Professional Documents
Culture Documents
Source: www.gov.uk/government/uploads/system/uploads/attachment_data/file/32277/11-1367-2011-skills-for-life-
survey-findings.pdf
Risky business
Exploring how different people 8
understand risk.
Stats Q&A
Putting to rest some common
statistical myths.
10
When stats go bad
Real-life examples of how 12
statistics can be misused.
real voices
Three people talk about how
they use science and stats.
14
ONLINE
Go to www.wellcome.ac.uk/bigpicture/
numbers for more teaching resources,
including extra articles, useful web links, lesson
ideas, curriculum links and more. You can
driver age
also download the PDF of this magazine and
subscribe to the Big Picture series.
Source: www.gov.uk/government/statistical-data-sets/ras30-reported-casualties-in-road-accidents
32 250 GB
games
console
GB
portable
mp3 player data amassed by CERN by mid-2012
200 petabytes
18 000 000
usable storage at wellcome trust
sanger institute by 2011
GB
3.3 4.3
(estimated)
billion
billion 10%
Source: www.reform.co.uk/client_files/www.
Source: www.radicati.com/?p=8417 reform.co.uk/files/the_value_of_mathematics.pdf
0.25
coin tossing probabilities
finding data
60%
too – different sources might define
‘numeracy’ or ‘adult’ differently.
Summer 2013 | 3
how science works
Scientists work to investigate, interpret and understand the
world around us. They use a set of tools and techniques known
as the scientific method and produce data.
Beating bias
iStockphoto
Researchers try to keep things objective
Evidence is what science is all test it, then analyse the data and How you design a study depends on the question you’re
about. It’s all well and good draw conclusions. The overall asking. In medicine, the most appropriate type of study
having an idea about how the quality of any study depends on depends on whether you are trying to diagnose, treat or
world works, but you need making sure each step is done calculate the likely outcome of a condition. For more on
something to back it up. The properly. this, see the diagram below.
scientific method allows Not all evidence gathered In ecology, as in medicine, samples are taken: examining
researchers to test their ideas using the scientific method is an entire population can be time-consuming and damage
through investigations. the same. Quantitative data the environment you’re looking at. The design of the
First, you need a question. are measurable (e.g. length or study depends on what you’re investigating. For example,
This normally comes after height), and because they are to estimate the size of an animal population, researchers
some initial observations that numerical they can be analysed often use a mark–release–recapture method. Marking
suggest something interesting statistically. Qualitative and releasing a set number of individuals, then capturing
is happening or from a problem data are descriptive (e.g. hair another set number and counting how many individuals
you want to solve. You can colour). Analysing qualitative got caught twice gives a good indication of how many
construct a hypothesis that data is more difficult and animals there are altogether. To sample plant populations,
could explain what you’ve seen interpretations can be subject to quadrats are used so that each sample comes from a specific
and carry out an experiment to personal opinion. area of ground.
Summer 2013 | 5
making sense of stats
Experiments yield data. How can we interpret this information using
statistics? What are some of the common pitfalls in data analysis
and interpretation?
Graphically thinking
Just about average Using graphs and diagrams to show data
There are different types of average When you have your data, you may want to
represent them graphically – for example, to
show whether two variables are correlated
When we talk about ‘an average’, doubled the mean, and the majority (e.g. a scatter graph plotting duck egg length
what we’re really trying to do is get of the numbers are below the mean. against duck egg width) or to show different
some sense of where the middle Imagine how the mean wealth of proportions (e.g. a pie chart showing the
is. We can then use that as a way biology teachers in a room might prey items of a lion). Which chart or graph
of comparing two groups of data. change if Bill Gates joined them. you use will depend on the type of data you
Unfortunately, there isn’t just one If you place the numbers in have. Common diagrams include bar charts,
type of average – there are several. ascending order and look for the pie charts, line graphs, scatter graphs and
To get the mean, add the data middle value in the list, you have histograms.
together and divide the total by the median. If there are an even See www.wellcome.ac.uk/bigpicture/
how many data points there are (see number of values, you take the numbers for a how-to guide on histograms
equation below). Beware, though mean of the middle pair. For the and weblinks to other resources on graphs
– outlying data can often skew the original list, this is 9. Outliers and charts.
mean to be artificially high or low. have a much smaller effect on the
Take the following number list: median than the mean, so adding
1, 3, 6, 9, 9, 11, 14. The mean here is 50 again does not alter the median.
