Content uploaded by Ben Goldacre
Author content
All content in this area was uploaded by Ben Goldacre on Oct 26, 2015
Content may be subject to copyright.
BUILDING EVIDENCE INTO EDUCATION
BEN GOLDACRE
MARCH 2013
2
3
Background
Ben Goldacre is a doctor and academic who writes about problems in science
and evidence based policy, with his Guardian column “Bad Science” for a
decade, and the bestselling book of the same name. He is currently a
Research Fellow in Epidemiology at London School of Hygiene and Tropical
Medicine.
To find out more about randomised trials, and evidence based practice, you
may like to read “Test, Learn, Adapt”, a Cabinet Office paper written by two
civil servants and two academics, including Ben Goldacre:
https://www.gov.uk/government/publications/test-learn-adapt-developing-
public-policy-with-randomised-controlled-trials
4
6
7
Building evidence into education
I think there is a huge prize waiting to be claimed by teachers. By collecting
better evidence about what works best, and establishing a culture where this
evidence is used as a matter of routine, we can improve outcomes for children,
and increase professional independence.
This is not an unusual idea. Medicine has leapt forward with evidence based
practice, because it’s only by conducting “randomised trials” - fair tests,
comparing one treatment against another - that we’ve been able to find out what
works best. Outcomes for patients have improved as a result, through thousands
of tiny steps forward. But these gains haven’t been won simply by doing a few
individual trials, on a few single topics, in a few hospitals here and there. A
change of culture was also required, with more education about evidence for
medics, and whole new systems to run trials as a matter of routine, to identify
questions that matter to practitioners, to gather evidence on what works best,
and then, crucially, to get it read, understood, and put into practice.
I want to persuade you that this revolution could - and should - happen in
education. There are many differences between medicine and teaching, but they
also have a lot in common. Both involve craft and personal expertise, learnt over
years of experience. Both work best when we learn from the experiences of
others, and what worked best for them. Every child is different, of course, and
every patient is different too; but we are all similar enough that research can
help find out which interventions will work best overall, and which strategies
should be tried first, second or third, to help everyone achieve the best outcome.
Before we get that far, though, there is a caveat: I’m a doctor. I know that
outsiders often try to tell teachers what they should do, and I’m aware this often
ends badly. Because of that, there are two things we should be clear on.
Firstly, evidence based practice isn’t about telling teachers what to do: in fact,
quite the opposite. This is about empowering teachers, and setting a profession
free from governments, ministers and civil servants who are often overly keen
on sending out edicts, insisting that their new idea is the best in town. Nobody in
government would tell a doctor what to prescribe, but we all expect doctors to be
able to make informed decisions about which treatment is best, using the best
currently available evidence. I think teachers could one day be in the same
position.
8
Secondly, doctors didn't invent evidence based medicine. In fact, quite the
opposite is true: just a few decades ago, best medical practice was driven by
things like eminence, charisma, and personal experience. We needed the help of
statisticians, epidemiologists, information librarians, and experts in trial design
to move forwards. Many doctors – especially the most senior ones - fought hard
against this, regarding “evidence based medicine” as a challenge to their
authority.
In retrospect, we’ve seen that these doctors were wrong. The opportunity to
make informed decisions about what works best, using good quality evidence,
represents a truer form of professional independence than any senior figure
barking out their opinions. A coherent set of systems for evidence based practice
listens to people on the front line, to find out where the uncertainties are, and
decide which ideas are worth testing. Lastly, crucially, individual judgement isn’t
undermined by evidence: if anything, informed judgement is back in the
foreground, and hugely improved.
This is the opportunity that I think teachers might want to take up. Because
some of these ideas might be new to some readers, I’ll describe the basics of a
randomised trial, but after that, I’ll describe the systems and structures that exist
to support evidence based practice, which are in many ways more important.
There is no need for a world where everyone is suddenly an expert on research,
running trials in their classroom tomorrow: what matters is that most people
understand the ideas, that we remove the barriers to “fair tests” of what works,
and that evidence can be used to improve outcomes.
How randomised trials work
Where they are feasible, randomised trials are generally the most reliable
tool we have for finding out which of two interventions works best. We simply
take a group of children, or schools (or patients, or people); we split them into
two groups at random; we give one intervention to one group, and the other
intervention to the other group; then we measure how each group is doing, to
see if one intervention achieved its supposed outcome any better.
This is how medicines are tested, and in most circumstances it would be
regarded as dangerous for anyone to use a treatment today, without ensuring
that it had been shown to work well in a randomised trial. Trials are not only
9
used in medicine, however, and it is common to find them being used in fields as
diverse as web design, retail, government, and development work around the
world.
For example, there was a longstanding debate about which of two competing
models of “microfinance” schemes was best at getting people out of poverty in
India, whilst ensuring that the money was paid back, so it could be re-used in
other villages: a randomised trial compared the two models, and established
which was best.
At the top of the page at Wikipedia, when they are having a funding drive, you
can see the smiling face of Jimmy Wales, the founder, on a fundraising advert.
He’s a fairly shy person, and didn’t want his face to be on these banners. But
Wikipedia ran a randomised trial, assigning visitors to different adverts: some
saw an advert with a child from the developing world (“she could have access to
all of human knowledge if you donate…”); some saw an attractive young intern;
some saw Jimmy Wales. The adverts with Wales got more clicks and more
donations than the rest, so they were used universally.
It’s easy to imagine that there are ways around the inconvenience of
randomly assigning people, or schools, to one intervention or another: surely,
you might think, we could just look at the people who are already getting one
intervention, or another, and simply monitor their outcomes to find out which is
the best. But this approach suffers from a serious problem. If you don’t
randomise, and just observe what’s happening in classrooms already, then the
people getting different interventions might be very different from each other, in
ways that are hard to measure.
For example, when you look across the country, children who are taught to
read in one particularly strict and specific way at school may perform better on a
reading test at age 7, but that doesn’t necessarily mean that the strict, specific
reading method was responsible for their better performance. It may just be that
schools with more affluent children, or fewer social problems, are more able to
get away with using this (imaginary) strict reading method, and their pupils
were always going to perform better on reading tests at age 7.
This is also a problem when you are rolling out a new policy, and hoping to
find out whether it works better than what’s already in place. It is tempting to
look at results before and after a new intervention is rolled out, but this can be
very misleading, as other factors may have changed at the same time. For
example, if you have a “back to work” scheme that is supposed to get people on
10
benefits back into employment, it might get implemented across the country at a
time when the economy is picking up anyway, so more people will be finding
jobs, and you might be misled into believing that it was your “back to work”
scheme that did the job (at best, you’ll be tangled up in some very complex and
arbitrary mathematical modelling, trying to discount for the effects of the
economy picking up).
Sometimes people hope that running a pilot is a way around this, but this is
also a mistake. Pilots are very informative about the practicalities of whether
your new intervention can be implemented, but they can be very misleading on
the benefits or harms, because the centres that participate in pilots are often
different to the centres that don’t. For example, job centres participating in a
“back to work” pilot might be less busy, or have more highly motivated staff:
their clients were always going to do better, so a pilot in those centres will make
the new jobs scheme look better than it really is. Similarly, running a pilot of a
fashionable new educational intervention in schools that are already performing
well might make the new idea look fantastic, when in reality, the good results
have nothing to do with the new intervention.
This is why randomised trials are the best way to find out how well a new
intervention works: they ensure that the pupils or schools getting a new
intervention are the same as the pupils and schools still getting the old one,
because they are all randomly selected from the same pool.
At around this point, most people start to become nervous: surely it’s wrong,
for example, to decide what kind of education a child gets, simply at random?
This cuts to the core of why we do trials, and why we gather evidence on what
works best.
Myths about randomised trials
While there are some situations where trials aren’t appropriate - and where
we need to be cautious in interpreting the results - there are also several myths
about trials. These myths are sometimes used to prevent trials being done, which
slows down progress, and creates harm, by preventing us from finding out what
works best. Some people even claim that trials are undesirable, and even
completely impossible, in schools: this is a peculiarly local idea, and there have
been huge numbers of trials in education in other countries, such as the US.
However, the specific myths are worth discussing.
11
Firstly, people sometimes worry that it is unethical to randomly assign
children to one educational intervention or another. Often this is driven by an
implicit belief that a new or expensive intervention is always necessarily better.
