Jeffrey Saltzman's Blog

Enhancing Organizational Performance

Prediction 2017

leave a comment »

The start of a new calendar year is upon us.  Everyone wants to know what is in store for us, and there is no shortage of those willing to make predictions. After such an abysmal year in trying to predict election outcomes you would think there would be some hesitation in offering prognostications.

There are those who extensively study prediction making, such as Phil Tetlock and his Good Judgement Project. He documented his work in a book he co-authored with Dan Gardner, called Superforecasting, the Art and Science of Prediction. He identified people, by nature of their approach to problem solving and prediction, who were much better than average at determining likely outcomes to an event or situation. The very best of them knew that they did not have all the answers, were always questioning themselves, acknowledged their mistakes and failures, were always looking for more information, and were willing to adjust and change their predictions based on that new information. He summed up their approach as “try, fail, analyze, adjust, try again”.

Phil’s recommendations for doing good forecasting include:

  1. Use predictive techniques on problems that can be predicted. Don’t try to predict, for instance, what life will be like 50, 100 or 150 years from now. The answers will be sheer nonsense.
  2. Break complex predictions into a series of less complex predictions. You are better off trying to predict one aspect of life 50 years out, not everything.
  3. Consider both insider and outsider sources of information. An insider (your contractor) might say, in predicting how much a kitchen remodel might cost, that it would cost X. By doing your research you would find that kitchen remodels more than a third of the time run over budget by substantial amounts (outsider source of information) and you should adjust your contractors estimate accordingly.
  4. Don’t over or underreact to evidence as it becomes available and don’t fall into the trap of wishful thinking (just because you want something to be true does not make it true).
  5. Open yourself up to considering both sides on an issue. (But make your consideration based on scientific evidence).
  6. Think about your predictions with percentages of certainty and uncertainty. Very few things in life are binary, either/or. Think about your prediction as a likelihood number – say 65% or 75% certain.
  7. Don’t become frozen by “analysis paralysis”. Balance making a prediction needed to take action, with the need to continually refine your predictions.
  8. Learn from your mistaken predictions, but don’t fall into the cognitive traps and biases commonly found.
  9. Listen to others and consider their insights.
  10. You get better at making good decisions by practicing and honing your decision-making skills and abilities.
  11. Don’t blindly follow these or any other rules. Realize that each situation can be different.

In today’s environment, I would add, 12. Don’t give into fear. Realize that many people will be trying to influence your personal predictions and your perceptions of future events by manipulating your emotions and thinking patterns in order to achieve their own ends. And 13. With the abundance of fake news and fake information that is now all around us, check your sources. Make sure they are legitimate. We all need to take lessons on how to be better at predicting, for the decisions and predictions that we make have real consequences on people’s lives.

Psychologists spend a lot of time trying to predict behaviors. In my first job after graduate school, I was tasked with trying to predict which person would make the best steel worker, the best bearing manufacturer, executive assistant, or corporate manager among other occupations. All of those predictions were based on probabilities and not absolute judgments. And I would have to say that many (maybe most) did not understand that. In order to make better predictions in this area, one technique used by psychologists is called a “multiple hurdle” approach. This means that the prediction of who would be the best steel worker, for instance, had multiple decision points where a candidate either passed for failed. Did they score high enough, compared to other successful steel workers, on a test of math ability? The kind of math required to do the job. If they did, they moved onto the next hurdle. If not, they were rejected.  Could a person who failed the math test make a good steel worker? Yes, but the odds were longer. The more hurdles, in general the better the prediction. The cost of each hurdle had to be taken into consideration against the added value it gave in prediction. (In my work, at the time, I determined that the most predictive assessment for steel workers was the Bennett Test of Mechanical Comprehension, it measured people’s understanding of how the physical world operated. It was a test, mostly non-verbal, of innate understanding of the properties of mechanics and physics).

At my current company, OrgVitality, which I founded with several partners, a good portion of our work on assessing organizational culture is aimed at prediction as well. Some of the questions we attempt to answer for clients include: What pattern of responses to an organizational survey, will best enable, making it more likely, that an organization can fulfill its strategic mission? (And how do you increase the likelihood of success?) Where in the organization are there response patterns that are indicative (more likely to occur) of higher levels of innovation, customer service, sales success, safety, ethics, etc.? What response pattern is indicative of less turnover and more future success for employees? And we like to examine those factors in terms of present performance and future potential. Scott Brooks and I wrote a book called Creating the Vital Organization which examines our approach in detail. One aspect of prediction the company is working upon and continually fine-tuning, is to determine, through various algorithms, which comments an employee or customer may make which are the most valuable in terms of organizational improvement and to have the very best rise to the top out of a pile of tens or hundreds of thousands.

Some pressing questions of prediction in the public sphere today revolve around violence and mass killings.  Can we predict who will cause mayhem and violence in our society and importantly can we prevent the violence from happening? If we look at 85 tracked mass shootings from 1982 to 2016 the demographics of those who committed these crimes in our society, a pattern emerges. Most of the mass murders were committed by around 30 year old, white men, a significant portion of which had a history of some sort of mental illness. The vast majority, as far as can be determined were not Muslim (about ½ of 1% were Muslim), even though those committed by Muslims garner much attention. Most obtained their guns legally and were not prohibited from owning the weapon.  Who has access to guns to commit these crimes? The largest percent are more likely to live outside of the northeastern part of the USA, 41% are white, 51% live in rural environments, 49% self-identified as Republicans (22% as Democrat) and 41% identify themselves as having a conservative ideology.  Using this profile, logic and statistics (and I am doing this to point out a flaw in this reasoning), if we take guns away from 30 year old white, non-Muslim men, who live in rural environments and have conservative beliefs we will greatly reduce the incidence of mass murders in our country. The flaw in this logic should be obvious to you. Of the 30 year old, white, non-Muslim men, who live in rural environments and have conservative beliefs, less than a fraction of one-tenth of 1% will commit mass murders. 99.99% of them will not commit any crimes with their weapons. Taking away the guns from all of them is like putting out a match with the Pacific Ocean. It simply does not make sense.  Yet there are those who are willing to use this very same type of flawed logic to castigate all Muslims or those with mental illnesses. It doesn’t work there either. Americans are being skillfully manipulated to come to erroneous conclusions.

I am reminded of an adult education class discussion I attended, where there was much discussion of how violent the 20th century was, with huge numbers being slaughtered in WWI and WWII. I made the statement that in general humanity over the millennia was becoming less violent (and there is one hypothesis that we are self-domesticating, weeding out the most violent among us). My statement was met with much derision. Those who ridiculed my statement where falling prey to at least one human bias – what you see is all there is. If you look at the larger context, which they were not, you realize that historically, the conquerors, the crusaders and warlords of millennia ago, killed much larger percentages of the population than what occurs today. For instance it is estimated that Genghis Khan killed 40% of the total global population as he conquered much of Asia. I am certainly not condoning or dismissing the levels of violence that occur today.

So what can we predict for 2017? Anything beyond the sun will come up in the morning and set at night? Yes, probabilistic predictions can be made, as long as they are about specific manageable, measurable issues which are surround by scientific facts, (important facts, not red herrings), all the information available is taken into consideration and you are careful not to fall into the traps of human bias and predisposition. Happy predicting!

 

Written by Jeffrey M. Saltzman

December 25, 2016 at 10:49 am

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: