from BBC Learning English.

I'm Neil.

And I'm Sam. Predicting the future is not easy,

but that's exactly the job of opinion pollsters - researchers who ask people

questions to discover what they think about certain topics. Often their aim

is predicting which political party will win in an election

by asking members of the public how they intend to vote.

But predicting the future is never 100 per cent accurate,

and opinion polls don't always get it right.

In 2016, few pollsters predicted a victory for Donald Trump over Hillary Clinton

in the U S presidential election. And in the 2020 US elections,

most polls predicted

Trump would lose to Joe Biden, by a much larger amount

than he actually did. These mistakes sometimes called 'misfires',

when things do not work in the way intended,

have damaged the reputation of opinion pollsters. In this programme,

we'll be taking a look into the opinion polling industry and of course,

learning some useful vocabulary as well.

But first I have a question for you,

Sam, it's about another time when the opinion polls got it wrong.

Few pollsters predicted that Britain would vote to leave the European Union

in the 2016 Brexit referendum,

which in the end, it did.

But what was the final split between those who voted to leave

and those who wanted to remain?

Was it a) 51 leave to 49 remain,

b) 52 leave to 48 remain, or

c) 52 remain to 48 leave?

I think it was b)

52 per cent voted to leave and 48 per cent to remain.

OK, Sam, I'll reveal the answer at the end of the programme.

One of the biggest polling companies

was founded by George Gallup, born in 1901

on a farm in Iowa, Gallup was a student of journalism.

He wanted to know people's opinion on a range of subjects

and came up with a simple idea -

why not try asking them?

Here's G Elliott Morris,

a data journalist from the Economist, explaining more to BBC

World Service Programme, More or Less.

And,he publishes his dissertation

on this - how to measure what people want, basically.

And he gets hired by a much bigger advertising agency

in New York called Young and Rubicam, and they basically give him

a blank cheque to do their research, to figure out how to call people,

how to talk to them, to figure out if they remember or liked a certain product.

Basically, to figure out early methodologies

in advertising and then by 1931 or so, he's wondering

well, if it works for toothpaste

why not politics?

George Gallup tried to figure out what customers wanted to buy.

If you figure something out,

you finally understand it or find a solution to a problem

after thinking about it a lot.

Later, he was hired by a New York advertising agency to find out

people's opinion of consumer products like toothpaste and soft drinks.

George was given a 'blank cheque' - an unlimited amount of money

and freedom to do his job.

At this time, polling was focused on consumer preferences,

not politics.

But asking people about their political views is

a lot more complicated than asking them about toothpaste.

Making accurate election predictions.

depends on polling a sample group of people

who accurately represent the population as a whole. One of the reasons

for pollsters' failure to predict Trump's election in 2016,

is that they didn't ask enough white, non-college educated voters.

So, polling is a very complex process -

one which is never totally reliable,

according to G Elliott Morris, speaking again here to BBC

World Service's More or Less...  

If people were understanding this process that is generating all these polls,

then they would understand polls as less, sort of, precise tools,

tools they definitely can't offer the laser-like predictive accuracy

we've come to expect from them.

then the difference between pollings'

expectations and performance wouldn't be so stark.

Opinion polls can estimate the outcome of an election,

but they can't give us laser-like accuracy.

If you describe something as 'laser-like' you mean it

it's very accurate and focused, like a laser.

If people understand how hard

it is to predict the future,

they might be more realistic about

how accurate opinion polls can be. Then differences between a prediction

and the final result would not be so stark - obvious and easily visible or harsh.

Predicting the future is difficult,

otherwise everyone would be a lottery winner by now.

Maybe it's not opinion polls that have broken,

but our desire to know the future that's the problem.

OK, it's time to reveal the answer to my question about the Brexit referendum.

I said the final result was 52 per cent for leave,

and 48 per cent for remain.

Which was the correct answer. And another example of an opinion poll 'misfire' - a situation

where something does not work as intended.

OK, let's recap the rest of the vocabulary

from this programme about opinion pollsters -

people who conduct polls

asking the public their opinion on particular subjects,

especially politics.

If you figure something out.

you finally understand it or find the solution to a problem

after thinking long and hard about it.

If someone gives you a blank cheque,

you have unlimited money and freedom to complete a task.,

When you describe something as 'laser-like', you mean that

it's very accurate and precise.

And finally, the adjective 'stark' has several meanings,

including 'obvious', 'harsh' and 'plain'.

Once again, our six minutes are up. Bye for now.

Bye bye.