When people believe this, they also worry that it’s wrong to deprive people of the
new intervention. It’s important to be clear, before we get to the detail, that a
trial doesn’t necessarily involve depriving people of anything, since we can often
run a trial where people are randomly assigned to receive the new intervention
now, or after a six month wait. But there is a more important reason why trials
are ethically acceptable: in reality, before we do a trial, we generally have no idea
which of two interventions is best. Furthermore, new things that many people
believe in can sometimes turn out, in reality, to be very harmful.
Medicine is littered with examples of this, and it is a frightening reality. For
many years, it was common to treat everyone who had a serious head injury with
steroids. This made perfect sense on paper: head injuries cause the brain to swell
up, which can cause important structures to be crushed inside our rigid skulls;
but steroids reduce swelling (this is why you have steroid injections for a
swollen knee), so they should improve survival. Nobody ran a trial on this for
many years. In fact, it was widely argued that randomising unconscious patients
in A&E to have steroids or not would be unethical and unfair, so trials were
actively blocked. When a trial was finally conducted, it turned out that steroids
actually increased the chances of dying, after a head injury. The new intervention,
that made perfect sense on paper, that everyone believed in, was killing people:
not in large enough numbers to be immediately obvious, but when the trial was
finally done, an extra two people died out of every hundred people given
steroids.
There are similar cases from the world of education. The “Scared Straight”
programme also made sense on paper: young children were taken into prisons
and shown the consequences of a life of crime, in the hope that they would be
more law abiding in their own lives. Following the children who participated in
this programme into adult life, it seemed they were less likely to commit crimes,
when compared with other children. But here, researchers were caught out by
the same problem discussed above: the schools - and so the children - who went
on the Scared Straight course were different to the children who didn’t. When a
randomised trial was finally done, where this error could be accounted for, we
found out that the Scared Straight programme - rolled out at great expense, with
great enthusiasm, good intentions, and huge optimism - was actively harmful,
making children more likely to go to prison in later life.
So we must always be cautious about assuming that things which are new, or
12
expensive, are necessarily always better. But this is just one special case of a
broader issue: we should always be clear when we are uncertain about which
intervention is best. Right now, there are huge numbers of different
interventions used throughout the country - different strategies to reduce
absenteeism, or teach arithmetic, or reduce teenage pregnancies, or any number
of other things - where there is no evidence to say which of the currently used
methods is best. There is arbitrary variation, across the country, across a town,
in what strategies and methods are used, and nobody worries that there is an
ethical problem with this.
Randomisation, in a trial, adds one simple extra chink to this existing
variation: we need a group of schools, teachers, pupils, or parents, who are able
to honestly say: “we don’t know which of these two strategies is best, so we don’t
mind which we use. We want to find out which is best, and we know it won’t
harm us.”
This is a good example of how gathering good evidence requires a culture
shift, extending beyond a few individual randomised trials. It requires everyone
involved in education to recognise when it’s time to honestly say “we don’t know
what’s best here”. This isn’t a counsel of despair: in medicine, and in teaching, we
know that most of what we do does some good (if we’re not better than nothing,
then we’re all in big trouble!). The real challenge is in identifying what works the
best, because when people are deprived of the best, they are harmed too. But this
is also a reminder of how inappropriate certainty can be a barrier to progress,
especially when there are charismatic people, who claim they know what’s best,
even without good evidence.
Medicine suffered hugely with this problem, and as late as the 1970s there
were infamous confrontations between people who thought it was important to
run fair tests, and “experts”, who were angry at the thought of their expertise
being challenged, and their favourite practices being tested. Archie Cochrane was
one of the pioneers of evidence based medicine, and in his autobiography, he
describes many battles he had with senior doctors, in glorious detail. In 1971,
Cochrane was concerned that Coronary Care Units in hospitals might be no
better than home care, which was the standard care for a heart attack at the time
(we should remember that this was the early days of managing heart attacks, and
the results from this study wouldn’t be applicable today). In fact, he was worried
that hospital care might involve a lot of risky procedures that could even,
conceivably, make outcomes worse for patients overall.
