A car crash, a gradient, and the power of a single word

It’s 1974, a whole year before Betamax will be released in North America and two years before Japan will invent the VHS.

But that isn’t stopping Elizabeth Loftus and John Palmer. They’ve got their hands on a rare Cartivision tape player, one of the first ever made. And the cassette (which is roughly the size of a 400-page hard-book book) contains just one short piece of footage – a car crash.

Elizabeth and John wheel their tape player into an unremarkable room in the Guthrie Hall at the University of Washington and hook it up to a television. There’s a knock at the door. It’s the first case study. They enter and sit at a chair.

The cassette is loaded into the Cartivision player and, with a fizz and a crackle, the TV screen flickers to life. The case study in the chair watches as the unmistakable image of an automobile comes into focus, rams head-first into another vehicle and then, almost as abruptly as it started, the video ends.

Later in 1974, Elizabeth and John publish an article titled An Example of the Interaction Between Language and Memory. In it, they argue that the choice of words in leading questions can influence the way we respond.

They note that changing even one word can sway and distort our memories. They posit we could even create false memories. They also point out a correlation between the strength of a word received and the strength of the response given. This suggests a ‘gradient of effectiveness’ for words.

And that was 1974.

How would you describe a car crash?

We’ll talk more about Loftus & Palmer – and what their research means for us copywriters – in just a moment.

For now, let’s understand exactly why they felt the need to show videos of car crashes to unsuspecting members of the Seattle public and how they managed to radically change our understanding of human memory. Our story starts in the early 1970s.

This was a bad era to be a driver in the United States. It was an even worse era to be a traffic officer or city transport official. Car crashes were a national epidemic, and the authorities were at their wits’ ends. In 1972 alone, 26 in every 100,000 people died as a result of a traffic collision. Officers were working overtime just to respond to the sheer number of incidents.

This invariably meant the courts were packed too, and it was here that law officials noticed another worrying trend. Through their own research, district lawyers had pointed out that accounts given by eyewitnesses were often rife with false information and errors of judgement. At a time when forensic evidence was nowhere near as mature as it is today, prosecutions often relied on personal testimony – yet eyewitnesses were prone to reporting things that did not align with physical evidence, contradicting the reports of other eyewitnesses and even going against reports they themselves had made in the past. So adamant were some witnesses that, even when proved wrong, they would stand by their false testimonies anyway. The courts needed a way to sort the wheat – the real, honest, reliable eyewitness accounts – from the chaff.

On this basis, Elizabeth Loftus and John Palmer concocted an experiment with a simple aim: they wanted to understand why eyewitness inconsistencies were so… consistent. They could have chosen to focus on any type of court case, from homicides to fraud investigations, but they decided to focus on car crashes: not only were there an abundance of court cases for them to analyse, unlike other incidents, they were mostly free from emotional motives and could be measured with statistical accuracy.

Furthermore, car accidents happened to be an area where eyewitness inconsistencies could be either positive or negative, great or small. For instance, when asked to judge the speed of a vehicle at the time of a collision, some eyewitnesses would go higher while others went lower. Some reports would be off by 2 or 3 mph, some by 12-15mph and, in some cases, they could be off by as much as 30mph. People could have seen a car crash at 40mph and reported it as travelling anywhere between 10 and 70mph.

So Loftus & Palmer kicked off a two-phase experiment. In phase one, as you already know, they showed participants a video of a car crash. In phase two, they asked participants to fill in a questionnaire – some participants did this immediately after seeing the video while others were asked to wait a week.

The questionnaire focused mainly on innocuous details in the video (‘Was the driver a man or a woman?’, ‘How many lanes were on the road?’) and were all exactly the same except for one question, which read: ‘About how fast was the red car going when it _____ the white car?’ Here, the blank was randomly filled with one of five words: smashed, collided, bumped, hit or contacted.

Their results will surprise you. No wait… they’ll shock you.

What Loftus & Palmer found was that their choice of word affected their participants’ responses. In the above order, average responses ranged from 40.5mph for ‘smashed’ to 31.8mph for ‘contacted’, nearly 10mph difference altogether.

Furthermore, there was a correlation between the strength of this question and responses to other questions. For instance, one question asked ‘did you see any broken glass?’ There was no broken glass in the video, but the participants who had been asked with the stronger words had a higher chance of answering ‘yes’.

