Professor Armstrong is internationally known for his pioneering work on forecasting methods, which he has worked on for 55 years.

Professor Armstrong, can you tell me, what is a forecaster?

A forecaster is someone who states what is likely to happen in a given situation, and how likely that is. In order to provide useful forecasts, forecasters need to use evidence-based procedures. Knowledge on evidence-based procedures is summarized as 139 evidence-based principles in my book, Principles of Forecasting. The knowledge comes from reviews by 40 forecasting experts from diverse fields of decades of research. The principles have been freely available at forprin.com since 2000.


Is that forecasting knowledge relevant, no matter what the forecasting problem is? In other words, can it be applied to all fields?

The forecasting principles apply to all situations. However, many people claim that it does not apply in their situation.


Are forecasts in some fields better than those in others?

Weather forecasting is a field that makes good use of forecasting principles. For instance, when weather forecasters predict a 60% chance of rain tomorrow, close to 60 out of 100 times they say that, it rains.

In contrast, business managers often use forecasts as motivational tools, and thus do not use scientific forecasting principles. Public policy is another field that often violates forecasting principles. This allows government officials to obtain forecasts that support favoured projects, such as mass transit schemes. Over-forecasting of revenues is common for such projects.


Is global warming a forecasting problem?

The question is not what happened in the past. The issue is what will happen to the climate in the future. Governments and major industries around the world argue for costly policies to stop what they forecast as the effects of man-made global warming. So, yes, it is first and foremost a forecasting problem.


Are forecasts of dangerous man-made global warming valid?

From the scientific point of view they are not valid.

The United Nations Intergovernmental Panel on Climate Change (IPCC) concluded that global mean temperature [changes] cannot be forecast because climate is too complex. They nevertheless rely on complex computer modelling to represent their assumptions about how the climate works. The outputs of such models are called “scenarios.” In effect, the IPCC tells stories, illustrated with computer graphics, about what would happen if their assumptions were correct.

Unfortunately, governments treat the IPCC scenarios as forecasts. Scenarios produce misleading forecasts, as described in Principles of Forecasting (p.519-540). Moreover, the IPCC assumptions lack scientific support, as explained in the three edited volumes of Climate Change Reconsidered published from 2009 to 2014.

Global warming alarmists claim that nearly all scientists agree that dangerous man-made warming will occur. However, surveying scientists’ opinions is not a scientific forecasting method. Nor is the claim true. It is difficult to find a list of scientists who do agree with the alarmist forecasts. Meanwhile the Oregon Petition shows that over 31,000 American scientists disagree.


What is the scientific truth about forecasting global mean temperatures?

Kesten Green, Willie Soon, and I are alone in our willingness to claim that we provide scientific forecasts of long-term change in global mean temperatures. Given the high uncertainty about the net effect of human carbon dioxide emissions on global temperatures, we only see natural changes in climate. Thus, we forecast no long-term trend, either cooling or warming. Our tests of forecast accuracy over the period from 1851 to 1975 found that for forecasts for 91 to100 years ahead, the models used by the IPCC had errors that were more than 12 times larger than errors from our “no-trend” model.

In an effort to encourage climate forecasters to test the accuracy of their forecasts, I proposed a bet with former U.S. Vice President Al Gore in 2007. I suggested we each put up $10,000, with the sum to go to the charity designated by the winner. The challenge was to forecast global mean temperatures for the next ten years. I forecast no trend. (Given natural variation, I faced a 1/3 chance of losing.) He refused the bet. Theclimatebet.com tracks monthly data on global mean temperatures to show how our bet would have fared had Mr Gore been willing to bet the IPCC “business as usual” scenario against my bet on “no trend.”


Why is the debate on climate warming so vicious?