Because of this, Cochrane tried to set up a randomised trial comparing home
13
care against hospital care, against great resistance from the cardiologists. In fact,
the doctors running the new specialist units were so vicious about the very
notion of running a trial that when one was finally set up, and the first results
were collected, Cochrane decided to play a practical joke. These initial results
showed that patients in Coronary Care Units did worse than patients sent home;
but Cochrane switched the numbers around, to make it look like patients on
CCUs did better. He showed the cardiologists these results, which reinforced
their belief that it was wrong of Cochrane to even dare to try running a
randomised trial of whether their specialist units were helpful. The room
erupted:
“They were vociferous in their abuse: “Archie,” they said “we always
thought you were unethical. You must stop this trial at once.” … I let them
have their say for some time, then apologized and gave them the true
results, challenging them to say as vehemently, that coronary care units
should be stopped immediately. There was dead silence and I felt rather
sick because they were, after all, my medical colleagues.
Similar confrontations are reported in many new fields, when people try
subjecting ideas and practices to fair tests, in randomised trials. But being open
and clear about the need for research - when there is no good evidence to help us
choose between interventions - is also important because it helps make sure that
research is done on relevant questions, meeting the needs of teachers, pupils and
parents. When everyone involved in teaching knows a little about how research
is done - and what previous research has found - then we can all have a better
idea of what questions need to be asked next.
But before we get on to how this can happen, we should first finish the myths
about trials. From now on, these are all cases where people overstate the benefits
of trials.
For example, sometimes people think that trials can answer everything, or
that they are the only form of evidence. This isn’t true, and different methods are
useful for answering different questions. Randomised trials are very good at
showing that something works; they’re not always so helpful for understanding
why it worked (although there are often clues when we can see that an
intervention worked well in children with certain characteristics, but not so well
in others). “Qualitative” research - such as asking people open questions about
their experiences - can help give a better understanding of how and why things
worked, or failed, on the ground. This kind of research can also be useful for
generating new questions about what works best, to be answered with trials. But
14
qualitative research is very bad for finding out whether an intervention has
worked. Sometimes researchers who lack the skills needed to conduct or even
understand trials can feel threatened, and campaign hard against them, much
like the experts in Archie Cochrane’s story. I think this is a mistake. The trick is to
ensure that the right method is used to answer the right questions.
A related issue involves choosing the right outcome to measure. Sometimes
people say that trials are impossible, because we can’t capture the intangible
benefits that come from education, like making someone a well rounded member
of society. It’s true that this outcome can be hard to measure, although that is an
argument against any kind of measurement of attainment, and against any kind
of quantitative research, not just trials. It’s also, I think, a little far-fetched: there
are lots of things we try to improve that are easy to measure, like attendance
rates, teenage pregnancy, amount of exercise, performance on specific academic
or performance tests, and so on.
However, we should return to the overly exaggerated claims sometimes
made in favour of trials, and the need to be a critical consumer of evidence. A
further common mistake is to assume that, once an intervention has been shown
to be effective in a single trial, then it definitely works, and we should use it
everywhere. Again, this isn’t necessarily true. Firstly, all trials need to be run
properly: if there are flaws in a trial’s design, then it stops being a fair test of the
treatments. But more importantly, we need to think carefully about whether the
people in a trial of an intervention are the same as the people we are thinking of
using the intervention on.
The Family Nurse Partnership is a programme that is well funded and
popular around the world. It was first shown to be effective in a randomised trial
in 1977. The trial participants were white mothers in a semirural setting upstate
from New York, and people worried at the time that the positive results might
have been exceptional, and occurred simply because the specific programme of
social support that was offered had suited this population unusually well. In
1988, to check that the findings really were applicable to other settings, the same
programme was assessed using a randomised trial in African-American mothers
in inner city Memphis, and again found to be effective. In 1994, a third trial was
conducted in a large population of Hispanic, African-American, and Caucasian
mothers from Denver. After this trial also showed a benefit, people in the US
were fairly certain that the programme worked, with fewer childhood injuries,
increased maternal employment, improved “school readiness”, and more.
Now, the Family Nurse Partnership programme is being brought to the UK,
15
but the people who originally designed the intervention have insisted that a
randomised trial should be run here, to see if it really is effective in the very
different setting of the UK. They have specifically stated that they expect to see
less dramatic benefits here, because the basic level of support for young families
in the UK is much better than that in the US: this means that the difference
between people getting the FNP programme, and people getting the normal level
of help from society, will be much smaller.