7.57. However, add a much higher The mode is the value that A number of significance
number to the end of the list, say occurs most often in a list. For this
Significance has a special meaning in stats
50. The mean is now 12.88. Just one list, the mode is 9.
particularly large outlier has almost If you want to accept your alternative
hypothesis, you must first reject your null
hypothesis. There is a chance, however, of
rejecting the null hypothesis when it is actually
true. It is usually possible to calculate the
probability (p-value) that what you observed in
an experiment was due just to chance.
You use a significance level to decide
whether you will reject the null hypothesis,
and this is often set at the 0.05 or 5 per cent
What’s your type? level. If your measured p-value equals 0.04, for
Not all data are the same example, then this is less than 5 per cent, so you
can reject the null hypothesis and accept your
Researchers define data in different ways. For example, data are categorical if the alternative. Still, this doesn’t mean that you
values can be sorted into non-overlapping categories (e.g. by blood type, species have proved the alternative hypothesis: if the
or sex). Every value should belong to only one category, and it should be clear null hypothesis were true, there would still be
which one it belongs to. Categorical data are also known as ‘nominal data’, or a 4 per cent chance of getting your result.
‘frequencies’, as the research looks to find out how frequently data fall into each If an unscrupulous investigator keeps on
category. Ordinal data, by contrast, can be ranked or have some sort of rating doing experiments on useless treatments, they
scale. Ordinal data often come from surveys and questionnaires. will still get results ‘significant at the 5 per cent
Data can also be defined as discrete level’ on 1 in 20 occasions. If only those ‘positive’
iStockphoto
or continuous. Data are discrete, or trials are reported, we will get a very misleading
discontinuous, if they can take only impression. This is why it is essential to have
isolated values. Continuous data can access to all the evidence, whether positive
take on any value and are limited only by or negative. ‘All trials registered, all results
how accurate your measurements are. reported’ is a campaign started by researchers,
So, while foot length is continuous, shoe doctors and others to try and get all clinical
size is discrete because you can’t be a size trials past and present to be registered and to
7.234434 – you have to be a 7 or a 7.5. have their results reported.
What is normal?
iStockphoto
Many things follow a normal distribution
Datasets can be spread shape of the distribution normal distribution,
out in many different resembles a bell and so is approximately 68 per
ways. The majority of sometimes called a ‘bell cent of the data will fall
the data can sit above curve’. This is also called between one standard
the mean or below it. In a normal distribution deviation on either side
many datasets, however (see graph below). of the mean. Around 95
– particularly large Standard deviation per cent will fall within
Jumping to
ones – the data points is a measure of how two standard deviations conclusions
seem to settle equally on spread out the numbers on either side. In such Take care with correlation
either side of the mean. are around the mean. circumstances, the mean,
Plotted on a graph, the If a dataset follows a median and mode of the
data are all equal. So you’ve collected your data and
There are many noticed a strong correlation between
MEAN
everyday biological two of your variables. It might be very
examples that follow tempting to assume that a change
a roughly normal in one is causing a change in the
SD = STANDARD
distribution, including other, but don’t fall into the trap.
