On Beginnings: Part 23

This essay (serialized here across 24 separate posts) uses words and numbers to discuss the uses of words and numbers — particularly examining evaluations of university degrees that employ statistical data to substantiate competing claims. Statistical analyses are crudely introduced as the mode du jour of popular logic, but any ratiocinative technique could likely be inserted in this re-fillable space and applied to create and defend categories of meaning with or without quantitative support. Questions posed across the series include: Is the data informing or affirming what we believe? What are the implications of granting this approach broader authority? The author, Melanie Williams, graduated from UA in 2006, with a B.A. in Anthropology and Religious Studies.

 

“[I]rrational behavior in the markets may result precisely because individuals are responding rationally to their incentives.”  The Signal and the Noise, p.357.

Surely the incentives for generating $tati$tical analy$e$ are the same incentives I have to keep sweeping up crayons.  Molly Ball says it better:

“[P]ollsters get paid by the poll, ad makers by the ad, phone-calling firms by the call, direct-mailers by the piece. They all have an incentive to promote their services, whether or not doing so helps the campaign win—and they face few consequences if it doesn’t.”

Nate Silver’s success depends upon him nailing his forecasts.  Other types of analyses may be funded whether they prove correct or no.  We can assume, given the uncertainty into which these predictions are cast, they aren’t as concerned with whether or not the statistics of an unknown future will vindicate them, as they are with the more immediate demands of saying something interesting, selling more copy, meeting a deadline, responding to a critique – i.e., getting about the business of circumlocution that constitutes the ever-changing public discourse in which we participate and by so doing, authorize.  I can’t debunk them – they might be right.  Like Fukuyama, they’ve chosen premises that aren’t falsifiable.  But as producers/consumers of this information, I think we could afford to be a bit more skeptical about the weight we give a statistical projection of a job market two, three, or five years into the futureI also just want to point out that we’re doing it – that inferences we represent as meaningful are simultaneously explanations of mathematical correlations and assertions of mathematical correlations.  And if you’re like me, it’s tempting to let these assertions inform the decisions we make when we ponder the costs and benefits of choosing an educational or career path, thanks to our procrustean tendencies to apply blanket abstractions to individuals.  But if I can’t assume I know anything about the creative processes of corporate CEOs based on my own stereotypes, maybe I likewise shouldn’t hold myself to projections of an unknown future based on deductions retrospectively drawn from a range of data within an arbitrary set of criteria. In the end, I am being imagined, too – just as the notion of a past is an imaginative exercise, and the assertion of a future based upon it is an extension of that conceit.  These will likely always be contested, unless Fukuyama’s History plays out to its natural End – or we could say:  someone else’s Beginning, Middle, or Tuesday.  I’m ready to believe in the possibility of anything, now that we’ve invented Furbies.  When faced with the showdown between “truth in the data” vs. “common sense,” I’ll put my money on the even chance that we’re making it up as we go along.  For “what is now proved was once only imagined,” so sayeth my favorite 18th century Englishman.  I’m with you, liberal arts major Newt Gingrich.  Let’s colonize this moon.  I bet I could get hosed there off two beers.

Part 24 coming today at noon…

On Beginnings: Part 13

This essay (serialized here across 24 separate posts) uses words and numbers to discuss the uses of words and numbers — particularly examining evaluations of university degrees that employ statistical data to substantiate competing claims. Statistical analyses are crudely introduced as the mode du jour of popular logic, but any ratiocinative technique could likely be inserted in this re-fillable space and applied to create and defend categories of meaning with or without quantitative support. Questions posed across the series include: Is the data informing or affirming what we believe? What are the implications of granting this approach broader authority? The author, Melanie Williams, graduated from UA in 2006, with a B.A. in Anthropology and Religious Studies.