This is just one example of why we need to be thoughtful about whether the
results of a trial in one population really are applicable to our own patients or
pupils. It’s also an illustration of why we need to make trials part of the everyday
routine, so that we can replicate trials, in different settings, instead of blindly
assuming we can use results from other countries (or even other schools, if they
have radically different populations). It doesn’t mean, however, that we can
never trust the results of a trial. This is just another example of why it’s useful to
know more about how trials work, and to be a thoughtful consumer of evidence.
Lastly, people sometimes worry that trials are expensive and complicated.
This isn’t necessarily true, and it’s important to be clear what the costs of a trial
are being compared against. For example, if the choice is between running a trial,
and simply charging ahead, implementing an idea that hasn’t been shown to
work - one that might be ineffective, wasteful, or even harmful - then it’s clearly
worth investing some time and effort in assessing its true impact. If the
alternative is doing an “observational” study, which has all the shortcomings
described above, then the analysis can be so expensive and complex - not to
mention unreliable - that it would have been easier to randomise participants to
one intervention or the other in the first place.
But the mechanics and administrative processes for running a trial can also
be kept to a minimum with thoughtful design, for example by measuring
outcomes using routine classroom data, that was being collected anyway, rather
than running a special set of tests. More than anything, though, for trials to be
run efficiently, they need to be part of the culture of teaching.
Making evidence part of everyday life
I’m struck by how much enthusiasm there is for trials and evidence based
practice in some parts of teaching: but I’m also struck that much of this
enthusiasm dies out before it gets to do good, because the basic structures
needed to support evidence based practice are lacking. As a result, a small
16
number of trials are done, but these exist as isolated islands, without enough
bridges joining the people and strands of work together. This is nobody’s fault:
creating an “information architecture” out of thin air is a big job, and it might
take decades. The benefits, though, are potentially huge. Some individual
randomised trials from the UK have produced informative results, for example,
but these results are then poorly communicated, so they don’t inform and change
practice as well as they might.
Because of this, I’ve sketched out the basics of what education would need, as
a sector, to embrace evidence based practice in a serious way. The aim - which I
hope everyone would share - is to get more research done, involving as many
teachers as possible; and to get the results of good quality research disseminated
and put into practice. It’s worth being clear, though, that this is a first sketch, and
a call to arms. I hope that others will pull it apart and add to it. But I also hope
that people will be able to act on it, because structures like these in medicine
help capture the best value from the good work - and hard work - that is done all
around the country.
Firstly - and most simply - it’s clear that we need better systems for
disseminating the findings of research to teachers on the ground. While
individual studies are written up in very technical documents, in obscure
academic journals, these are rarely read by teachers. And rightly so: most
doctors rarely bother to read technical academic journals either. The British
Medical Journal has brief summaries of important new research from around the
world; and there is a thriving market of people offering accessible summary
information on new “what works” research to doctors, nurses, and other
healthcare professionals. The US government has spent vast sums of money on
two similar websites for teachers: “Doing What Works”, and the “What Works
Clearing House”. These are large, with good quality resources, and they are
written to be relevant to teachers’ needs, rather than dry academic games. While
there are some similar resources in the UK, these are often short-lived, and on a
smaller scale.
For these kinds of resources to be useful at all, they then need to land with
teachers who know the basics of “how we know” what works. While much
teacher training has reflected the results of research, this evidence has often been
presented as a completed canon of answers. It’s much rarer to find all young
teachers being taught the basics of how different types of research are done, and
the strengths and weaknesses of each approach on different types of question
(although some individual teachers have taught themselves on this topic, to a
very high level). Learning the basics of how research works is important, not
17
because every teacher should be a researcher, but because it allows teachers to
be critical consumers of the new research findings that will come out during the
many decades of their career. It also means that some of the barriers to research,
that arise from myths and misunderstandings, can be overcome. In an ideal
world, teachers would be taught this in basic teacher training, and it would be
reinforced in Continuing Professional Development, alongside summaries of
research.
In some parts of the world, it is impossible to rise up the career ladder of
teaching without understanding how research can improve practice, and
publishing articles in teaching journals. Teachers in Shanghai and Singapore
participate in regular “Journal Clubs”, where they discuss a new piece of
research, and its strengths and weaknesses, before considering whether they
would apply its findings in their own practice. If the answer is no, they share the
shortcomings in the study design that they’ve identified, and then describe any
better research that they think should be done, on the same question.
