Johnson Graduate School of Management Prof Examines FiveThirtyEight Election Predictions using @RISK
The midterm Senate race is fast approaching—and so are the speculations on its outcome. The stats-centered news site, FiveThirtyEight, builds interactive forecasts that are updated almost daily. As of this writing, the model predicts that Democrats have a 42.4% chance of retaining the Senate next year.
"They’ve also included a correlation, of a type in their model," says Dr. Lawrence Robinson, Professor of Operations Management at Cornell University’s Johnson Graduate School of Management. "But they do not explicitly use a correlation coefficient—instead, they change the distribution of the candidate’s lead." Robinson explains that FiveThirtyEight introduces correlation through an additional random variable representing what they’ve labelled "national error," which they generate and add to the mean margin of victory of every candidate. "Unfortunately, nowhere in their post do they specify the probability distribution for his new ‘national error’ random variable," Robinson says. "Thus it is not possible to know how correlated the individual races are with one another."
Robinson wanted to devise a way to arrive at these forecasts using his own statistical methods, and to use a correlation that is explicitly defined. Instead of just using FiveThirtyEights’s "Leader’s chance of winning," which was only given to the nearest percent, Robinson started with the mean and (estimated) standard deviation of the margin of victory, and calculated the probability of winning by assuming that the margin of victory on Election Night was normally distributed. "Although Silver says he assumes that the victory margin is leptokurtic [has fat tails] for finding the probability of winning, he never specifies its probability distribution," says Robinson. "I found that the standard assumption that the margin was normally distributed better matched his reported analysis."
Robinson built a Monte Carlo simulation model in Excel using @RISK, treating the outcome of each race as a Bernoulli (0/1, win/lose) random variable. He then introduced a correlation matrix that captured the correlation between every pair of races, and ran 27 different simulations (each one simulating 400,000 elections) for correlations ranging between 0% and 100%. His results closely match that of FiveThirtyEight’s, showing the probability that Democrats will retain control of the Senate as a function of the correlation among races. Robinson points out the benefit:
The advantage of this approach is that
we specify the correlation precisely, and
that we conduct robustness analysis to
see how the results change with the
Dr. Lawrence Robinson - Cornell University
Johnson Graduate School of Management
Professor of Operations Management
» More detail about Robinson’s model
» Download Robinson’s spreadsheets
» More about @RISK
Join Other Thought Leaders Next Month in New Orleans:
Palisade Risk & Decision Analysis Conference
Palisade’s Annual Risk Conference is right around the corner, taking place on November 19th and 20th in the heart of the downtown New Orleans. As our headline event of the year, the annual Risk Conference features four tracks of informative sessions of case studies and software training, along with one-on-one expert sessions and of course, lots of time for networking and collaboration.
Also, don’t miss the keynote presentation by Senior Manager of Innovation Strategy Dave MacAdam of Novelis Global Research & Technology Center, which will take you through the method he used (and continues to use) to bring @RISK into his company and improve how his team values high-risk R&D projects.
- Dr. Martin Pilch, Sandia National Laboratories
- Dr. Robert Forbes, University of California, Irvine
- James Joyner, NASA Kennedy Space Center
- Dr. Robert C. Patev, US Army Corps Of Engineers
- JD Solomon, CH2M Hill
» Register now to attend
There are three easy ways to get the DecisionTools Suite for your students:
Course Licenses available from Palisade. These are economical, annually renewing bulk licenses for both
network lab and laptop installation. Palisade offers a complimentary Live Web Training session to all instructors who order a Course License.
» Learn more
Student Versions from Palisade. These are 12-month versions available for purchase individually by students through either the Palisade web site or the school store.
» Learn more
Textbook Editions available bundled with textbooks.
These are time- and model-limited licenses.
» Learn more
Get DecisionTools Suite 6 for your course by contacting:
Palisade Academic Sales
800-432-7475 / 607-277-8000
The DecisionTools Suite is a great way to present quantitative techniques in a straightforward, easy-to-understand way for students.
Jobshop and Production Floor Scheduling
Excerpted from Evolver Solutions for Business,
by Roy Nersesian, published by Palisade Corp.
A jobshop is characterized by individual jobs taking different amounts of time for each operation. Suppose that a jobshop consists of four operations: sanding, varnishing, etching, and finishing. Each job is unique with regard to the amount of time required for each operation. This example has 6 jobs. Once a job is sanded, the sanding operation can proceed immediately to the next without delay. Once the first job is sanded, it can immediately be varnished, etched and finished without intervening delays. However, there are two conditions that have to be met for the remaining scheduling. For the second job to be
varnished, it must first have been sanded and the first job must have completed being varnished. What is the length of time to do all the jobs in sequential order? What is the sequence of jobs to optimize performance; that is, minimize the length of time to complete all six jobs? Find out by using Evolver and this downloadable example.
» Download example file and view step-by-step instructions
Delimiters and Discrete Distributions
Q: I have a RiskPoisson(3) distribution, and I click Define Distributions, or Browse Results after a simulation. I set the delimiters to 0 and 6, and @RISK shows a probability of 91.7% between them. But Excel's POISSON.DIST(6,3,TRUE) shows a cumulative probability of 96.6%. Which one is right?
This seems strange at first, but there's an explanation. This is nothing special about the Poisson distribution; it applies to RiskBinomial, RiskDiscrete, and all the other discrete distributions.
@RISK and Excel are both right, but they're measuring different things. Excel is reporting the cumulative probability from x=minus infinity to x=6. @RISK reports the cumulative probability from x=0 to x=6. But the Poisson distribution doesn't extend to negative x, so why aren't those two the same?
» Find out why in Palisade's Knowledge Base
The DecisionTools Suite is part of complex analyses published in white papers.
Quantitative Risk Assessment on the Public Health Impact of Pathogenic Vibrio parahaemolyticus in Raw Oyster
Center for Food Safety and Applied Nutrition
Food and Drug Administration, U.S. Department of Health and Human Services
The Center for Food Safety and Applied Nutrition Food and Drug Administration
of the U.S. Department of Health and Human Services analyzed scientific data that was used to model and predict the public health impact of pathogenic Vibrio parahaemolyticus in raw oysters. The assessment focused on comparing the relative risk among different geographic regions, seasons, and harvest practices. The scientific data and the mathematical models developed during the risk assessment, in conjunction with @RISK and Monte Carlo simulation, facilitate a systematic evaluation of strategies to reduce the public health impact of pathogenic Vibrio parahaemolyticus associated with the consumption of raw oysters. Questions that were addressed include:
- What is known about the dose-response relationship between consumption of Vibrio parahaemolyticus and illnesses?
- What is the frequency and extent of pathogenic strains of Vibrio parahaemolyticus in shellfish waters and in oysters?
- What environmental parameters (e.g., water temperature, salinity) can be used to predict the presence of Vibrio parahaemolyticus in oysters?
- How do levels of Vibrio parahaemolyticus in oysters at harvest compare to levels at consumption?
- What is the role of post-harvest handling on the level of Vibrio parahaemolyticus in oysters?
- What reductions in risk can be anticipated with different potential intervention strategies?
Read the full paper here
The Academic Global Leader in
Risk and Decision Analysis
Palisade is the world’s leading provider of risk and decision analysis software and solutions. Founded in 1984, its products @RISK and the DecisionTools Suite are used by over 95% of the Fortune 100, in nearly every industry around the world. Palisade is headquartered in New York State, and has offices in London, Sydney, Rio de Janeiro, and Tokyo.
What is The DecisionTools Suite?
The DecisionTools Suite is the world’s only integrated set of risk and decision analysis programs. The Suite includes @RISK for Monte Carlo simulation, PrecisionTree for decision trees, and TopRank for “what if” sensitivity analysis. In addition, DecisionTools Suite comes with StatTools for statistical analysis and forecasting, NeuralTools for predictive neural networks, and Evolver and RISKOptimizer for optimization. All programs work together seamlessly, and all integrate completely with Microsoft Excel for ease of use and maximum flexibility.
The Suite is offered in three affordable and flexible licensing options for those in academia: Student Versions, Course Licenses and Full Academic Versions.
» Schools using Palisade software
» Textbooks using Palisade software
|©2014 Palisade, visit www.palisade.com
798 Cascadilla Street, Ithaca, NY 14850 USA, +1 607 277 8000 firstname.lastname@example.org
Av. Presidente Vargas 435, Sala 501, Centro, Rio de Janeiro CEP 20071-001, +55 (21) 2586-6334 email@example.com
Palisade Europe UK Ltd
31 The Green, West Drayton, Middlesex, UB7 7PN, UK, +44 1895 425050 firstname.lastname@example.org
Palisade Asia-Pacific Pty Limited
Suite 404, L4, 20 Loftus Street, Sydney NSW 2000 Australia, +61 2 9252 5922 salesAP@palisade.com
Palisade Asia-Pacific, Tokyo Office
〒 150-8512 東京都渋谷区桜丘町 26-1 セルリアンタワー15F 電話 03-5456-5287