Roger Federer defied the expectations of many when he won the 2017 Australian Open against Rafael Nadal.
Before the final, Federer had won just 11 out of 34 matches against Nadal. Numerous experts believed that Roger’s backhand would break down as Nadal unleashed top-spinning, high-bouncing forehands across to Roger’s single-handed backhand, just like in many of their past matches.
So how is this even remotely relevant to the world of HR and selection?
Experts tried to predict the manner in which Federer and Nadal would play against each other, and therefore the outcome of the match. Likewise, HR professionals try to predict how a candidate will behave, and therefore how well they will perform in a role.
In HR and organisational psychology circles, it is oft quoted that “the best predictor of future behaviour is past behaviour”. Unfortunately, this is sometimes treated more like a fundamental law of physics than a handy rule of thumb.
To say past behaviour is “the best” predictor oversimplifies the issue, yet it is true that it often gives an indication of probable future behaviour. Imagine, Heidi has been exercising three times per week for the last year, so we would not be surprised to find her exercising next week, or in future weeks. This might even give an indication of how well she will perform in a fun run in 3 months’ time. In a work setting, a sales person’s high level of people contact and networking in their last role might predict similar behaviours in the role you are considering them for, hopefully translating into high revenue generation.
The Best Predictor
“The best predictor of future behaviour is past behaviour”, has been attributed to everyone from psychologists, such as Albert Ellis, Walter Michel, and B.F. Skinner, to writers such as Mark Twain.
One of the people to explore this idea in depth was the American psychologist Paul Meehl. He wrote, “…behavior science research itself shows that, by and large, the best way to predict anybody’s behavior is his behavior in the past…”. However, 'Meehl’s Malignant Maxim' was focused specifically on legal judgments about whether someone would re-offend. He also believed there were substantial “sources of error” in using simple predictions based on an offender’s past behaviour.
But back to predicting job performance and tennis matches.
Caveats for Basing Predictions on Past Behaviour
If predicting future behaviour were as simple as looking at past behaviour, you might expect that Federer would have a good chance of losing, based on his previous backhand performance against Nadal and 32% win rate in those matches. But psychological research shows us there are several crucial provisos to take into account, before assuming past behaviour will strongly predict future behaviour:
1. The future situation should be very similar to the past situation
Yes, there were many similarities between the Australian Open final 2017 and previous encounters. Nadal would very likely be unleashing big top-spinning forehands to Federer’s single-handed backhand. Moreover, previous grand slam matches heavily favoured Nadal, 9 to 2.
However, there were some key differences from past situations. The courts were much faster than usual, suiting Federer’s overall style of play on his backhand (and his forehand). In addition, it made a difference that Nadal only had a two-day gap between his semi-final and the final, compared to Federer's three days, particularly with both players being older than in past encounters.
While these situational differences would not nullify the history between the two players, it might make the odds closer to 50/50, or at least make us more cautious in our predictions.
Selection Example
A sales candidate, David, has had excellent sales in his previous role. He is hired, but only performs at a mediocre level. Here’s the problem - his previous role was selling combine harvesters to farmers. His new role is selling strategic branding services to corporations. That’s two very different situations, different types of customers, and different products, requiring a different approach (i.e. set of behaviours).
2. Prediction is best when there is a short time between past and future behaviour
In the case of Nadal and Federer, they hadn’t played against each other since 2015, when Federer won, and before that since 2014. Their matches stretch back to 2005. Basing a prediction over such a long time period reduced the validity of prediction based on previous matches.
Imagine basing a selection decision on behaviours stretching back 12 years in a candidate’s career. Over that time period people often change. Which brings us to the next point.
3. The person must still be basically the same now as they were then
People change for a variety of reasons. General experience can give us a different perspective on both ourselves and the world, as can significant life events. A person who at 22 had been insensitive or cheeky to customers because of a lack of maturity might not have any such issues by the time they are 32.
We don’t know exactly how Federer might have changed as a person, but we know he had been using a tennis racquet with a larger head since 2014 – this allowed him to hit his backhand flatter and more aggressively, with less risk of missing the sweet spot. He also reacted unexpectedly in the fifth set when behind, lifting and playing aggressively rather than becoming nervous and tentative. It turns out that he had prepared with a different mindset, that he would play the ball, rather than his opponent, and would continue to play ‘attacking’ tennis regardless of the risk.
People adapt because of the feedback they get from the world. The consequences of a behaviour can lead to a change in that behaviour. Sports people are a prime example of this, frequently changing their approach based on whether they win or lose.
Selection Example
Let’s say that very early in their career a nurse, Alex, occasionally misses key details in patient information and administering care. This is clearly an undesirable work behaviour which you might believe will predict the same behaviour in future. One day, Alex makes a mistake that puts a patient at risk. There is an investigation and Alex is thoroughly mortified by her mistake. The consequences of the behaviour cause her to check the details far more carefully in future, and there are no further incidents. Why? Because the poor behaviour has been eliminated through feedback.
4. Frequent, reflexive behaviours are more predictive than behaviours based on conscious intentions.
The intentions of people can make a big difference to whether they carry out a behaviour. This has been written about by researchers Icek Ajzenand Martin Fishbein, and others such as Sheppard, Hartwick & Warshaw (1988).
True, you could say hitting a tennis ball is a high-frequency, reflexive behaviour, resulting in a fairly consistent style of play. But for tennis players, they have the capacity to rework their style through conscious effort and coaching. That is exactly what Federer has done.
Selection Example
In a work situation, high-frequency, unconscious behaviours might include the manner in which someone interacts with others. Picture John, a negotiator who is only moderately effective. Two of his intentional behaviours are making counter-offers quickly after an offer is made and the use of phrases such as “this is a very fair offer”.
Through training, John becomes aware of research that shows these two behaviours are actually counterproductive, reducing his effectiveness. He removes them from his repertoire and voila, he improves as a negotiator and becomes an eminently employable candidate.
Conclusion
Past behaviour may be a fairly good predictor of future behaviour in some cases, but before jumping to conclusions about just how predictive, you should take into account similarity of situations, the time period, how much the person may have changed, and how habitual the behaviour might be.
And who'd have thought, Federer actually won!
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behaviour. Englewood Cliffs, NJ: Prentice Hall.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behaviour: An introduction to theory and research. Reading, MA: Addison-Wesley.
Franklin, K. (2013). The best predictor of future behavior is...past behavior". Psychology Today.
Meehl, P. E. (1989). Law and the fireside inductions (with Postscript): Some reflections of a clinical psychologist. Behavioral Sciences and the Law, 7, 521-550.
Sheppard, B. H., Hartwick, J., & Warshaw (1988). The theory of reasoned action: A meta-analysis of past research with recommendations for modifications and future research. Journal of Consumer Research, 15, 325-343.