 

Nate Silver makes it clear that his preference for Bayes’ theorem is a conscientious nod to the role of uncertainty in any depiction of the future.  His attempts to mitigate uncertainty, we can imagine, are addressed in his model; his accounting for it is evident in the probabilities in which he lays out his forecasts, acknowledging some estimate of chance either way of getting it right or getting it wrong – if we could say, in a sense, that Nate Silver could be “wrong.”  Because each candidate’s odds are expressed in probabilities, Silver calculates his misses into his projections.  You could say when an election result falls within the minority of his probability set, it’s a playing out of what he had always acknowledged was a possibility – which is just about the best form of hedging anyone might devise.  And yet, his renown turns on his getting it right where others were mistaken – using polling data available to all of the pundits attempting to forecast the election results.  So what gives?  Why is it so difficult to find the “signal” within the “noise?” Continue reading

On Beginnings: Part 12

This essay (serialized here across 24 separate posts) uses words and numbers to discuss the uses of words and numbers — particularly examining evaluations of university degrees that employ statistical data to substantiate competing claims. Statistical analyses are crudely introduced as the mode du jour of popular logic, but any ratiocinative technique could likely be inserted in this re-fillable space and applied to create and defend categories of meaning with or without quantitative support. Questions posed across the series include: Is the data informing or affirming what we believe? What are the implications of granting this approach broader authority? The author, Melanie Williams, graduated from UA in 2006, with a B.A. in Anthropology and Religious Studies.

 

Nate Silver’s methodology, to the extent he describes it, seems to be stitching together a few different types of analyses.  His aims and limits are fairly well defined:  Forecast the results of elections in various states in which a nomination process whittles the initial field of candidates down to a handful.  For each open seat, one candidate will triumph.  Silver then sets about combining political polling data, from which he derives the probabilities of his forecasts, the way you would bet on Natter Phineas to win at 6:1.  We could stop there, and chalk his success up to the inviolable laws of math.  Or we could ask a few more questions.  After all, elections don’t offer a clean analogy to sporting events, or poker, or red cars vs. blue cars.  How often are elections contests between equals?  Or contests, for that matter, at all?  I voted last week.  Of the near-dozen positions on the ballot, two were contested.  A closer examination of the 2008 senatorial election results suggests there were a great many seats in the U.S. Senate in which the incumbent was not challenged aggressively, or at all.  I don’t mean to diminish Nate Silver’s achievements.  I just want to point out that many candidates have ostensible advantages in our elections, making any survey of individual odds less clear cut than it might appear at the outset.  Nate Silver’s data analyses may work, then, because he doesn’t solely rely on them.  There are other sources of data, after all, for deriving an estimate of a candidate’s ephemeral popularity.  When polling data does not inspire an adequate level of confidence, or when the results yield no statistically significant edge to any party, you might turn to the effects of factors that may or may not be reflected in the polling data:  the exigencies of re-districting lines, political alliances and legacies, successful state and federal social or economic programs, popular scandals, infamous gaffes, (it’s so hard to choose), or any other histrionic hoo-ha of our political stage that invites the intrepid to nestle up to the glow of a wider world’s bad reality tv.

What happened this year?

Part 13 coming tomorrow morning…

On Beginnings: Part 11

This essay (serialized here across 24 separate posts) uses words and numbers to discuss the uses of words and numbers — particularly examining evaluations of university degrees that employ statistical data to substantiate competing claims. Statistical analyses are crudely introduced as the mode du jour of popular logic, but any ratiocinative technique could likely be inserted in this re-fillable space and applied to create and defend categories of meaning with or without quantitative support. Questions posed across the series include: Is the data informing or affirming what we believe? What are the implications of granting this approach broader authority? The author, Melanie Williams, graduated from UA in 2006, with a B.A. in Anthropology and Religious Studies.

 

Nate Silver’s B.A. in Economics landed him his own crappy job, which led him to cultivate an interest in baseball statistics that preceded his fortuitous entrée into political forecasting.  Silver’s early influences included the pioneers of sabermetrics, so I dug up an edition of the New Baseball Historical Abstract of liberal arts major Bill James.  Maybe Bill can speak to his own motivations for collecting the baseball statistics he helps innovate:

“Baseball statistics are simplifications of much more complex realities.  It may be unnecessary to say this because, of course, all human understanding is based on simplifications of more complex realities….  Baseball statistics are interesting not because they answer questions for us, but because they may be used to study issues.  The value of baseball statistics in identifying the greatest players is not that they answer all of the questions involved, but that they provide definitive answers to some of the questions involved, which enables us to focus on others.” Continue reading