Advocates of the global warming hypothesis do not want to treat it as a subject for scientific study. They refuse to compare alternative hypotheses or to cite literature that is inconsistent with global warming or to debate sceptical scientists. They claim it is too late for more research, given the horrific stories about the threat to the Earth. They claim that “climate skeptics” are not proper scientists, and they try to block skeptics’ ability to present their findings in journals and newspapers.


Do you believe that global warmers are cheating?

There are documented cases of cheating by those favouring the global warming hypotheses. “Climategate” revealed many of these. Also, unexplained revisions to temperature data have led to increases in the historical temperature trends.


You had a long controversy and a 2008 Senate Hearing about the extinction of polar bears? Right?

Yes. The polar bear population has been increasing in recent decades due to restrictions on hunting. Despite the strong upward trend in the polar bear population, government researchers forecast an immediate sharp downward trend based on forecasts of global warming. They violated 69% of the relevant forecasting principles. For example, they did not fully disclose their data. They refused to provide me with their data on the population of polar bears. (See my 6-minute testimony on YouTube for details.)

What other issues are you working on at the moment?

We have done research on forecasting for conflict situations.

How would you make forecasts about current conflicts such as those in the Middle East and the Iran nuclear program negotiations?

Our preferred method is Simulated Interaction. We ask naïve subjects to take on the roles of the key parties in a conflict, brief them on the situation, and ask them to interact with each other. After conducting simulations for a given situation with ten or more groups, we use the decision selected most often across the groups as the forecast of the actual decision. We tested the approach using real conflicts that were unfolding at the time and well documented (disguised) past situations. This approach reduced forecast errors by 47% compared to forecasts by game-theory experts.


Have you been commissioned to make these simulations?

Not recently. Regrettably, many analysts seem to prefer using their judgmental methods.

You also mention that you are applying your evidence-based approach to education. What is that about?

Benjamin Franklin, the founder of the University of Pennsylvania, said that our job as faculty is to develop and disseminate useful knowledge. That happens in some areas of the university (e.g. engineering) and is currently followed in the medical field. But teaching of evidence-based principles and techniques is less common in some other areas, such as business schools.


Is your skepticism over higher education in business schools inconsistent with the fact that you are a business school professor?

I make suggestions on how to improve things. Although business school professors produce much useful knowledge, it is seldom taught.

In addition, formal education removes students’ responsibility for their learning. So it is difficult to disseminate useful knowledge as programs are currently designed.


Are you changing your role as a teacher?

The Internet offers a more effective approach at a much lower cost. My paper Natural Learning in Higher Education summarizes evidence on this issue.

I am trying to adapt my role to education on the Internet. I make evidence-based knowledge available to anyone, for free, and provide learning exercises so that students can develop skills. This requires that students take responsibility for their learning, which I view as the most important advantage of the Internet.

In joining this movement, I will work to implement Ben Franklin’s idea that higher education should involve discovery and communication of useful knowledge. I regard useful knowledge as principles and procedures that have been verified by experimental studies.

Education Key Wide

Do you consider yourself controversial?

My career has focussed on examining evidence about important problems. This is what scientists should do. Often the evidence conflicts with common beliefs. I view that as progress. Unfortunately some people are offended, and this will often be the case. Scientists who discover things (I have been involved with about 70 discoveries) all suffer criticism. That’s life.


What gives you scientific credibility?

My credibility derives from following the scientific method, as it does for any scientist. For example, I try to build upon cumulative knowledge, and I provide full disclosure of data and methods. Many of my findings have been replicated and extended. To date, none of my “controversial findings” have been found to be in error.

What about your status in the academic world, in your University?

I think I was born for this work. I am a full-time professor and plan to keep going in my career as long as I can make contributions to the world through science. It is an exciting profession.





April 8, 2015

J. Scott Armstrong is Professor, University of Pennsylvania, Editor of Principles of Forecasting, a founder of the International Journal of Forecasting and the Journal of Forecasting, author of Long-Range Forecasting, and founder of forecastingprinciples.com. All of his papers are available at jscottarmstrong.com