DEVIATION
blood pressure, height A correlation shows that there is a
and foot length. Along relationship between your variables;
with these examples, you it doesn’t prove there is a causal
could also try looking at relationship.
stalk height in daisies, Think about Andrew Wakefield’s
the length of holly 1998 claim that using the MMR
-3SD -2SD -1SD 0 +1SD +2SD +3SD
leaves or the diameter of vaccine can result in autism. It’s true
68%
95% lichens (commonly found that use of the MMR vaccine had
99.7%
on gravestones). increased up to that point, as had the
number of cases of autism recorded,
so there’s a correlation between the
Choose your method two. This doesn‘t necessarily mean
Exploring different statistical tests that the jab is causing the increase in
autism – there could be a third factor
There are several different statistical and your predicted frequencies and (called a confounder) causing one
tests that you can use, depending on calculate the chi-squared value. You or both of the variables to increase,
the type of data you are dealing with. compare this to the critical value to such as an increase in maternal age.
Two examples are given here, and see whether the difference between Wakefield’s work has been discredited
there are more online – plus a worked- them is likely to have occurred by and he was struck off in 2010,
through example for chi-squared (‘chi’ chance. If your calculated value is meaning he can no longer practise
is pronounced ‘ki’ to rhyme with ‘eye’) bigger than the critical value, you as a doctor.
– at www.wellcome.ac.uk/bigpicture/ reject your null hypothesis. True causation can only be tested
numbers. The t-test enables you to see with carefully controlled studies,
The chi-squared test is used with whether two samples are different which often compare two groups who
categorical data to see whether any when you have data that are are matched in every way except for
difference in frequencies between your continuous and normally distributed. the variable of interest. This limits
sets of results is due to chance. For The test allows you to compare the role of confounders as much as
example, you could use the test with the means and standard deviations of possible.
the null hypothesis that ‘there is no the two groups to see whether there Oh, and watch your ‘u’s and ‘s’s – a
difference in the frequency of worms is a statistically significant difference causal relationship is very different
on different types of ground’. between them. For example, you could from a casual one!
In a chi-squared test, you draw a test the heights of the members of two
table of your observed frequencies different biology classes.
Summer 2013 | 7
Risky business
It is impossible to live a life without risk. So, how do we understand
and weigh up the risks associated with different activities, behaviours
and events?
Take a chance
Nobody lives a risk-free life
Risk is generally understood responsibility for reducing addition, risk in the workplace spend a lot of time and money
as an exposure to the chance risk, from individuals to has become a hot topic in on health campaigns to warn
of injury, loss or harm. governments. On the roads, recent years. about risks in many areas, from
Throughout our lives we come for example, laws about Along with conventional unprotected sex to smoking and
across countless situations speed limits, seat belts and risk reduction devices such as from drinking while pregnant
that present such risks; they’re drink-driving are intended fire extinguishers and alarms, to the dangers of drugs.
impossible to avoid. We often to reduce the risk of accidents attention is increasingly
act to limit risks by undertaking and injuries, and we all have turning to reducing other,
iStockphoto
a particular action or by responsibility for our less obvious risks at work.
reducing or stopping a certain own behaviour as a driver This can include supplying
behaviour. Sometimes we do it or passenger. appropriate office furniture
without giving it much thought In the home, people install to reduce the risk of back pain
– by taking an umbrella if the smoke and intruder alarms to and taking steps to reduce the
forecast says it’s likely to rain, reduce the risk of fire damage risk of stress. On a larger scale,
for example. or burglary. The same is true of national and international
Everyone has some carbon monoxide detectors. In government organisations
Life-changing findings?
We don’t all respond to risk the same way
Living the (micro) life People perceive and respond to risks deal with short-term choices, where the
One way to quantify risk in different ways. How we think and potential benefits are immediate, better
behave (psychological factors) and how than long-term ones.
society works (sociological factors) In some situations we have to weigh
US scientist Ronald A Howard first play a part in this – we often go with up relative risks – for example, with
introduced the concept of ‘micromorts’ our gut feeling or are affected by the preventative medicines such as anti-
– a unit of risk measuring a one-in- behaviour of people around us. It seems malaria drugs. Drugs like doxycycline
a-million probability of death. This most people underestimate the risk in and mefloquine can have unpleasant
allows us to compare the risk of day-to- activities where they have control and side-effects, but if you don’t take them
day events. For example, in the medical overestimate the risk of things they can when travelling to malaria hotspots,
world, going under anaesthesia for a do little about. you run the risk of contracting the
non-emergency operation exposes you The timescale involved can also be potentially fatal condition.
