Papers by Scott Armstrong
International Encyclopedia of Statistical Science, 2011
Forecasting is concerned with making statements about the as yet unknown. There are many ways tha... more Forecasting is concerned with making statements about the as yet unknown. There are many ways that people go about deriving forecasts. This entry is concerned primarily with procedures that have performed well in empirical studies that contrast the accuracy of alternative methods. Evidence about forecasting procedures has been codified as condition-action statements, rules, guidelines or, as we refer to them, principles. At the time of writing there are 140 principles. Think of them as being like a safety checklist for a commercial airliner-if the forecast is important, it is important to check all relevant items on the list. Most of these principles were derived as generalized findings from empirical comparisons of alternative forecasting methods. Interestingly, the empirical evidence sometimes conflicts with common beliefs about how to forecast. Primarily due to the strong emphasis placed on empirical comparisons of alternative methods, researchers have made many advances in forecasting since 1980. The most influential paper in this regard is the M-competition paper (Makridakis et al. 1982). This was based on a study where different forecasters were invited to use what they thought to be the best method to forecast many times series. Entry into the competition required that methods were fully disclosed. Entrants submitted their forecasts to an umpire who calculated the errors for each method. This was only one in a series of M-competition studies, the most recent being Makridakis and Hibon (2000). For a summary of the progress that has been made in forecasting since 1980, see Armstrong (2006). We briefly describe valid forecasting methods, provide guidelines for the selection of methods, and present the Forecasting Canon of nine overarching principles. The Forecasting Canon provides a gentle introduction for those who do not need to become forecasting experts but who nevertheless rightly believe that proper knowledge about forecasting would help them to improve their decision making. Those who wish to know more can find what they seek in Principles of Forecasting: A Handbook for Practitioners and Researchers, and at the Principles of Forecasting Internet site (ForPrin.com). Forecasting methods As shown in Figure 1, the Forecasting Methodology Tree, forecasting methods can be classified into those that are based primarily on judgmental sources of information and those that use statistical data. There is overlap between some judgmental and statistical approaches.
We show that absolutely minimizing functions relative to a convex Hamiltonian H:R^n →R are unique... more We show that absolutely minimizing functions relative to a convex Hamiltonian H:R^n →R are uniquely determined by their boundary values under minimal assumptions on H. Along the way, we extend the known equivalences between comparison with cones, convexity criteria, and absolutely minimizing properties, to this generality. These results perfect a long development in the uniqueness/existence theory of the archetypal problem of the calculus of variations in L^∞.
Proceedings of the American Mathematical Society
We examine the question of when the ∗ * –homomorphism λ : A ∗ D B → A ~ ∗ D ~ B ~ \lambda : A*_D ... more We examine the question of when the ∗ * –homomorphism λ : A ∗ D B → A ~ ∗ D ~ B ~ \lambda : A*_D B\to \widetilde {A}*_ {\widetilde {D}}\widetilde {B} of full amalgamated free product C ∗ ^* –algebras, arising from compatible inclusions of C ∗ ^* –algebras A ⊆ A ~ A\subseteq \widetilde {A} , B ⊆ B ~ B\subseteq \widetilde {B} and D ⊆ D ~ D\subseteq \widetilde {D} , is an embedding. Results giving sufficient conditions for λ \lambda to be injective, as well as classes of examples where λ \lambda fails to be injective, are obtained. As an application, we give necessary and sufficient conditions for the full amalgamated free product of finite-dimensional C ∗ ^* –algebras to be residually finite dimensional.
Leukemia, 2009
Recent years have seen great advances in the understanding of epigenetic gene regulation. Many of... more Recent years have seen great advances in the understanding of epigenetic gene regulation. Many of the molecular players involved have recently been identified and are rapidly being characterized in detail. Genome scale studies, using chromatin immunoprecipitation followed by expression arrays ('ChIP-Chip') or next generation sequencing ('ChIP-Seq'), have been applied to the study of transcription factor binding, DNA methylation, alternative histone use, and covalent histone modifications such as acetylation, ubiquitination and methylation. Initial studies focused on yeast, and embryonic stem cells. Genome-wide studies are now also being employed to characterize cancer and specifically leukemia genomes, with the prospect of improved diagnostic accuracy and discovery of novel therapeutic strategies. Here, we review some of the epigenetic modifications and their relevance for leukemia.