On Beginnings: Part 6

This essay (serialized here across 24 separate posts) uses words and numbers to discuss the uses of words and numbers — particularly examining evaluations of university degrees that employ statistical data to substantiate competing claims. Statistical analyses are crudely introduced as the mode du jour of popular logic, but any ratiocinative technique could likely be inserted in this re-fillable space and applied to create and defend categories of meaning with or without quantitative support. Questions posed across the series include: Is the data informing or affirming what we believe? What are the implications of granting this approach broader authority? The author, Melanie Williams, graduated from UA in 2006, with a B.A. in Anthropology and Religious Studies.

 

If you’re still here, forgive me.  I don’t claim to grasp nor be in any position to explain the finer points of calculating probability, which is beyond my purview (if you have further interest, allow me to suggest any number of excellent and more expert books on the topic.)  I only mention Bayesian priors in reference to Nate Silver’s methods in order to point out that he uses a method.  Silver’s predictions, however you classify them, use a model to process data selected by him to arrive at a conclusion – a conclusion that is the result of his operation upon what he has chosen to pay attention to.  Nate Silver, in short, is using statistical data to calculate probabilities.  The forecasts he derives from those calculations we can call a variety of statistical inference.  Since his Bayesian approach relies on probabilities, it may prove less helpful in systems of increasing uncertainty and complexity, when implications of given variables are not limited to known sets – a courtroom being an example of such a complex (social) setting, in which statistical data may conveniently suit a purpose more than “unveil a truth.”  And yet, even within the bounds he has drawn for his analyses, Silver’s success is predicated on his competitors’ “getting it wrong,” using the same data sets with the same spectrum of outcomes.  If, as Silver suggests, there is a “signal” hidden in the “noise” of statistical data (terms lifted from the lingo of electrical engineers), why can’t everyone concoct a model to predict the winners of political elections?  Statistical data that feed widely varying conclusions suggest, to me, that such inferences have more in common with the rhetorical techniques used in a courtroom than with calculating how many blue cars may drive through my town on any given day. Continue reading

On Beginnings: Part 4

This essay (serialized here across 24 separate posts) uses words and numbers to discuss the uses of words and numbers — particularly examining evaluations of university degrees that employ statistical data to substantiate competing claims. Statistical analyses are crudely introduced as the mode du jour of popular logic, but any ratiocinative technique could likely be inserted in this re-fillable space and applied to create and defend categories of meaning with or without quantitative support. Questions posed across the series include: Is the data informing or affirming what we believe? What are the implications of granting this approach broader authority? The author, Melanie Williams, graduated from UA in 2006, with a B.A. in Anthropology and Religious Studies.

 

Nate Silver’s political forecasting methods employ a permutation of conditional probability known amongst Number Munchers as Bayesian inference.  In the 18th century, Thomas Bayes authored an unpublished essay offering a method for advancing inquiries in the face of undefined variables, using probabilities. Silver’s application of Bayes’ theorem falls under the subjectivist umbrella of its use, in which an experiential hunch is assigned some initial, arbitrary degree of likelihood, called the prior probability.  We could just call the prior a hypothesis – one with specific odds of being borne out by observation.  Observations themselves then “condition” the prior, either supporting or refuting the hypothesis according to the value each bit of data brings to the equation, expressed in probabilities.  These conditional probabilities of data, more often called “evidence,” are calculated somewhat circularly, given some combination of objective and subjective measures of the likelihood of the hypothesis predicting the evidence, and the likelihood of the evidence implying the hypothesis.  These conditional probabilities are then compounded with the prior probability to yield a new likelihood of the hypothesis, accounting for the evidence that may have implied the hypothesis that might have predicted the evidence.  That’s right.  This process can continue in the same helical form until any adequate series of adjusted probabilities (called posteriors) lead one to accept, redefine, or discard the prior.  Bayes’ theorem has many specialized derivatives, but is best known for allowing the user to estimate a range of likelihoods with meager information, because it assigns both hypothesis and conditional data those interdependent, never-100% probabilities, from which the composite probability of the outcome is calculated – the idea being that each trial will either contradict or confirm your prior, allowing you to update and refine its likelihood as your trials progress.  Contrast this to the frequentist approach championed by Ronald Fisher in the 1920’s, in which a theory with no initial value slowly acquires credibility over a series of rigorous trials, the results of which must be filtered through their p-values:  a standard measure of the odds that any deviation from the expected values of those results would be the product of chance alone.  Each method seems to have its advantages – the frequentist one claiming more objectivity and a process of validation that relies more on exhaustive trials and peer review; the Bayesian one allowing the user to pursue a line of inquiry well into the realm of unknowns, using a pliant estimate of likelihoods to alternately spur or yoke a premise.  Bayesian inference seems to offer more flexibility in this regard, as an informal, ad hoc tool for assessing chance; it also has the benefit of smelling like what most of us would consider plain vanilla reason, wherein our existing beliefs and values can be tweaked in light of new circumstances while continuing to inform the decisions we make as we go along.  As a forecasting tool, Bayesian inference would appear to have its limits – most notably, when a prior probability is not available, in the face of unprecedented events or complex circumstances.  I would trust Bayesian priors, for instance, if I were trying to remember where I parked my bike last night.  I would not trust Bayesian inference to predict what will happen if I press this button.