to an average of ten micromorts. In the a factor, particularly when it comes to Research into our perception of
UK, giving birth (all births combined) health. It can be hard to be motivated risk has tried to place a monetary
is worth 120 micromorts; a Caesarean to change short-term behaviour – such value on how much the average person
section increases this to 170. Sky- as eating or drinking habits – because would have to be paid to willingly
diving, rock-climbing and hang-gliding of a risk of health problems in the accept a one-in-a-million chance of
come in at ten micromorts or lower. distant future. This can often lead death – see ‘Living the (micro) life’. You
Calculations suggest each mission people to put off an action to another might think it would be high, but the
flown by a member of Bomber day, saying things like “I’ll stop findings estimate that it is just $50.
Command in World War II carried smoking when I get older”. When it
iStockphoto
We’re not always good at doing maths are both 1/2, no matter what’s happened
quickly. Often we go by gut feeling, rather previously. The situation is different if the
than what the numbers tell us. Take events are dependent – if you pull an ace
childbirth, for example; imagine a woman from a pack of cards without returning it,
has given birth to three children, who are the probability of picking another ace goes
all boys. If she becomes pregnant again then down from 4/52 to 3/51.
people might say – because all of her current This is especially true when unlikely
children are boys – that there is a strong events occur. Take natural disasters, for
chance her next child will also be a boy, or example. It is often said that events like
Are you absolutely
that she must be ‘due’ a girl next. And yet floods, tsunamis and hurricanes are ‘once
sure? biology tells us there is still a 50 per cent (or every 100 years events’, and people are
Why risk should be reported 1:1) chance of her having a boy because each surprised when they occur more often.
responsibly conception is an independent What it actually means is there is
event and is unaffected by a 1 per cent (1 in 100) chance of
The way numbers are presented the existence of her previous the event happening in any
can be misleading, so watch out. children. given year. Again, however,
In 1995, UK news outlets reported The same is true of coin these are independent events:
advice from the Committee tosses. Just because a coin if a ‘once in 100 years’ flood
on Safety of Medicines, which comes up heads ten times happened last
suggested that a new version of in a row, a head is no year, that
the contraceptive pill doubled more likely than a tail doesn‘t mean
a woman’s risk of venous on the 11th flip that it can’t
thromboembolism (VTE) – a (provided the coin happen again
condition in which blood clots is not fixed). The this year. It is
form in the veins around the probability of unlikely – but
legs. In the wake of the news, the having a boy, or unlikely things
British Pregnancy Advisory Service the coin coming happen all
estimated that the number of down heads, the time.
abortions in the UK rose by 13 000,
reversing a previously downward
trend, as a result of a falling trust
in the Pill. The number of births to
teenage mothers also increased.
The actual findings relating to
the Pill and VTE are as follows: for
every 7000 women that took the
previous Pill, one would have VTE.
For every 7000 women that took
the new Pill, two would have VTE.
The claim of ‘a doubling’ of
the risk of VTE was not wrong:
the number of women affected
had doubled from 1 in 7000 to 2
in 7000. This is the relative risk.
But was this the best number to
publish? The difference in terms
of women affected is just 1 in 7000
(the difference between 2 in 7000
and 1 in 7000 women), or 0.014
per cent.
Summer 2013 | 9
stats q&a
From lottery mythbusting to understanding medical test results, your
niggling number questions are answered using science and statistics.
