Pedro Magalhães

Margens de Erro

Teorias sobre o fiasco das sondagens em NH

Posted January 10th, 2008 at 10:48 am4 Comments

Para todos os gostos, mas alguma convergência naquelas que têm já algum apoio empírico:

1. What was going on with the New Hampshire polls? Excessiva filtragem dos votantes menos prováveis, ponderações amostrais e o "Bradley effect" (eleitores dizem que votam em candidato negro e depois não o fazem)..

2. Why presidential polls are wrong? Recuperação de última hora de Hillary.

3. Pollsters Review Tactics After Forecast `Fiasco' . Recuperação de última hora de Hillary.

4. Pollsters flummoxed by New Hampshire primary. The New Hampshire contrarians.

5. Ballot Changes Cited in Vote's Discrepancy With Polls. A ordem dos candidatos nos boletins. isto é para levar a sério: Krosnick é uma das pessoas na academia que mais sabe sobre sondagens.

6. New Hampshire: So What Happened? Recuperação de última hora de Hillary e "Bradley effect".

7. Why were the polls wrong in New Hampshire? Mostly all of the above, excelente síntese por Anthony Wells.

E outras aqui, aqui e até uma conspiração aqui. Uma síntese (irónica): How the world will explain Clinton’s win despite final polling showing her way behind Obama

"Didn’t factor in the 24 hours of tears? The fact that New Hampshirites like to make news? That independents turned out for McCain and Clinton as well as for Obama? That without Huckabee as a factor, the McCain-Romney fight was taken more seriously in the end? That all those comments that she was bussing in people from New York and Massachusetts to pad the crowds were the nonsense some always suspected them to be? That Bill Clinton acting out and saying crazy things reminded people that they were once sympathetic to his wife? That her debate performance was mocked by the pundits but loved by the voters? The lingering impression of Billy Shaheen’s pre-Iowa words? The shadowy hand of Michael Whouley? That negative attacks — through the mail on issues such as abortion — work? The appeal of Clinton to 20 somethings? The 'doer versus talker' message? All those prominent women supporters in a state that has a lot of women elected leaders? The Chris Matthews hug? That Obama has limited appeal to blue-collar Democrats in places like Epping? That she took questions in town meetings at the end, which New Hampshire voters really like? That Obama had a Tom Bradley-Doug Wilder problem? More coming…."

by Pedro Magalhães

Mudando de assunto (ou, num certo sentido, não)…

Posted January 9th, 2008 at 4:11 pm4 Comments

Fonte:Ellen Nolte e C. Martin McKee (2008), "Measuring the Health of Nations: Updating an Earlier Analysis", Health Affairs 27: 58-71. Resumo aqui.

"The concept of amenable mortality - that is, deaths from certain causes before age 75 that are potentially preventable with timely and ef-fective health care - was developed in the 1970s to assess the quality and performance of health systems and to track changes over time. (...) In a Commonwealth Fund-supported study comparing preventable deaths in 19 industrialized countries, researchers found that the United States placed last. While the other nations improved dramatically between the two study periods—1997–98 and 2002–03—the U.S. improved only slightly on the measure."

"The largest reductions in amenable mortality were seen in countries with the highest initial levels, including Portugal, Finland, Ireland, and the U.K, but also in some higher-performing countries, like Australia and Italy. In contrast, the U.S. started from a relatively high level of amenable mortality but experienced smaller reductions. "

E a cobertura noticiosa do estudo, aqui.

by Pedro Magalhães

E o outro fiasco…(actualizado)

Posted January 9th, 2008 at 1:45 pm4 Comments

Bad Bet: Why were the political futures markets so wrong about Obama and Clinton?
By Daniel Gross

"So, I've been watching the action in one of the political futures markets this evening, Intrade. And the action in this prediction market has reinforced my opinion that these are less futures markets than immediate-past markets. The price movement tends to respond to conventional wisdom and polling data; it doesn't lead them. Throughout the day and into the early evening, while polls were still open, Democratic investors, mimicking the post-Iowa c.w. and polls, believed Obama was highly likely to be the Democratic nominee. The Obama contract was trading in the lows 70s, meaning investors believed he had a 70 percent chance of being the nominee, while Hillary Clinton contracts were in the 20s. But between 7 p.m. and 8 p.m., as the Concord Monitor began to post early returns showing Hillary Clinton in the lead, the contracts started to move quickly."