Picture a police officer at the scene of the accident asking you a similar question about the speed of the cars. You know the cars impacted at 40mph, but he asks, “what speed were the cars going when they smashed into each other?” You think, smashed? Well, it seemed more like a bump to me, but if the police officer is asking about the ‘smash’ I guess I’ll say…“About 50mph, officer.” 50mph is 10mph over the speed limit. Suddenly we’ve got ourselves a court case.

Meanwhile, in a parallel universe, the police officer asks you a different question. He asks, “what speed were the cars going when they contacted one another?” Err, contact is an understatement, I mean just look at the state of them! But come to think of it, I guess it wasn’t much more than a contact. Probably happened at around… “30mph, officer.” 30mph? Well there’s no point taking this to court. One party must have just stopped paying attention for a moment. The insurance companies can pick this one up.

And you, straddling the borders of these parallel universes, have noted the same thing that Loftus & Palmer drew in their conclusion, namely…

Stronger questions lead to stronger responses (and vice versa).

Loftus & Palmer gave two explanations for their findings: first, they said the choice of even a single word could not only impact people’s choices, it could distort their memories; second, they acknowledged that response-bias factors could be at play, i.e. participants who weren’t sure of their answers may have given answers they feel they were ‘supposed’ to give (“Well, it looked like 30mph, but I’ll sound stupid if I guess too low, so I’ll go with 40”).

Ultimately, Loftus & Palmer argued that responses to questions can be changed according to a ‘response gradient’, determined by the strength of the active verb in that question. This suggests that we remember things not by facts and statistics but by a general feeling for the event, and that over time this feeling can be changed, manipulated or exacerbated depending on the language we associate with the feeling.

The thought of two cars “smashing” draws connotations of carnage. But the thought of two cars “contacting” is insignificant by comparison. You’re probably more likely to remember a “smash”, just as you’re more likely to think of “contact” in a positive light.

Yeah but what does this have to do with copywriting?

It should go without saying that your choice of words is vital to the effectiveness of your copy. After all, we copywriters developed a whole discipline around the infinite permutations of language and called it tone of voice.

But Loftus & Palmer’s study helps us go a bit further. Their suggestion of the existence of a ‘response gradient’ helps us to understand how and why certain words can change feelings (and even facts).

We can use this understanding to affect almost any kind of statement or question. For instance, see a how single word can alter declarations (‘The car smashed into the wall’ vs. ‘The car bumped into the wall’), exclamations (‘What an awful day!’ vs. ‘What a bad day!’) and imperatives (‘Give a kind donation today’ vs. ‘Give a life-changing donation today’). In theory, we can change almost any word in almost any sentence and expect a different response.

And you know what that means…

Test, test, test.

Now I digress. You’re probably not going to A/B test every single word in every single email, letter or landing page (but if you do, please publish your results!), but you could take Loftus & Palmer’s study into account when testing your most important phrases, such as your calls to action.

Why not start by testing words on yourself and your colleagues? For instance, when writing a new call to action, ask which of these hits you hardest:

“Provide a donation today.”

“Help us with a donation today.”

“Give a donation today.”


“Donate today.”

Your intuition will give you a good starting point. From there, it’s a case of implementing a testing strategy. That’s up to you. But with the right testing criteria in place, you can start to build up a little black book of ‘power words’ and phrases, recorded in the form of your own response gradient. This will give you a proven base from which to elicit stronger responses from your audience.

A final word on Loftus and Palmer.

In their conclusion, Loftus & Palmer say memory ‘is not like a tape recorder’. At a time when tape recorders barely existed, this comment shows just how ahead of the time their study was.

Elizabeth Loftus and John Palmer helped us to understand how susceptible we are to manipulation, and how easily we can be swayed by even a single word. Therefore beware every word you use. Know that even one small change could radically alter the overall tone and effectiveness of your copy.

And finally – drive safely. Or should that be drive carefully? Drive cautiously? Drive gingerly? Drive daintily? Drive fastidiously? Hmm… what do you reckon?



This is just one of the behavioural lessons we apply to our work at On Agency. To find out how our behavioural approach could help boost your fundraising income, contact us today.

Written by Jonjo Maudsley, Copywriter.

Notice: ob_end_flush(): failed to send buffer of zlib output compression (0) in /home/onagency/public_html/cms/wp-includes/functions.php on line 3783

Notice: ob_end_flush(): failed to send buffer of zlib output compression (0) in /home/onagency/public_html/cms/wp-content/plugins/really-simple-ssl/class-mixed-content-fixer.php on line 111