Clinical Cancer Research, 2010
Purpose-Patients with mixed lineage leukemia (MLL)-rearranged B-lymphoblastic leukemias (B-ALL) h... more Purpose-Patients with mixed lineage leukemia (MLL)-rearranged B-lymphoblastic leukemias (B-ALL) have an unfavorable prognosis and require intensified treatment. Multiple MLL fusion partners have been identified, complicating the diagnostic evaluation of MLL rearrangements. We analyzed molecular markers of MLL rearrangement for use in rapid diagnostic assays and found the immunomodulatory protein, Galectin-1 (Gal-1), to be selectively expressed in MLL-rearranged BALL. Experimental Design-Transcriptional profiling of ALL subtypes revealed selective overexpression of Gal-1 in MLL-rearranged ALLs. For this reason, we analyzed Gal-1 protein expression in MLL-germline and MLL-rearranged adult and infant pediatric BALLs and cell lines by immunoblotting, immunohistochemistry, and intracellular flow cytometry of viable tumor cell suspensions. Because deregulated gene expression in MLL-rearranged leukemias may be related to the altered histone methyltransferase activity of the MLL fusion protein complex, we also analyzed histone H3 lysine 79 (H3K79) dimethylation in the LGALS1 promoter region using chromatin immunoprecipitation. Results-Gal-1 transcripts were significantly more abundant in MLL-rearranged BALLs. All 32 primary MLL-rearranged BALLs exhibited abundant Gal-1 immunostaining, regardless of the translocation partner, whereas only 2 of 81 germline-MLL BALLs expressed Gal-1. In addition, Gal-1 was selectively detected in newly diagnosed MLL-rearranged BALLs by intracellular flow
AJP: Cell Physiology, 2006
It has been increasingly appreciated that aldosterone elicits acute vascular effects through nong... more It has been increasingly appreciated that aldosterone elicits acute vascular effects through nongenomic signaling pathways. Our previous studies demonstrated that aldosterone attenuated phenylephrine-mediated constriction in intact vessels [via phosphatidylinositol 3-kinase-dependent nitric oxide synthase activation] but enhanced vasoconstrictor responses in endothelium-denuded arteries. To determine the mechanism of this vasoconstrictor response, we assessed the effect of aldosterone on myosin light-chain phosphorylation and contraction in clonal adult human vascular smooth muscle cells. Acute aldosterone exposure mediated dose-dependent myosin light-chain phosphorylation, inhibited by spironolactone and phosphatidylinositol 3-kinase inhibition. These rapid effects of aldosterone were mimicked by estradiol and hydrocortisone and were also inhibitable by both spironolactone and eplerenone. In parallel to its effects on myosin light-chain phosphorylation, aldosterone mediated dose-de...
Forecasting global climate change: A scientific approach
The Golden Rule of Forecasting requires that forecasters be conservative by making proper use of ... more The Golden Rule of Forecasting requires that forecasters be conservative by making proper use of cumulative knowledge and by not going beyond that knowledge. The procedures that have been used to forecast dangerous manmade global warming violate the Golden Rule. Following the scientific method, we investigated competing hypotheses on climate change in an objective way. To do this, we tested the predictive validity of the global warming hypothesis (+0.03°C per year with increasing CO 2) against a relatively conservative global cooling hypothesis of-0.01°C per year, and against the even more conservative simple no-change or persistence hypothesis (0.0°C per year). The errors of forecasts from the global warming hypothesis for horizons 11 to 100 years ahead over the period 1851 to 1975 were nearly four times larger than those from the global cooling hypothesis and about eight times larger than those from the persistence hypothesis. Findings from our tests using the latest data and other data covering a period of nearly 2,000 years support the predictive validity of the persistence hypothesis for horizons from one year to centuries ahead. To investigate whether the current alarm over global temperatures is exceptional, we employed the method of structured analogies. Our search for analogies found that environmental alarms are a common social phenomenon, with 26 similar situations over a period of two hundred years. None were the product of scientific forecasting procedures, and in all cases the alarming forecasts were wrong. Twenty-three of the alarms led to government actions. The government actions were harmful in 20 cases, and of no benefit in any.