Part 5 coming tomorrow morning…

On Beginnings: Part 2

This essay (serialized here across 24 separate posts) uses words and numbers to discuss the uses of words and numbers — particularly examining evaluations of university degrees that employ statistical data to substantiate competing claims. Statistical analyses are crudely introduced as the mode du jour of popular logic, but any ratiocinative technique could likely be inserted in this re-fillable space and applied to create and defend categories of meaning with or without quantitative support. Questions posed across the series include: Is the data informing or affirming what we believe? What are the implications of granting this approach broader authority? The author, Melanie Williams, graduated from UA in 2006, with a B.A. in Anthropology and Religious Studies.

 

You have probably heard of Nate Silver.  His forecasts set political and social media atwitter in 2008 when he correctly estimated the Electoral College votes of the U.S. presidential election in 49 out of 50 states, as well as the District of Columbia; he also predicted the outcomes of 35 out of 35 senatorial elections that year.  I initially attributed Silver’s success to an extraordinary bit of luck – after all, somewhere in the prognostic clamor surrounding any national election, someone is bound to come within the mark of the final results, the way every eleventh snowball I launch makes contact with my co-worker.  Then, in 2012, Nate Silver correctly forecast the results of the U.S. presidential election in all 50 states and the District of Columbia, and picked the winners of 31 of 33 senatorial elections.  This is how I was suckered into reading his book.  I was hoping to glean some insights on the statistical data analyses that inform Silver’s forecasting techniques – perhaps not enough to re-present in essay form, but at least a dirty trick or two I could take to the races.  The Signal and the Noise, as it turns out, is scant on the specifics of Silver’s particular methods, but he does describe some general premises of statistical inference he applies in making his forecasts, with none of the usual esoteric symbols that lull laypeople like me into a mental parade Religious Studies graduate Jennifer Goodman once succinctly described:  “monkey…banana…bicycle…ball.”  I don’t agree with many of Nate Silver’s presuppositions nor the conclusions he draws from them, but I think the methods he shares for approaching statistical inference are worth exploring, not only to understand his uncanny success, but to confront the growing popular relevance of statistics as we march (onward, forward) into the titillating age of Big Data, amassing ever-larger caches of minutiae to formulate anything from optimized marketing strategies to post-graduate career prospects.  If I can’t use statistics to make my fortune, could I at least use them to elucidate the arc of my “career?”  Or yours?  For an eloquent and thorough rendition of statistical applications, let me refer you to Silver’s book.  Read it and come back, I’ll wait here for you.  If you’re satisfied with the train wreck version, climb aboard.  If you’re admiring a pony in the mental parade, just close your eyes.  I’ll wake you at the end with a rousing musical number.

 In order to talk about statistics we should also talk about probability.  Please stay.  I love you.

Part 3 coming tomorrow morning…