Nobody knows anything
By Paul Krugman
"But to be more specific, the prediction markets — which you see, again and again, touted as having some mystical power to aggregate information, know no more than the conventional wisdom. (...) From inevitability to pitiful failure to front-runner again in just a few days. There’s no hint that the market saw either Iowa or New Hampshire coming, or knew anything beyond the bloviations of the talking heads."

E um paper de Setembro de 2007:
Are Political Markets Really Superior to Polls as Election Predictors?
ByRobert Erikson and Christopher Wlezien
"By our tests, the IEM election markets are not better than trial-heat polls for predicting elections. In fact, by a reasonable as opposed to naïve reading of the polls, the polls dominate the markets as an election forecaster. This is true in the sense that a trader in the market can readily profit by 'buying' candidates who, according to informed readings of the polls, are undervalued. Moreover, we find that market prices contain little information of value for forecasting beyond the information already available in the polls."

by Pedro Magalhães


Posted January 9th, 2008 at 11:11 am4 Comments

A vitória de Hillary Clinton em New Hampshire é daqueles acontecimentos que acabam por ser muito positivos para quem faz sondagens, se bem que não pareça. Quando tudo parece correr bem, ninguém se preocupa em investigar o que falhou, mesmo que muito possa ter corrido mal: pode "acertar-se" pelas razões erradas, ou mesmo por acaso. Mas quando há um desafasamento tão óbvio entre as sondagens e os resultados - a média das últimas cinco sondagens dava 38% para Obama e 31% para Clinton, quando os resultados foram 39% para Clinton e 37% para Obama - não há alternativa se não investigar o que correu mal e claro, aprender. Isto dito por quem, felizmente, não teve de fazer sondagens em New Hampshire...

Há um risco, contudo: a da multiplicação de explicações ad hoc e post hoc sobre o que se passou. Ontem de madrugada, já andava o inefável Wolf Blitzer na CNN a explicar que a lágrima ao canto do olho de Clinton"mostrou o seu lado humano" e que isso pode ter influenciado os resultados e yadda, yadda, yadda. A minha sugestão é que não acreditem. Aliás, não acreditem em nenhuma das dezenas de explicações que vão aparecer agora para dar conta do que aconteceu. Vai passar algum tempo e vai ser precisa muita análise até que se perceba qual ou quais dessas razões poderá estado realmente por detrás do falhanço das sondagens.

No mesmo sentido, ver Gary Langer, sobre o New Hampshire's Polling Fiasco:

"There will be a serious, critical look at the final pre-election polls in the Democratic presidential primary in New Hampshire; that is essential. It is simply unprecedented for so many polls to have been so wrong. We need to know why. But we need to know it through careful, empirically based analysis. There will be a lot of claims about what happened - about respondents who reputedly lied, about alleged difficulties polling in biracial contests. That may be so. It also may be a smokescreen - a convenient foil for pollsters who'd rather fault their respondents than own up to other possibilities - such as their own failings in sampling and likely voter modeling."

by Pedro Magalhães

E o antónimo de "momentum" é…

Posted January 7th, 2008 at 5:30 pm4 Comments

by Pedro Magalhães

Momentum em acção

Posted January 7th, 2008 at 3:32 pm4 Comments

by Pedro Magalhães

Obama e os efeitos de Iowa para NH

Posted January 7th, 2008 at 11:20 am4 Comments

Algumas ideias:

1. NH: View from Sunday Morning, no habitual Pollster. Merece ser lido integralmente, mas:

"Obama is certainly rising, the only question is by how much;"

"Firm conclusions are premature for two important reasons. The first involves the issue of weekend interviewing, or more specifically, surveys based on interviews completed entirely on Friday night and Saturday. (...) The second and more important reason to be cautious about this Sunday morning snapshot is that New Hampshire voters are still in the midst of a difficult decision;"

"I cannot point to an academic study to prove this, but most campaign pollsters will tell you that when a candidate is gaining, vote preference is usually the last thing to change. The movement usually shows up first on internal measures."

2. Polls Picking Up an Obama Surge?, no também habitual The Fix:

"In the days since Clinton's third-place finish in Iowa, her campaign has insisted that New Hampshire would pay little attention to what happened in the Hawkeye State. "Voters in New Hampshire are fiercely independent," argued Clinton deputy communications director Phil Singer in the spin room following last night's Democratic debate. 'They make their own decisions [based on] what they see, not what happens in other places.' Tonight's polls seem to contradict that argument."

