Examination of Witnesses (Questions 60-80)
DR JOHN
GOLDTHORPE, PROFESSOR
STEPHEN MACHIN
AND DR
JO BLANDEN
23 JANUARY 2008
Q60 Mr Chaytor: In terms of the schooling
system. You commented on diversity and expressed some scepticism
about that with regard to an increase in social mobility.
Dr Goldthorpe: Yes.
Q61 Mr Chaytor: I am interested in
the issue of hierarchy, of a system of primary schools and secondary
schools that are structured in a hierarchical and highly selective
way. What conclusions do you draw from that? Secondly, do you
think there is any relevance to the issue of social mobility in
the system of testing and assessment we have? That is the subject
of another inquiry that the Committee is involved in at the moment.
Do you think that an intensive testing system in which certain
groups of children are repeatedly failed has any relationship
to the level of social mobility or not? That question is for Dr
Goldthorpe, to start with.
Dr Goldthorpe: On testing, I have
to say that I have not done research in this area and I am not
very expert about consequences of it. My gut feeling, for what
it is worth, is that testing has become overdone, and there is
evidence that exposing children to the possibility of repeated
failure is not good for their academic confidence and performance.
On school systems, I generally favour comprehensive systems. If
we go back to considering the Nordic countries, accepting that
they are quite small and perhaps peculiar countries, they do have
fairly well developed comprehensive systems and there is some
evidence that that has played a part in the fact that they have
lower class differentials in educational attainment than we have
in this country. That is comprehensive education combined with
what in British parlance we would call fairly extensive setting
at older ages. That seems to me the best way of avoiding the essential
arbitrariness of earlier selection, while still making adequate
provision for children of differing aptitudes in different directions.
In general, I am in favour of what you call flatter rather than
more hierarchical educational systems.
Professor Machin: On one aspect
of the hierarchy, there is some evidence that countries that have
earlier tracking in the education system have higher educational
inequalities. Germany is a very clear example. They track at age
11 into the three
Q62 Mr Chaytor: Earlier trackingearlier
selection?
Dr Blanden: Yes. Selection.
Professor Machin: Yes. Putting
people on a track at an earlier age. There is evidence that that
leads to educational inequalities, largely because people are
not able to move across tracks very much, so that is having some
impact on mobility in the early years and then subsequently
Q63 Chairman: Decreasing or increasing?
Professor Machin: It increases
inequalities, because it restricts people to being on a particular
track. On the testing regimes, I do not know of any evidence that
particularly links the presence of a national testing regime to
aspects of social mobility. You could make theoretical arguments
either way, I would have thought. If everybody takes the same
national tests, that is at least levelling the playing field so
that everybody has the potential to do quite well, rather than
selecting people according to certain examinations that they may
or may not take, which was the old system, certainly for O-levels
rather than GCSEs. If we want to make the talent argument and
if people are not constrained by other things, presumably letting
everybody have a go is quite a good thing. Whether that is placing
much more stress on people from different social backgrounds is
an open question. Again, I do not know. I do not know of any research
that has looked at that in any detail and I cannot quite work
out how you would be able to do very good research on that question,
notwithstanding the fact that it is an important question. It
is not one I know much about myself.
Dr Blanden: I just wanted to follow
on from the testing point. High-stakes testing throughout the
schooling system may encourage parents to exercise their choices
more at an earlier age. To that extent, it could be associated
with selection by postcode. It could be promoting that to a certain
extent, but that is only a theoretical thought that I had just
as Stephen was speaking.
Q64 Mr Chaytor: May I ask one other
question? It relates to your conclusion about the impact of access
to the expanded higher education system being disproportionately
monopolised by young people from middle-class families. Surely
the real test of the value of an expanded HE sector will come
only when we see the entry into the job market of a broader base
of HE entrants. Inevitably, when the system expanded from the
mid-1980s, it was the group of middle-class young people who had
done quite well at school but had not previously gone into HE
who gobbled up the opportunities, because they were best placed
to do so. Surely we will have to wait until the post-1997 cohort,
which is a much wider social base of graduates, enters the labour
market to see whether the expansion has had any significant effect
on social mobility. Is it not too early to say whether there is
an effect?
Professor Machin: I would mostly
agree. Some of the cohorts who benefited from it in the early
stages entered the labour market, but to evaluate the whole expansion,
yes, we probably need to wait a while.
Q65 Mr Chaytor: So when do you think
we will have reliable data about the real effect on social mobility
of a significantly expanded HE system?
