By: Leopold Kong, PhD
The polling crisis: How to tell what people really think
The conflicting polling results for the US presidential elections have been a source of no small confusion for American voters. Skepticism over polling is further justified by recent failures, as in the 2013 provincial elections in British Columbia when the Liberal Party won against expectations, or the Brexit referendum. Two major challenges make polling less accurate, and changes are underway to address these issues.
The first major challenge is obtaining public opinion. In the past, pollsters can simply call people at home, but this is increasingly difficult with the rise of cell phone use. Currently, only 50% of US households have landlines compared to 80% in 2008. Federal regulations require mobile phones be called manually, and people often don’t answer cell phones from an unfamiliar number. People who do answer these numbers might represent a biased population. Despite these limitations, calling cell phones are more accurate than online polls, which are less regulated and could easily be manipulated. Using texting instead of direct calls could also increase response rates.
The second major challenge is predicting who will vote, which is particularly difficult in the US with low voter turnouts of about 45-50%. To predict this, each pollster organization uses a proprietary mix of factors such as voting history and political engagement. “Likely voter modeling is notoriously the secret-sauce aspect of polling,” says Courtney Kennedy, Director of survey research at the Pew Research Center in DC. Furthermore, these models may generate unconscious bias for pollsters to “herd” polling to better reflect predicted expectations. Improvements are underway, including using a probability model versus a discrete yes/no model, and greater transparency in methodology.
With the changing face of demographics and technologies, polling science is evolving to keep pace. (Ramin Skibba, Nature)
Two HPV shots instead of three
Human papilloma virus (HPV) is responsible for about 5% of all cancers in the world, including 70% of throat, neck and oral cancers, and 90% of all anal cancers. Originally, an effective vaccine was approved in 2006 for a three-dose regimen to confer protection. Since then, clinical data reviewed has shown protective efficacy with only two doses in Costa Rica. The Advisory Committee on Immunizations Practices at the Centers for Disease Control and Prevention (CDC) has now recommended two doses of the vaccines for pre-teen boys and girls.
“The pediatricians and other people I talked to said the new recommendation is a game changer with that schedule,” said Kevin Ault, MD, professor of OBGYN at the University of Kansas Hospital. “It’ll make it easier for the doctors, easier for the parents and easier for the kids.”
This recommendation is very timely, and may boost vaccination rates, which have risen very slowly so far. Teen girls getting the vaccine only increased from 60% in 2014 to 62.8% in 2015. Doctors have been timid about promoting the shots with parents, who may not want to have discussions about their children having sex. A lighter vaccination schedule may help. Furthermore, it reduces cost significantly for implementing the vaccine in low and middle-income countries, and thus may greatly aid in curbing the global cancer burden. (Associated Press, STAT)
Have an interesting science policy link? Share it in the comments!