"Unlike in years past, however, there is so little time between the votes in Iowa and New Hampshire that even a temporary bounce could be enough to carry Obama to victory in the primary, a win that would be another major step forward in his quest for the nomination."

by Pedro Magalhães

Ano 4

Posted January 6th, 2008 at 10:49 am4 Comments

Começa aqui.

by Pedro Magalhães

Um ponto adicional sobre Iowa…

Posted January 4th, 2008 at 3:07 pm4 Comments

...é o seguinte (e que me foi sugerido hoje ao almoço pelo José Tavares): quando olhamos para os resultados das sondagens "à boca das urnas" (por assim dizer), vemos que não há uma grande correlação entre o voto em Obama e características socio-demográficas ou mesmo atitudinais dos eleitores. Obama dominou em quase todas as "auto-identificações ideológicas" e níveis de rendimento. As correlações mais fortes são com a idade e com o estado civil (provavelmente espúria esta última), mas isso, como vimos aqui, é o sinal de uma vantagem na capacidade de mobilização de novos eleitores. Isto não se passa da mesma forma com os candidatos republicanos, cada um deles aparentemente mais representante de determinados segmentos ou nichos do eleitorado. Se a isto somarmos o facto de um negro ter ganho num estado de brancos religiosos e conservadores, temos aqui duas coisas: por um lado, um sintoma da natureza "transversal" da candidatura de Obama; e por outro, um potencial efeito de demonstração que esse facto terá sobre as considerações dos eleitores nas futuras primárias. Há "electability" para dar e vender na candidatura de Obama.

by Pedro Magalhães

Notas sobre Iowa

Posted January 4th, 2008 at 11:34 am4 Comments

1. Obama e os novos eleitores:
"Obama's field operation -- led by Iowa state director Paul Tewes, adviser Steve Hildebrand and caucus director Mitch Stewart-- deserves a MASSIVE amount of credit for the work they did to recruit first-time caucus-goers. The Iowa Democratic Party was estimating turnout at 236,000 -- a huge increase from the 125,000 or so who turned out in 2004 (and an even larger leap over 2000's tiny 59,000 turnout). Tewes and Hildebrand were widely regarded as two of the best in the business, but even those who spoke glowingly of them didn't think they could grow the electorate over 200,000. Well, they did that and much, much more."
E ver isto.

2. A mobilização dos Democratas
"Republicans could well be in deep, deep trouble next November if turnout patterns in tonight's Iowa caucuses are born out across the country. More than 230,000 Democrats turned out, more than double the number of Iowa Republicans who did the same. The energy deficit has been clear for much of the past few years and led to Democrats' gains in the House and Senate in 2006. That chasm appears to be growing wider."

3. Iowa é importante, mas apenas na medida em que influencia New Hampshire:
"New Hampshire is the early state that has the biggest impact. Not Iowa. Iowa has a habit of picking losers. It is easy for media types to forget that because in its most recent outing, 2004, it single-handedly determined the winner. But historically, Iowa does not make much of a big ripple nationwide. The big question: will Iowa move New Hampshire? Obama needs it to. He is in second there right now. We don't have an answer yet - and history provides a mixed message. Sometimes Iowa does move New Hampshire. Sometimes it doesn't."
Ver também aqui.

4. Romney é o candidato dos Republicanos anti-Bush. Não parece que isso lhe seja particularmente favorável.
"Tonight was bad for Romney. Really bad. He lost by a lot. He lost by more than anybody expected. He lost after having led for a year. He lost after a monumental effort. This loss was bigger than Clinton's. He is not his party's frontrunner. He cannot afford to lose a state he tried so hard to win. Worse for Romney - McCain had already surged ahead of him in New Hampshire prior to tonight's loss."

"The press is not interpreting this as 'Clinton ties Obama among Democrats in entrance poll' or 'Mormon Romney finishes strong second in evangelical Iowa.' This matters. Watch how the press continues to interpret these results over the next few days. It is the source of information for persuadable voters in New Hampshire."
- Nos Democratas. Iowa acaba com Edwards. Agora é só entre Obama e Clinton. E se Obama ganha New Hampshire...Nas sondagens ainda está atrás de Clinton. Mas nos mercados electrónicos, que já estão a reagir a Iowa, já está à frente.
- Nos Republicanos, McCain, que parecia perdido, pode voltar à lista dos favoritos com New Hampshire. Romney sofre uma derrota muito grave e acaba se não ganhar New Hampshire, o que é cada vez mais provável. Thompson já acabou. Huckabee já não é uma curiosidade. Giuliani em suspenso.

by Pedro Magalhães