Dr Goldthorpe: The next cohort
after the 1970 one is the 2000 one, so we will be in the 2020s
before we can see that. But something could be done if those running
the General Lifestyle Module (GLM), formerly the General Household
Survey, could be persuaded to reintroduce questions that would
help us with social mobility. They would not have to do it every
year, just once every three or five years. Another thing that
is very important in Government statistics generally is that we
need to break down the unitary category of tertiary education
or degree. I know that would be invidious and create problems,
but we have to recognise that as tertiary education has expanded
it has become increasingly stratified. We need to make some distinction
between different levels of university, however controversial
that might besay, Oxbridge, Russell Group and then so-called
modern universities, with whatever is in between. One could discuss
that. We also need to collect information on the subject areas
in which people get degrees, because it is clear that returns
to higher education, whether in earnings or social class, will
become increasingly variable. If we want to evaluate the outcome
of the expansion of tertiary education, we must treat it in a
far more differentiated way than hitherto.
Q66 Mr Chaytor: So in respect of
the changes necessary to the GLM to provide the kind of evidence
that you need to draw significant conclusions, is there an agreed
set of questions that you or other academics have argued for?
One of the most saddening things about the background papers was
the change in 1993, which took out the question of social background.
Should that be restored?
Dr Goldthorpe: There is a big
difference. With the birth cohort studies, there are standing
committees of academics to discuss the questions. With the GLM,
we have little inputat least, as far as I know, not of
any formalised kind.
Q67 Mr Chaytor: Who determines the
questions that go into it?
Dr Goldthorpe: The Office for
National Statistics, and various Government Departments can
Q68 Mr Chaytor: This is a highly
politically charged issue, is it not? The questions that are being
asked are not delivering the relevant answers that academics need
to assess the impact of Government policy.
Dr Goldthorpe: I tried very hard
to find out why they stopped collecting the relevant data in 1993,
and I have never been able to get a satisfactory answer. It may
just have been an accident.
Q69 Chairman: So would you academics
like it reinstituted? Would you like this Committee to recommend
that that kind of data be reinstituted?
Dr Goldthorpe: Yes please.
Professor Machin: That is of direct
relevance to John's work, and I concurit would be a good
thing. It would be of high practical relevance to John's work
on social class.
Q70 Fiona Mactaggart: I have found
the papers that we have looked at very hard, which is relevant
to the discussion. The things that I know about statistics I mostly
read in a book published in 1954 that has been read by 1.5 million
other people, by Darrell Huff, called How to Lie With Statistics.
His first chapter relates to the built-in bias in a sample. He
refers to the 1924 class of Yale graduates who earned $25,111
a year. He pointed out that in that group of people, those who
earned the least had fallen out so that the stayers were likely
to be higher earners. He pointed out that it was not clear from
the way the questions were asked what the difference was between
salary and other income. He also suggested, with some evidence,
that you cannot trust people when they report income. Every single
point seems to be relevant to your work, but is not addressed
in it.
Professor Machin: All the things
you said are part of our day to day research. We have to face
all such questions when we analyse survey data and look at who
responded to the surveywhether they are people in the labour
market, employers or consumers. Those are the kinds of things
that we have to look at when we do quantitative statistical research.
We are very aware of all of those points. We spend a lot of time
in our research papers concentrating on whether our samples are
representative, whether there is a measurement error in people's
responses to the survey, what the framing of the questions is
and whether the questions are framed in the same way across cohorts
or whether there are discrepancies in the way questions are asked.
We look at all that and many other minute details when we do research.
Things have moved on since the 1950s. We have various statistical
techniques for dealing with issues such as measurement error.
Q71 Fiona Mactaggart: But these are
very specific measurement errors. They are not classic measurement
error. Such errors will produce bias in your results.
Professor Machin: They may produce
bias.
Q72 Fiona Mactaggart: They will,
because we know that the people who fall out of these long cohort
studies are absolutely most likely to be the least prosperous.
The 2,000 missing people from the figures in 1958 are likely to
be the poorest. The odds on that are massive.
Professor Machin: Yes, we know
that. I would not have said that the odds were massive. We have
various tables in various papers that we have written in which
we start with a full sample. We work down on the sample restrictions
that we make to end up with the sample that we look at. Inevitably,
people are lost. Even when people respond to a survey, some do
not answer all the questions. If we use a sample of x thousand
people out of an initial sample of y thousand, it is very important
to say how we have got from the representative sample of y thousand
down to x thousand. We spend a lot of time in our research being
very careful about that. It is true that it is more of an issue
in some data sets than other data sets. In certain papers, we
write with different data sets and we have to look at that. That
is something that we are very aware of in our research and we
have to very rigorous about it. We have to assess the robustness
of our results. Sure, people can lie with statistics. They can
also tell very representative stories about what is going on as
well.
Q73 Fiona Mactaggart: Do not think
that I am accusing you of intending to lie. What I am concerned
about is this whole business of the conversation between politicians
and researchers. The conversation is, in my view, a pathetic one.