Parents do not send their children to school to learn how to speak. How then do children learn to... more Parents do not send their children to school to learn how to speak. How then do children learn to speak? The objective becomes obvious to children due to the frustration of being unable to communicate. Learning tasks allow for practice. Feedback is immediate and clear because adults love to help young learners. Applications of new knowledge are made so as to continue learning. Children take responsibility for all aspects of this “natural learning process.” Natural learning obviously works. Adults often use the natural learning process when they need to learn something that is important to them (Tough 1971). For example, people have taught themselves to speak a new language, use computers, learn math, play chess, or to play a musical instrument. They set objectives and manage this process by seeking resources and help from others, engaging in active learning tasks, getting feedback from others, and practicing applications. The motivation is intrinsic. Natural learning obviously works...
Parrot, the mascot of the PollyVote, provided predictions of the two-party, popular vote shares b... more Parrot, the mascot of the PollyVote, provided predictions of the two-party, popular vote shares by averaging forecasts within and across four categories of methods: trial-heat polls, forecasts from the Iowa Electronic Market (IEM), quantitative models, and experts' forecasts. Combining forecasts from methods within a category aggregates forecasts based on a similar approach and similar information. It is more powerful, however, to average across methods, especially when the methods draw upon di erent information (Armstrong, 2001). Using equal weights, the nal Polly-Vote forecast is computed as an average of the forecasts across the four categories, giving equal weight to each.
The ATLAS experiment at the Large Hadron Collider (LHC) will face the challenge of efficiently se... more The ATLAS experiment at the Large Hadron Collider (LHC) will face the challenge of efficiently selecting interesting candidate events in©¨collisions at 14 TeV center-of-mass energy, whilst rejecting the enormous number of background events. The High-Level Trigger (HLT = second level trigger and Event Filter), which is a software based trigger will need to reduce the level-1 output rate of kHz to Hz written out to mass storage. In this talk an overview of the current physics and system performance of the HLT selection for electrons and photons is given. The performance has been evaluated using Monte Carlo simulations and has been partly demonstrated in the ATLAS testbeam in 2004. The efficiency for the signal channels, the rate expected for the selection, the global data preparation and execution times will be highlighted. Furthermore, some physics examples will be discussed to demonstrate that the triggers are well adapted for the physics programme envisaged at the LHC.
The ATLAS experiment at the Large Hadron Collider (LHC) will face the challenge of efficiently se... more The ATLAS experiment at the Large Hadron Collider (LHC) will face the challenge of efficiently selecting interesting candidate events in©¨collisions at 14 TeV center-of-mass energy, whilst rejecting the enormous number of background events. The High-Level Trigger (HLT = second level trigger and Event Filter), which is a software based trigger will need to reduce the level-1 output rate of kHz to Hz written out to mass storage. In this talk an overview of the current physics and system performance of the HLT selection for electrons and photons is given. The performance has been evaluated using Monte Carlo simulations and has been partly demonstrated in the ATLAS testbeam in 2004. The efficiency for the signal channels, the rate expected for the selection, the global data preparation and execution times will be highlighted. Furthermore, some physics examples will be discussed to demonstrate that the triggers are well adapted for the physics programme envisaged at the LHC.
Journal of Marketing, 1975
The subject of social irresponsibility in marketing is of great current interest. However, the ma... more The subject of social irresponsibility in marketing is of great current interest. However, the marketing literature is vague about just what "social irresponsibility" is, why there is a problem, and what solutions are possible.
Psychological Reports, 1997
Studies suggest a bias against the publication of null (p > .05) results. Instead of significance... more Studies suggest a bias against the publication of null (p > .05) results. Instead of significance, we advocate reporting effect sizes and confidence intervals, and using replication studies. If statistical tests are used, power tests should accompany them. A number of studies have shown that peer review is biased against the publication of null (p > .05) or so-called " negative, " results. Sterling, Rosenbaum, and Weinkam's (1995) recent article cited ten studies on this. They also conducted a further analysis showing that 95.6% of psychology papers using tests of significance rejected the null hypothesis, while for medical journals it was 85.4%. We briefly mention the results of four additional empirical studies on the incidence of publication of articles rejecting the null hypothesis.
