The importance of contextual data

In primary school, as I embarked on the long road that eventually led to my career in chemistry, one of the first things I recall being taught was about the importance of good scientific practice and avoiding bias. As I progressed through school, university and into industry I have become increasingly aware that completely removing bias, which seemed simple to me at that early age, is actually quite complicated.

Recently I have been thinking more deeply about bias in science and the wide effect this can have on industry and society. For example, during the current coronavirus situation, it was widely reported by the media that people who had received the Astra-Zeneca vaccine were developing blood clots.  Often the claims were supported by evidence such as ‘X in 100,000 people who had the AstraZeneca vaccine developed a blot clot’. This caused the public to draw conclusions about the safety of the vaccine. Although these conclusions were drawn from data, this data did not represent the whole picture, this is something we might refer to as sampling bias. Once data was published about the rate of blood clot development in unvaccinated people, people who had covid and people who took other commonly used medicines, this gave a different picture and a different conclusion was drawn by many people. This demonstrates the importance of contextual data.

Another situation has arisen pertaining to the adhesives industry in particular, which again has caused me to think about contextual data and bias. Articles have been published about adhesively bonded brick slips falling from height. These describe several case studies and some are asking for people to provide information about other cases. Again, although there is scientific data within the articles, I find myself drawing similarities between this and the above example of data on coronavirus vaccines. I worry that unless there is contextual data about failures of mechanically fixed brick slips, or any other materials at height, available alongside such articles that any conclusions drawn will be based on a skewed subset of data. The publishers actively asking for people to supply additional data already appear to have drawn the hypothesis that adhesively bonded brick slips are prone to failure. I believe this also leaves the research vulnerable to what is known as confirmation bias, are they are looking for data subjectively and analysing it in such a way as to support a hypothesis they already believe to be true.

As part of a company that develops and manufactures adhesives, I am acutely aware of the damage such articles could have on the industry. This has even further ingrained in my mind the importance of all scientific articles being vigorously assessed to ensure that what is published is fully representative of the entire picture and that bias is limited to a minimum so that conclusions drawn will be astute and will benefit societal safety and industrial prosperity. With that being said, we would call for the safety of all materials above 2 storey’s in height that have the potential fail to be impartially assessed to the same testing standards.