This is an important quirk: understanding how research is done also enables
teachers to generate new research questions. This, in turn, ensures that the
research which gets done addresses the needs of everyday teachers. In medicine,
any doctor can feed up a research suggestion to NIHR (the National Institute for
Health Research), and there are organisations that maintain lists of what we
don’t yet know, fed by clinicians who’ve had to make decisions, without good
quality evidence to guide them. But there are also less tangible ways that this
feedback can take place.
Familiarity with the basics of how research works also helps teachers get
involved in research, and to see through the dangerous myths about trials being
actively undesirable, or even “impossible” in education. Here, there is a striking
difference with medicine. Many teachers pour their heart and soul into research
projects which are supposed to find out whether something worked; but in
reality the projects often turn out to be too small, being run by one person in
isolation, in only one classroom, and lack the expert support necessary to ensure
a robust design. Very few doctors would try and run a quantitative research
project alone in their own single practice, without expert support from a
statistician, and without help from someone experienced in research design.
In fact, most doctors participate in research by playing a small role in a larger
research project which is coordinated, for example, through a research network.
Many GPs are happy to help out on a research: they recruit participants from
among their patients; they deliver whichever of two commonly used treatments
18
has been randomly assigned to their patient; and they share medical information
for follow-up data. But they get involved by putting their name down with the
Primary Care Research Network covering their area. Researchers interested in
running a randomised trial in GP patients then go to the Research Network, and
find GPs to work with.
This system represents a kind of “dating service” for practitioners and
researchers. Creating similar networks in education would help join up the
enthusiasm that many teachers have - for research that improves practice - with
researchers, who can sometimes struggle to find schools willing to participate in
good quality research. This kind of two-way exchange between researchers and
teachers also helps the teacher-researchers of the future to learn more about the
nuts and bolts of running a trial; and it helps to keep researchers out of their
ivory towers, focusing more on what matters most to teachers.
In the background, for academics, there is much more to be said on details.
We need, I think, academic funders who listen to teachers, and focus on
commissioning research that helps us learn what works best, to improve
outcomes. We need academics with quantitative research skills from outside
traditional academic education departments - economists, demographers, and
more - to come in and share their skills more often, in a multidisciplinary fashion.
We need more expert collaboration with Clinical Trials Units, to ensure that
common pitfalls in randomised trial design are avoided; we may also need -
eventually − Education Trials Units, helping to support good quality research
throughout the country.
But just as this issue stretches way beyond a few individual research projects,
it also goes way beyond anything that one single player can achieve. We are
describing the creation of a whole ecosystem from nothing. Whether or not it
happens depends on individual teachers, researchers, heads, politicians, pupils,
parents and more. It will take mischievous leaders, unafraid to question
orthodoxies by producing good quality evidence; and it will need to land with a
community that - at the very least - doesn't misunderstand evidence based
practice, or reject randomised trials out of hand.
If this all sounds like a lot of work, then it should do: it will take a long time.
But the gains are huge, and not just in terms of better evidence, and better
outcomes for pupils. Right now, there is a wave of enthusiasm for good quality
evidence, passing through all corners of government at the moment. This is the
time to act. Teachers have the opportunity, I believe, to become an evidence
based profession, in just one generation: embedding research into everyday
19
practice; making informed decisions independently; and fighting off the odd
spectacle of governments telling teachers how to teach, because teachers can use
the good quality evidence that they have helped to create, to make their own
informed judgements.
There is also a roadmap. While evidence based medicine seems like an
obvious idea today - and we would be horrified to hear of doctors using
treatments without gathering and using evidence on which works best - in
reality these battles were only won in very recent decades. Many eminent
doctors fought viciously, as recently as the 1970s, against the very idea of
evidence based medicine, seeing it as a challenge to their expertise. The case for
change was made by optimistic young practitioners like Archie Cochrane, who
saw that good evidence on what works best was worth fighting for.
Now we recognise that being a good doctor, or teacher, or manager, isn’t
about robotically following the numerical output of randomised trials; nor is it
about ignoring the evidence, and following your hunches and personal
experiences instead. We do best, by using the right combination of skills to get
the best job done.