? L
Q: My dad refuses to Q: My teacher says there’s Q: I just read online about Q: I’ve passed my driving
pick 1, 2, 3, 4, 5 and 6 no such thing as ‘truth’ a man who cured his test. I’ve been told by car
on the Lotto draw. in science. So why do we cancer with carrot juice. insurance companies that
Is this sensible? bother? Why can’t we all use there’s no discount for
natural remedies? being female, even though
A: Those six numbers are just A: Think for a second about women are statistically less
as likely to come up as any how you’d prove something A: Remember a key saying likely to have an accident.
other six numbers. In the UK to be true. You could say, in statistics: correlation does Why?
Lotto, six numbers are drawn for example, that all swans not equal causation. In other
at random from numbers 1 to are white – but you can’t words, just because there A: You’re right that women are
49, giving 13 983 816 possible prove that’s true because it’s is a relationship between statistically less likely to have
combinations. The chance of impossible for you to observe the volume of carrot juice an accident. In fact, young men
winning is approximately 1 in every single swan. However, the man consumed and the under the age of 22 used to pay
14 million, no matter which just one observation of a black disappearance of his cancer, it an average of £1000 more a year
numbers you pick. However, swan would prove that you were doesn’t mean that one caused in insurance premiums than
there is a very good reason why wrong. It’s much easier to prove the other. women of the same age. This
your dad shouldn’t play 1, 2, 3, something is untrue than to Even if it did, think what was because the statistics show
4, 5 and 6 – if he did win, he prove it’s true. you’re not hearing about. If that young men are twice as
would win less money than with Accordingly, the scientific he had died in spite of all the likely to suffer a serious collision
another combination. method works by trying to carrots, would an article have as young women (and ten
An estimated 10 000 players falsify a statement. When been written about it? Maybe times more likely to be killed
a week choose the combination a scientist puts forward a there are thousands of failures or seriously injured in one than
of balls 1–6. If you picked those theory, the rest of the scientific out there that are not being those aged 35 or over).
numbers too and they came up, community try to disprove it. reported. This is an example However, an EU ‘gender
you’d have to share your cash It’s only those theories that of ‘selection bias’ – where you directive’ that came into force
with all those other people, survive the onslaught – the ones only hear about the ‘successes’. on 21 December 2012 made it
which would severely dilute that can’t (yet) be proven wrong A related phenomenon, illegal to discriminate according
your winnings. – that we stick with. That ‘publication bias’, happens in to sex when pricing financial
Although you can’t increase doesn’t mean that one day a new scientific research. Journals and products, such as pension
your chances of winning, piece of evidence won’t come some researchers are more likely annuities and insurance. The
there are other ways you might along to disprove it, but you to publish new, exciting results change has seen premiums for
increase the potential amount trust it for now. For example, than ‘negative’ findings that women under 40 rise and those
you’d win. People pick numbers we think that dropping a brick a particular variable does not for men of a similar age fall.
between 1 and 31 more often, will mean it falls to the ground, cause an effect. Some have argued that this
for example, because they but we can never prove that it Medical science does not is unfair, but others have said
correspond to the birthdays of wouldn’t one day float off. So work on the basis of anecdotal equality can’t be selective. The
people they know. Selecting six far, though, the theory that evidence (based on personal change is unlikely to affect the
numbers between 32 and 49 is it will fall has withstood all experiences), but through premiums of those aged over 40
more likely to set you apart from attempts at falsification. carefully controlled trials. – the age at which the statistics
the crowd. show that men and women
become equally likely to have
an accident.
Q: My mum doesn’t
like that I smoke, but I Q: My doctor has told me that if I get a ‘positive’ result in a medical
don’t think it’s a big deal. test, it doesn’t definitely mean that I have the disease. Why not?