The only politician who is cited in any of this workapart
from quoting politicians who quote youis Dr Goldthorpe's
reference to Ed Miliband looking at some of his data. It seems
to me that there is a failure to converse and a failure to understand,
which leads politicians sometimes to lay greater weight on some
of your conclusions than perhaps even you would. Of course, you
want people to use your conclusions. That is quite important if
you are going out looking for research grants from elsewhere.
Partly because of this conversation failure, we can be wrong about
what the data show. I heard David Cameron cite your work to imply
that, under the Labour Government, social mobility had massively
reduced, which nobody who has read it carefully would claim. Am
I right?
Dr Blanden: Yes. In fact, I was
involved with the Channel 4 spotting of mistakesI am not
sure what it is calledand picking up on Cameron's exact
quote and trying to set it right. We do trythat is the
best we can say. That is why we have tried to follow it up. In
terms of attrition and measurement error, we are okay if neither
of those things change substantially between the two cohorts.
To a certain extent, we can never know absolutely if those things
have not changed.
Q74 Fiona Mactaggart: Exactly; you
have to guess. There is no magic way.
Dr Blanden: It is not guessing.
Professor Machin: Indeed. We have
to be as scientific as we can from a research perspective. That
is not guessing. Guessing would be someone just saying "Oh
yeah, it looks representative," or "No, it doesn't."
That is not what we do; we spend an enormous amount of time trying
to be as scientific as we can in doing the research, otherwise
there would be no point in doing it.
Q75 Fiona Mactaggart: I understand
that. I am trying to point out that one consequence of that kind
of research is that people put weight on some examples that they
do not bear outI have cited one such example with which
Dr Blanden agreedand that, because you cannot know certain
things, you have to extrapolate using the best evidence that you
can in the circumstances. I called it guessing, and I might have
been exaggerating, but I did that for effect. You cited some other
work about the impact of Head Start and early years work. Some
of the work that you cited was presumably based on a quite controlled
experiment with many fewer people than the big panels that you
usedthe High/Scope Perry research for example. In such
experiments, the same individuals, not one panel in 1958 and another
in 1970, are tracked over time and the differences between the
different groups are analysed. Is that a better way of studying
these issues?
Professor Machin: There are two
points that I should like to make in response to that. First,
returning to your point about dialogue between policy makers and
researchers, there is a potential conflict of interest, because
policy makers have different questions from academic researchers.
Sometimes they coincide, and sometimes they do not. That is why
people might have different incentives for interpreting results
in different ways. Closer consultation on that is a good thing
in order to minimise that conflict of interest. On the second
point about experimental versus non-experimental data, that is
something that we face a lot in the social sciences. We do not
often have experiments that we can run in the same way as those
in the medical sciences. However, we try to approximate them to
the best of our ability, using non-experimental data to evaluate
particular questions of interest. There are no experiments out
there on social mobility. We cannot randomise somebody into a
particular family as they are growing up and compare them to a
family in which someone is not randomised into a higher or lower-income
or higher or lower social-class family. So we must do what we
can with non-experimental data. There are many studies out there
comparing experimental outcomes with non-experimental outcomes
in particular areas to see whether that matters and how you might
develop methodologies to get closer to the experimental kind of
studies. I would argue that we should still look at aspects of
social mobility, whether income mobility or social-class mobility,
even in the absence of experimental data. We should be doing that
and trying to do the best job that we can with non-experimental
data.
Dr Goldthorpe: I think that you
put your finger on a very serious mattercommunication between
politicians and social scientists. I would hope that thought could
be given to that. However, on a more positive note, it is natural
that politicians should, from time to time, try to make a political
point from a piece of research that appears. That is their job.
It is important, however, that they do not necessarily always
believe their own propaganda. They should be ready to talk to
researchers about this in a different mode. I think that a fair
bit of that does, in fact, go on. I have discussed our research
with Ed Miliband and other people in the Government, but I have
also had several long conversations with David Willetts about
it. At that level, communication is better.
Q76 Chairman: To build on Fiona's
point, is not there a problem in the way that your research has
been picked up? As I have said, I admire the work of the Sutton
Trust, particularly the work that it has done on the concentration
of independently educated people in the BBC, major corporations,
journalism and in almost every walk of life. That complements
the work that you have done on social mobility, which is very
thought-provoking and important for politicians to know about.
