230. To return to YELLIS for a moment, your
argument in favour of that is that it provides constancy because
it is the same each year, unlike Key Stage 3; but why does anybody
buy it if it is the same each year? Why do they not just photocopy
it previously?
(Professor Fitz-Gibbon) I suppose they
could.
231. Is it exactly the same test year after
year after year?
(Professor Fitz-Gibbon) Yes, but it is a lot of work
to mark tests and then they need to be nationally standardised,
so you need the national standardisation. We have a fantastic
team of very clever youngsters running computers. They do all
the matching and they do the value-added calculations; so the
teacher gets a huge amount of work for nothing, plus some free
software, plus conferences, plus help on the telephone and so
on. Uniquely in the world, schools are supporting 65 staff in
Durham, the biggest research group in UK education[2].
232. But you said earlier that YELLIS was primarily
for predicting GCSE scores; and now you have referred to its value
as a measure of value-added. Does it work equally well as both?
(Professor Fitz-Gibbon) If it did not
have good predictive validity, it would not be a good measure
for value-added; so it has to predict well.
233. How is it different from the methodology
that the DfES uses for its value-added calculation, which just
draws the difference between Key Stage 3 point scores and GCSE
point scores? What is your criticism of the methodology that has
been used in the value-added scores that were published last year
for the first time?
(Professor Fitz-Gibbon) If we can look at figures
3a distribution 3b: that refers to GCSE predicting A-Level, but
could just as well refer to Key Stage 3 predicting GCSE. What
we see is that each subject has a different line necessary, and
there is no way any single line could represent those adequately.
So where the DfES has gone wrong is in treating an A as an A as
an A, no matter whether it is in physics or communication studies.
It just is not true that the same challenge is in both subjects.
234. Your criticism is entirely based on the
variability of the grading in each subject, because in your submission
you make an issue of the contrast between physics and sociology
for example.
(Professor Fitz-Gibbon) They are graded so they give
a similar distribution . I have no problem with that. But you
must not then evaluate the school. If they put a lot of students
in for an easier subject they should not go up the league tables
as value-added.
235. Is it not conceivable that the value-added
achieved in sociology is greater than the value-added achieved
in physics? Is that not an equally valid conclusion that you could
draw?
(Professor Fitz-Gibbon) On what?
236. As I understand it, your argument is that
because the levels of difficulty of individual subjects vary—and
you quote sociology whereby the cohort entering sociology, by
and large, has a lower level of performance than that for taking
physics—but could not the reason for broadly the same level
of grade distribution be that there is more value-added generated
by sociology teachers? Is that not an equally valid conclusion?
(Professor Fitz-Gibbon) No, I think it is most unlikely.
You cannot compare the teaching of physics and sociology. We can
only look at the data and ask how difficult it is to get a child
through physics or through sociology. Most teachers would agree
that one was harder than the other. We can then see the effect
of calling them all the same. The maths enrolment has declined.
It is not easy to get people through A-Level maths and so on.
It is a matter of comparing like with like, and that is where
the DfES is unfair to schools. When I did a re-analysis of the
data of schools that were told that they were below average, they
had put students in for the difficult subjects in the same way
the others had. But the other schools that were told they were
above average, when in fact they were not on our calculations,
had put students in for a lot of easy subjects. That was having
an impact. The capping may have reduced that effect a little bit,
but it is still a very unfair system that does not compare like
with like. The first principle of a statistical model is that
it models the process that produces the data. The process that
produces each of these regression lines is the examining in that
particular syllabus; and that is what should be modelled statistically.
237. If there were absolute parity or absolute
equivalence of grading between the subjects, would the method
of calculating the point scores be acceptable? Do you think it
is a sound method in itself?
(Professor Fitz-Gibbon) I do not think what the DfES
does is sound. The grading of the subjects is perfectly sound
and I have no problem with the same distribution in sociology
as in physics. It is just let's not kid ourselves that it would
be good for the country to move everybody out of physics so they
would do well in sociology.
238. That is not the point I am trying to get
at. If the way in which the grading of physics and sociology,
for the sake of argument, were done that satisfied to you that
it accurately reflected the levels of difficulty, do you think
the DfES's method of measuring the progress from Key Stage 3 to
Key Stage 4 is adequate, simply on averaging of the point scores?
(Professor Fitz-Gibbon) It would be enormously difficult
to constrain them to be exactly the same level of difficulty.
The Scottish Office asked me this in 1990, they said, "What
would happen if all the subjects were equated and given a distribution
of grades that reflected aptitude?" No-one would have got
a fail in physics and nobody would have had an A in art. You should
be able to use the full range in every subject. Everybody knows
that they are different subjects and of course the distribution
of aptitudes will vary from subject to subject. So some students
in physics might be absolutely tongue-tied in sociology. However
those who have A-level maths have higher verbal scores than those
with A-level English, on average, so it is not a simple thing.
Then the subject should be taught in a way that is appropriate
for the group of students that take that subject, and it should
be graded to give a reasonable distribution. The analysis must
not penalise schools that put students in for tough subjects because
the country needs people to do maths, sciences and foreign languages.
Students who have A-level maths earn 10% more than students who
have not taken A-level maths and are doing the same job.[3]
We must not reward schools for moving people out of those subjects.
239. On GCSEs you make the point about the five
A-Cs and indicate that they are not being very helpful. What is
the issue there? You do not think that accurately reflects the
schools' achievement? How else could it be done? If it were five
A-Gs at GCSE, would that be more accurate?
(Professor Fitz-Gibbon) No, the average
points score.
2 Note by witness: In universities. Back
3
Note by witness: Report from LSE. Back
| 
