Search

Applying the “Bradley Effect” to COVID-Related Polls in the UK

In my opinion, the UK government has fallen victim to “the Bradley Effect” through its over-reliance on unreliable polls for developing and implementing COVID policy. This reliance has resulted in ultimately unpopular decisions, inconsistent policy actions and fractured messaging. Despite the fact that the policy choices of the government were initially popular, this popularity was ultimately based upon perceived fear which, deteriorated steadily as indeed indicated by polling. The polling, however, as a result of the Bradley Effect has drastically underestimated the rate of this deterioration.


When an election outcome is predicted incorrectly by polls because voters provide answers inconsistent with their actual voting behaviour, it is referred by pollsters as the “Bradley effect”. The Bradley effect harkens back to the 1982 election for California Governor in which Los Angeles Mayor Tom Bradley, despite being ahead in polls by a six percent margin and widely believed to be the political equivalent of a “sure thing,” narrowly lost the election to the Republican candidate George Deukmejian. The explanation provided for the results is that, when polled, a subset of white voters wanted to be seen as progressively supporting a black candidate. When these voters, however, entered the privacy of the voting booth, they proceeded to choose another candidate.


The past two US presidential elections both serve as further examples of the Bradley effect. In both 2016 and 2020, “shy” Trump supporters resulted in notoriously inaccurate pre-election polling. Immediately prior to the 2016 election, Hillary Clinton was predicted by the polling aggregation site Five thirty-eight to be 71.4% likely to win the election with 302 electoral votes. In fact, Donald Trump won with 304 electoral votes. While the polling in the 2020 election did correctly predict the winning candidate, the result was far from the “Blue Wave” predicted by the majority of polls.


Throughout the COVID pandemic in the UK, government policy has been clearly based on polling. Numerous journalists have reported this including Patrick O’Flynn, a journalist at the Telegraph, who stated that:

“The truth is that when it comes to the COVID pandemic both Boris Johnson and Keir are tracking public opinion polls almost obsessively, as well as running their own private surveys and focus groups”


The government’s dependency on polls and public opinion in deciding policy is seen most recently in the widely discussed implementation of a vaccine passport, with Michael Gove directly asking Telegraph readers for their opinion on vaccine passports and whether or not they should be implemented.


The problem with this form of “pollocracy” is that polls do not necessarily represent the reals views of a segment of the populace nor do they accurately gauge how the population will respond in reality.


According to The Independent, 93% of people in a March 2020 You-Gov commissioned poll said they supported locking down. Yet, in May 2020 the University of York released a survey stating that 31% of those polled admitted to either “breaking or bending the rules”, meaning that at least 24% of people that polled in favour of locking down in March chose to break COVID rules not much later on. Furthermore, in the month of April 2020 alone the police received over 194,000 calls from people reporting their neighbours for COVID rule violations. We also shouldn’t forget events such as the infamous Bournemouth beach debacle, which saw thousands of people heading to the beach in the summer 2020 despite government pleas to avoid such gatherings..


As evidenced, attempts by Boris Johnson to use polls to create popular policies that people both willingly support and actually follow have clearly failed. In fact, as lockdowns continued, symptoms of the Bradley Effect became more pronounced. According to a YouGov taken at the start of the second lockdown, 72 % of poll participants backed the PM’s plan for the lockdown, representing a relatively modest decline from the 96% support for the first lock-down, as indicated by the above mentioned Independent-Yougov poll.


Based upon the actions of the general population during each lockdown, however, there are clear indications that the decline in support may have been dramatically underestimated. According to an article by the BBC, road traffic fell in the first lockdown to 63% below average pre-COVID use. During the second one, however, road traffic was just 21% below average pre-COVID levels, representing a far greater differential than the 24% difference in polled support would have indicated. Similar percentage discrepancies can also be seen in both the use of public transport and general pedestrian traffic between the two lockdowns.


While the pandemic and its associated polices may be nearing an end, there are some clear lessons to be learnt by the Conservative Party from their overreliance on polls. Firstly, following polls and giving the public what they think they want, is not a successful means of ensuring public support for policy actions. Such overreliance can result in unintended and detrimental consequences. In fact, throughout the pandemic, trust in government and politicians fell dramatically . Secondly, as shown in this piece, it is clear that polls are often inaccurate means of gauging both behaviour in response to policy and strength of policy support. One can only hope that Boris Johnson has learnt his lesson and will finally realise that polls are not the way to giving the people the intended “bread and circuses”, nor will they necessarily keep him in power.


Written by Evan Robert Miller


Evan Robert Miller is a columnist at DecipherGrey.