Events pollsters watch

There were three interesting events for pollsters this week: (a) the final act of the US election, (b) the 2020 conference of the Asia chapter of the World Association for Public Opinion Research (WAPOR), and (c) the release of the report about the failure of polling in the 2019 federal election in Australia.

The United States election. Although the counting was unusually long, due to the mail-in voting, the full outcome seems to be as expected from the polls. On 11/12/20, 9:08 p.m. EST (10:08 a.m. Manila time, 11/13/20) CNN reported that Joe Biden, Democrat, will get between 51 and 52 percent of the popular vote. This compares well with the 51.3 percent call of the final pre-election polls of the popular vote for Biden, as averaged on 11/2/20 by WAPOR past president Claire Durand (see my “The tight US election,” 11/7/20).

In the United States’ quirky electoral college system, 51 percent is not a slim win. With only 52.9 percent of the popular vote in 2008, and 51.1 percent of it in 2012, Barack Obama easily won both times. Donald Trump clearly won in 2016 despite getting only 46.1 percent of the popular vote, versus Hillary Clinton’s 48.2 percent. This year, Biden’s win seems convincing, and Trump looks simply like a sore loser.

WAPOR Asia 2020. The WAPOR Asia 2020 conference was held last Monday-Tuesday by zoom at the International Islamic University Malaysia in Kuala Lumpur, opened by no less than Malaysian Prime Minister Muhyiddin Mohd Yassin.

Professor Durand had a plenary session on “Analysis of polling during the presidential election” that was quite well attended. Only a few papers were on elections, since pollsters are focusing more on the pandemic and other topics. SWS had pieces on: “The effect of joblessness on hunger experience among Filipinos in times of pandemic,” by Krisia S. Santos; “How often is essential? Exploring the frequency of Filipinos leaving their house for essential activities amidst the COVID-19 pandemic,” by Dankarl Leon DL. Magpayo; “The tolerance of Asians for certain neighbors,” by Gerardo A. Sandoval and Linda Luz B. Guerrero; “How dependent are Asians on their government?” by Leo Rando S. Laroza and Linda Luz B. Guerrero”; and “The refusal of Filipinos to pivot towards China,” by yours truly. All presentations will be at https://waporasia2020.com/.

The Australian 2019 polling failure. Last Wednesday, the Association of Market and Social Research Organisations (AMSRO) in Australia released the final report of its inquiry into the performance of the opinion polls at the 2019 Australian federal election. It aimed to determine why all the published polls incorrectly called the outcome of the May 2019 election, and how polling might be improved in the future.

From 2007 to 2016, Australian pollsters had a 96 percent success rate: 25 of 26 polls were correct across four federal elections. For 2019, however, the report concluded that there was a “polling failure” rather than simply a “polling miss,” as the polling error was statistically significant and all in the same direction, i.e., all in favor of the Australian Labor Party, whereas the actual winner was the Liberal National Party. It concluded that the source of the error was in the polls themselves and not the result of a last-minute shift in the preferences of the voters. It said that the polls used inadequately adjusted unrepresentative samples, and were very likely skewed toward over-representing more politically engaged and better-educated voters, thus over-representing the Labor Party.

In all countries, so much is expected of opinion polls, but they get harder to do all the time, due to changes in technology, and in the people themselves who are polled. The solution is not to stop relying on polls, but to do more of them, and to do them better. And all are welcome to give it a try.

——————

Contact mahar.mangahas@sws.org.ph.