We will pick up on the bits we like: if I raised the concentration
of independently educated people at the BBC, I would immediately
be hounded by the press unless I pointed to that piece of Sutton
Trust work. There are interesting points for us because we spend
our time in this Committee trying to scrutinise the work of Government
and whether the spending of taxpayers' money is good value. One
of the things that we have asked over a number of years is whether
something is evidence-based policy. For that, we very much rely
on you people. Coming out of your work on social mobility, there
is one strand that says, "Oh, all that SureStart stuff, free
nursery care and all that has all been a failure and has not worked",
as Fiona would say. Some policy makers have then said, "Well,
it's not worth doing anything because you cannot make any difference
through educational policy to social mobility." Others, such
as the Sutton Trust, have said, "You have got to renew and
be more vigorous in the way that you give more kids the opportunity
for a broader and better education." Therefore, the policy
implications of your work are quite diverse, are they not?
Professor Machin: Yes. As academic
researchers, I think that we can take two positions on that, one
of which is fairly productive and one of which is less so. We
could say that we have done the research and whoever wants to
use it can. On the other hand, we could engage in dialogue with
people who are interested in it and perhaps go into much more
detail on that in the way that we are doing in this Committee,
which I think is very useful, so that we can actually come to
a closer agreement on what the research findings mean and what
their relevance is for policy. Many people take that first view:
the research is done in an academic ivory tower and then if anyone
wants to pick up on it, that is fine, but it has been done and
published in a leading academic journal and is over and done with.
I would prefer to think of the second position as being much more
useful with regard to the impact that academic research can have
in the policy arena.
Q77 Chairman: I think that most people
in this Committee think that your research has been very useful.
Given you responses today, I would like to mention the importance
of setting research in context. I know that I was nasty about
some Nordic countries, but I have a thing about Finland. However,
we were comparing the UK, like with like, with big, mature, industrial
democracies. I got the feeling that, in one of your answers, there
were slight throw aways: you said France is different. My intuition
about the history of France is that it becomes a free-market,
industrial, mature democracy much later than the UK, and the UK
comes much later than the United States. If you are going to put
your research into context, is there a sense that some of those
societies mature in a different way? In a sense, the context is
lost when you look only at that snapshot piece of research.
Dr Goldthorpe: I have done quite
a lot of comparative work with various colleagues on social mobility
and think that you are absolutely right. Different countries follow
different historical trajectories of development and have very
different institutional forms. France is an interesting case in
that regard, because it used to have a fairly rigid society, in
terms of both class differences in education and social mobility.
As I mentioned, it is one of the societies that has most clearly
become more fluid, having starting off from a low base. One possible
explanation that my French colleagues are working on is that that
comes from the decline of the agricultural sector, which was particularly
rigid and continued to be quite a sizeable sector in France until
much later than it did in this country. It may well be that that
is the source. Also, France has a highly centralised education
system. When French Governments want to change educational policy,
say in the interests of reducing class differentials, they have
more powerful levers than exist in this country.
Chairman: Graham, as usual, wants the
last word.
Q78 Mr Stuart: You have looked at
the combination of class mobility and income mobility. If you
look at it in terms of educational outcomes at certain key stages,
how strong is the link between educational outcome and changes
in either class or income mobility over time?
Dr Goldthorpe: In my experience
there is not a perfect but a clear association. Those countries
that have been most successful in reducing class differences in
educational attainment most often show an increase in fluidity.
Q79 Mr Stuart: When we had our great
time of change in the post-war world, was there a transformation
in educational attainment among lower social classes?
Dr Goldthorpe: No. In Britain,
the evidence on the relationship between class and educational
attainment is remarkably constant. Of course, more children from
all class backgrounds have gone on to higher secondary and tertiary
education, but the differentials have remained remarkably constant.
We do not find any widening in class differentials in the way
that Stephen and Jo do in relation to family income. We find just
a constancy and that is rather disturbing.
Professor Machin: We do find evidence
that the relationship between educational attainment and family
income has strengthened in terms of attainment at higher education
and family income and in terms of early age test scores and family
income. One of the nice things about the cohort data is that the
children were tested in maths and reading as they were growing
up. We find a stronger relationship for the second cohort than
the first.
Q80 Mr Stuart: So a system of early
intervention to help lift up those who are falling behind, however
it is done, will be effective.
Professor Machin: Yes. If you
look at the latest cohort study, the MCS and the relationship
between age five test scores and family income, there is a very
strong gradient between the two. That was in 2005.
Chairman: A last word from Jo Blanden.
Dr Blanden: It is not a very good
last word.
Chairman: That is all right.
Dr Blanden: One thing that I have
noticed from looking at a lot of the sociological work in the
run-up to this is something that appears in John's workI
think he would agreein that the relationship between educational
background and social class seems to be falling. However, we do
not find that there is any fall in the relationship between education
and earnings at all. That might be another component of why we
find this slightly differing picture. That was a bad last word.
Chairman: No, it was not. This has been
more of a seminar than a regular question and answer session.
You have educated us wonderfully. We have enjoyed the experience
and will use your words today as we look further at social mobility.
We would be grateful if we could keep in touch with you. Thank
you very much for your time.
|