For those who read or listen to the traditional news media, the overwhelmingly negative messaging about the state of the UK and its constituent parts, as well as the world in general, is striking. This reflects the adage that ‘bad news or unusual events sell papers’; whereas the opposite is true for everyday events, that is, they do not result in the same level of readership and therefore, news sales. However, this apparent negative bias or emphasis on unusual events can cause despondency and extinguish hope and optimism, even while there are good things happening in the world. For example, there have been remarkable improvements in life expectancies worldwide prior to the COVID-19 pandemic disruption to all health systems, and global infant mortality and malaria continue to decline (World Health Organization, 2023).
In his award winning book ‘Factfulness’, the Swedish epidemiologist Hans Rosling (2018) has argued that our ignorance of facts is not due solely to portrayals in the media but due to a lack of critical thinking and skepticism of the evidence, and a reliance on intuition which yields an overdramatic worldview. In other words, our evolutionary development has encouraged our brains to come to swift conclusions without much thinking to avoid immediate dangers, as, in the past, only hearsay and dramatic stories were the source of news and useful information. Every day we are bombarded with lots of information which is beyond our mental capacity, so it is functional to filter what is absorbed.
Rosling (2018) explained that ‘popular’ misconceptions arise due to:
- The gap instinct: desire for a simple division or binary thinking, such as rich and/versus poor, which neglects the biggest group, namely, middle incomes
- The negativity instinct: proclivity to notice the bad rather than the good (with poor recall of the past – ‘how things really were’), selective reporting in the media and a desire not to appear heartless by acknowledging that while things are better, some continue to be bad
- The straight line instinct: assumption that data lines continue straight, but in reality, trends over time are often curves of different shapes
- The fear instinct: fear of physical harm, captivity (loss of freedom/control), and contamination, but in reality this is not the same as current dangers or risks in our lives
- The size instinct: tendency to get things out of proportion due to lack of comparative data. For example, a single figure is less meaningful than rates, especially when comparing different sized groups. When confronted by a long list, it is best to focus on the few largest (80/20 rule)
- The generalisation instinct: tendency to categorise, but it can be misleading and lead to stereotyping. This instinct can be countered by questioning your categories and looking for differences within and across groups and similarities across groups
- The destiny instinct: the belief that innate characteristics determine destinies of people, nations and cultures, and things are unchanging and unchangeable, thereby reinforcing false generalisations and binary thinking. However, in reality, change can be slow and knowledge can go out of date quickly or slowly depending on the topic. Talking to an earlier generation is revealing
- The single perspective instinct: dependence on a single source rather than on multiple and competing perspectives relevant to understanding a complex world
- The blame instinct: desire to identify a clear and simple cause or guilty party often confirming existing beliefs
- The urgency instinct: acting quickly often means deciding quickly and thinking less critically with insufficient information/data when a more methodical analysis based on more data results in better decision making.
‘Our evolutionary development has encouraged our brains to come to swift conclusions without much thinking to avoid immediate dangers, as, in the past, only hearsay and dramatic stories were the source of news and useful information.’
Rosling argued for a more balanced approach to understanding the context of reported events, namely, control of the 10 ‘instincts’, which he identified alongside skepticism of the presented evidence and critical thinking based on multiple perspectives with a willingness to change your opinion. He argued that people should believe that it is possible that the world in which we live can improve. He described himself as a possibilist in contrast to being a pessimist or optimist. Rosling's work continues as the Gapminder Foundation https://www.gapminder.org/ which you may find a useful resource.
In contrast to the news media, academic journals may have a positive bias because they rarely publish research findings which are negative or of no consequence, although future researchers may learn a great deal from ‘failed’ randomised controlled trials or other ‘failed’ research. Further journals may prioritise novelty and be reluctant to publish replication studies. van Aert et al (2019) have argued that understanding the extent of publication bias is important because meta-analyses can suggest overestimated effects or suggest the presence of non-existing effects. Their examination of 83 meta-analyses published in Psychological Bulletin and 499 systematic reviews from the Cochrane Database of Systematic Reviews focused on meta-analyses reporting homogenous subsets of primary studies. It is not possible to explore bias in heterogeneous subsets of primary studies using statistical tests. Publication bias tests did not find bias in the homogeneous subsets. While they found overestimation was minimal, it was statistically significant and confirmed the presence of mild publication bias in both psychology and medicine publications. However, van Aert et al (2019) acknowledged that their study may have had several limitations. Publication bias may be less likely if the meta-analysis was focused on a secondary outcome rather than a primary outcome, where the result may have influenced whether it was published in the first place. Additionally, some of the meta-analyses in this study included unpublished studies that may have decreased the detectability and severity of publication bias in the homogenous subsets. The primary studies may have been compromised by questionable research practices, which may have biased the effect size estimates, and they were included without the scrutiny of a journal review process. Most meta-analyses report medium or high heterogenous subsets; therefore, this study's findings regarding publication bias cannot be generalised to meta-analyses with heterogenous subsets of primary studies.
‘In contrast to the news media, academic journals may have a positive bias because they rarely publish research findings which are negative or of no consequence, although future researchers may learn a great deal from ‘failed’ randomised controlled trials or other ‘failed’ research.’
Another potential bias is confirmation bias, which has been acknowledged in the psychological literature for years (Nickerson, 1998). It refers to the systematic seeking or interpreting of evidence consistent with existing beliefs and expectations, or to support a hypothesis so that impartiality (rationality) is suspended contrary to good scientific practice. Indeed, candidates in all public and professional examinations are given numbers to disguise their identities to control for this bias in an educational context. While we all like to think we are careful to take rational decisions on the evidence, in all probability we cannot always seek out all relevant evidence from multiple sources when under time pressure. There is remarkably little research regarding the extent of confirmation bias. Kosnik (2008) tested for confirmation bias using a confidential questionnaire survey of US undergraduates (n=284) regarding a recent tax policy (during the President Bush Jr era) and found that the undergraduates' predispositions predicted their responses regardless of the evidence.
Modgil et al (2021) explored social media-induced polarisation during the COVID-19 pandemic and reported that there was an interplay between confirmation bias and ‘echo chambers’ at the height of the pandemic. With social media increasingly becoming a primary source of information for many people, Fernandes (2023) tested a social learning model using simulation data to explain how confirmation bias influences the development of opinions. Social learning describes how we learn through our networks with information accumulating through time and experience; however, it predisposes us to overvalue the views of friends with similar beliefs. Fernandes (2023) noted that only two types of opinion could be formed and both are biased. Yet, one opinion type was less biased than the other depending upon the state of knowledge, but the size of both biases depended on the ambiguity level and relative magnitude of the state of knowledge and confirmation biases. Thus, when the degree of partisanship is high, those with strong views may exacerbate the misinterpretation of ambiguous data, thereby exerting a disproportionate influence over others and amplifying conflicts in the social network. Different people may vary in the intensity of their confirmation biases, with polar opposite biases being on display. This variability in views can impact how the emergence of group consensus plays out. This study was based upon simulation data but it indicates how real world social networks may work.
Our capacity to assimilate lots of information is tested every day. We need to inevitably filter and take shortcuts; however, we need to be mindful that it is easy to be exposed to limited data, which predisposes us to partial knowledge of the subject matter.