My great-grandma has
smoked since she was 15 A: No test is 100 per cent accurate. women screened, 41 have a positive
and is still going at 90! Not all positive test results mean that result and are called back for more tests.
someone has a particular condition; not Of these, 8 women will be found to have
A: A common misconception all negative results mean that a disease cancer, and 33 will be found not to. So,
is that your own experience is absent. Positive and negative test an initial positive result means just a
– or those of your friends or results can be described as true or false, 20 per cent chance of cancer. Overall,
relatives – can be generalised depending on whether they classify 966.8 (8 + 958.8) women per 1000
to everyone else. As a famous correctly the person tested. screened will get a ‘true’ result, making
quote by English statistician The diagram shows the outcome of this technique 97 per cent accurate.
and biologist R A Fisher says, breast screening by an X-ray technique
“That is an experience, not an called mammography. For every 1000
experiment”. Just because your
great-grandma has survived
true
to that age doesn’t mean you
will; you can’t get meaningful 8 positive
(cancer)
data from a sample size of one.
41
This is why most scientists use positive further
test
experiments that collect large result
testing
volumes of data, rather than
relying on anecdotal false
case studies. 33 positive
(no cancer)
1000
women
screened
true
958.8 negative
(no cancer)
959
negative
test
result
false
images: iStockphoto
Carrot juice image: 123RF
Summer 2013 | 11
when stats go bad
Actually, this is more a case of ‘when people using stats go bad’. Look at
these real-life examples to see how numbers can be misused, misrepresented
or poorly explained, and explore the implications of substandard stats.
“Chance of cot
do if the two events were independent of
each other, like tosses of a coin. The chance
of getting two heads in a row is 1/2 x 1/2 (1/4).
deaths in brothers
However, two cot deaths in the same family
are not independent events; there could be
underlying genetic or environmental factors
‘1 in 73 million’”
that make them more likely. The Royal
Statistical Society deemed Meadow’s account
a “mis-use of statistics”.
Consider…
Q: Even if the chance of a double cot death
in the same family really was 1 in 73 million,
why would this not have meant there was
only a 1 in 73 million (0.0000014 per cent)
Sally Clark served three years of a life The prosecution’s assertion was based chance of the accused being innocent?
sentence for the murder of her two children on the expert testimony of Professor Sir Search for “prosecutor’s fallacy” online to find
before her conviction was overturned in Roy Meadow, a researcher in paediatrics. out more.
2003. In the original case, the defence had Meadow had said that the chances of one Q: What figure should this number have
claimed that sudden infant death syndrome child dying from SIDS in a non-smoking, been compared with to work out the relative
(SIDS) – commonly known as cot death – was affluent family was 1 in 8543. When working likelihood of guilt or innocence?
responsible for the death of both boys, who out the probability of two cot deaths in the Q: Should statistical evidence in court only
died just over a year apart. The prosecution same family, he squared this probability be presented by experts in statistics, rather
argued that such a double cot death was – multiplying 8543 by 8543 – to get 1 in than by experts in the field in which the
exceptionally unlikely and claimed murder. 73 million. statistics are being used?
MOT for
the advert implied that CT it might be unacceptable for
scanning in patients with no people without symptoms.
current symptoms could pick up
“Recommended by
In explaining their verdict, they Why risk should be
said: “We were concerned that reported responsibly
the...respondents entering the
any friends”. (which included a question about Q: Is it OK to provide incentives sell more papers? Would
Afterwards, whether they would recommend to people taking part in surveys? you feel differently if this
some text the product to a friend). If they Will you get the same results, were done by a government
flashed up on returned the survey, along with or will the incentive skew the health campaign or a cancer
the screen: a photo of themselves and a findings? charity trying to raise funds
“Participants short story, they had a chance Q: Is there any guarantee that for research? What if it were
in a survey of winning a trip to New York. the people filling in the survey used in adverts by a business
of 245 Red The ASA banned the advert actually tried the hair dye they that produces vegetarian
magazine because they believed the claim were sent? How could it have food?
readers, of 93 per cent was misleading. been done differently?
Summer 2013 | 13
real voices
Three people tell us about the role of statistics and science in their lives. Meet Vicky Peterkin, a
biostatistician at a pharmaceutical company; biology teacher David Colthurst; and Anthony
Underwood, who uses bioinformatics at Public Health England.