Politics Symposium, 2016
The PollyVote is an evidence-based formula designed to forecast election outcomes, using both wel... more The PollyVote is an evidence-based formula designed to forecast election outcomes, using both well-established methods and innovations. Forecasting error is reduced by combining forecasts within and across different methods, equally weighted. Following this rule, the PollyVote has accurately forecast the outcome of the last three presidential elections by as much as a year in advance of Election Day. Updated twice a week in 2004 (Cuzán, Armstrong, and Jones 2005a ; 2005b ) and in subsequent elections at least once daily, at no time has the PollyVote called the
election for any other than the winner. Moreover, on average across the past six elections, the PollyVote forecast has been more accurate than any of its component methods (Graefe et al. 2014a ; 2014b ). In the sections that follow, we successively elucidate the combination principle used in the PollyVote, summarize the methods incorporated into it, review its performance in forecasting presidential elections, issue a forecast for 2016, and conclude with remarks on the nature of the PollyVote.
Foresight, 2012
Choosing the right person for a given position is a highly complex task, yet experts believe tha... more Choosing the right person for a given position is a highly complex task, yet experts believe that their experience allows them to do this well. Michael Lewis’s 2003 book Moneyball and the recent film based on the book provide a counterpoint, showing that the statistical procedures used by Billy Beane, general manager of professional baseball’s Oakland Athletics, are more effective in predicting job performance than are experts’ judgments.
In this article, Scott Armstrong traces the emergence of the argument in favor of statistical procedures to writings in the 1950s by Paul Meehl and shows how Meehl’s principles, carried forward by Billy Beane, can be applied to improve business performance today.
International Journal of Forecasting, 2012
Soyer and Hogarth's article, " The Illusion of Predictability, " shows that diagnostic statistics... more Soyer and Hogarth's article, " The Illusion of Predictability, " shows that diagnostic statistics that are commonly provided with regression analysis lead to confusion, reduced accuracy, and overconfidence. Even highly competent researchers are subject to these problems. This overview examines the Soyer-Hogarth findings in light of prior research on illusions associated with regression analysis. It also summarizes solutions that have been proposed over the past century. These solutions would enhance the value of regression analysis.
Political Science Research and Methods, 2013
The Big-Issue Model predicts election outcomes based on voters’ perceptions of candidates’ abilit... more The Big-Issue Model predicts election outcomes based on voters’ perceptions of candidates’ ability to handle the most important issue. It provided accurate forecasts of the 2012 US presidential election. The results demonstrate the model’s usefulness if one issue clearly dominates the campaign, such as the state of the economy in the 2012 election.
It is also particularly valuable if economic fundamentals disagree, a situation in which forecasts from traditional political economy models suggest high uncertainty. The model provides immediate feedback to political candidates and parties on the success of their campaign and can
advise them on which issues to assign the highest priority.
Journal of Behavioral Decision Making, 2013
We used the index method to predict U.S. presidential election winners based on issues polls. Iss... more We used the index method to predict U.S. presidential election winners based on issues polls. Issues polls ask voters which candidate they expect to do a better job in dealing with the issues facing the country. A simple heuristic, which predicted that the candidate who is rated more favorably on a larger number of issues will win the popular vote, was correct for nine of the ten elections from 1972 to 2008. We then used simple linear regression to relate the incumbent's relative ratings to the actual popular twoparty vote-shares. The resulting model yielded out-of-sample forecasts that were as accurate as forecasts from the Iowa Electronic Markets and established econometric models. The model has implications for political decision-makers as it can help to decide which issues to focus on in campaigns.
Journal of Forecasting, 1984
Uploads
Papers by Scott Armstrong
election for any other than the winner. Moreover, on average across the past six elections, the PollyVote forecast has been more accurate than any of its component methods (Graefe et al. 2014a ; 2014b ). In the sections that follow, we successively elucidate the combination principle used in the PollyVote, summarize the methods incorporated into it, review its performance in forecasting presidential elections, issue a forecast for 2016, and conclude with remarks on the nature of the PollyVote.
In this article, Scott Armstrong traces the emergence of the argument in favor of statistical procedures to writings in the 1950s by Paul Meehl and shows how Meehl’s principles, carried forward by Billy Beane, can be applied to improve business performance today.
It is also particularly valuable if economic fundamentals disagree, a situation in which forecasts from traditional political economy models suggest high uncertainty. The model provides immediate feedback to political candidates and parties on the success of their campaign and can
advise them on which issues to assign the highest priority.