What do you do? which we can be certain that What do you do? you may do steps A, B, C and
I work as a statistician in the the effects we see are due to our I am a secondary school D in the hope you will get
pharmaceutical industry, drug, rather than chance. It’s science teacher. I am also result E, nine times out of ten
specialising in clinical trials the key piece of information the lead teacher on the MBP2 it won’t work! It has also given
for new treatments. we want from the trial. It tells project, which gives sixth-form the students the opportunity
us whether the drug is useful students the opportunity to to try lots of techniques that
How do you use statistics and whether we should carry carry out genuine research. they normally wouldn’t get to
in your job? on researching it. That’s what experience until university.
To find out whether the drug I love about statistics: you What is the MBP2 project? For example, our students
we’re testing significantly can boil a huge amount of Five years ago my wife was have developed a genetically
improves a measurement information down to a single, diagnosed with multiple modified yeast strain that can
of interest compared with a clear, yes-or-no decision. sclerosis. I wondered if I would make the human myelin basic
control drug. For example, be able to combine my 15 years protein. While doing this they
if we’re testing a new drug Why might people be wary of experience as a biochemist have learnt how to extract
to treat high blood pressure, of maths? with my 15 years of teaching. DNA, use the polymerase chain
the measurement of interest I think people imagine The Myelin Basic Protein reaction (PCR) to amplify DNA,
might be the change in statisticians sit at a computer Project (MBP2) is investigating and carry out Western blots to
blood pressure since starting working on their own all day. the role of this protein in study the proteins produced.
treatment; however, it’s not In my job that’s just not the multiple sclerosis using
enough to just look at the mean case. I’m involved at every stage genetically modified yeast. What is next for the project?
blood pressure change in each of a clinical trial – I help design The project is run in I believe MBP2 provides a model
treatment group. We need to it, check the data quality while collaboration with researchers for how students can carry out
adjust for other information it’s running and analyse it all at the University of Kent. research in schools. Inspired
(such as age and weight) and at the end – and I work closely When we first started, by this, Authentic Biology is a
show whether the difference with medics, scientists and they ran a workshop at the series of research projects led
is statistically significant. trial monitors on a daily basis. University to teach DNA and by sixth-form students in five
protein techniques to a small schools across the UK. Each
Why is that? What training do you have? group of students, who then one is drawing on the expertise
Adjusting for other I studied maths at university, taught them to their teachers of researchers at their local
information ensures that then a Master’s in statistics. and other students. It was a universities to investigate
any differences between the There’s a lot of on-the-job nice turnaround to have the topics relevant to them. In
treatment group results are training, too. students be the experts while London, for example, students
due to the drug, not due to I think students need to the teachers sat scratching are researching diabetes, and
differences in disease severity be exposed to statistics at their heads, thinking “How in Sheffield they are looking at
and demographics. Statistical an earlier age, so they can does this bit work?” heart disease. We are currently
significance indicates that the see for themselves how useful planning the second Authentic
size of the difference between it is. The industry is in What do your students gain? Biology Symposium. This is
treatment groups is too big to desperate need of young The real wake-up call for them an opportunity for the schools
have occurred by chance and statisticians. You can become was around experimental involved to share their research
must be due to the treatments. involved in important trials procedure – often, experiments with each other and with
very quickly, and your opinion don’t work as planned or give academics from the partner
What is a p-value? is highly valued. the results that are expected. universities.
When we run a statistical When this happens, you
test, we get a p-value at the Find out more about working have to tweak and change Find out more at www.
end. It shows the statistical as a statistician at the design. Very quickly the thelangtonstarcentre.org/
significance – the degree to www.psiweb.org/newcareers. students realised that while index.php/mbp-squared-link.
Summer 2013 | 15
Veer
Want to bring
cutting-edge
science to the
classroom?
We’re looking for trainee and
newly qualified post-16 biology
teachers to join the Big Picture
advisory board.
Are you a teacher in the UK? If so, you can order a class set. Email publishing@wellcome.ac.uk
Name:
Job title:
Organisation:
Address:
Email address: