We think of information as something to learn from. But we really use it to verify what we already think to be true, particularly for healthcare. We use healthcare information like a transaction: find some content that supports what we believe and use it to justify our beliefs.
It’s called confirmation bias, and it appears in places where there are few checks and balances on the way we think. This would be the internet. And the process just described would be what most people refer to when they say, “I did my own research”.
We can see the effects of this process repeated over and over by reviewing the most widely trafficked healthcare websites. We’ll see a pattern that reveals how most people consume healthcare content online through the ratio of backlinks to search traffic.
A backlink is a link created when one website links to another. It represents a “vote of confidence” from one site to another. Search traffic refers to the visitors who arrive at a website by clicking on the search results that lead to a website.
Websites that have more search traffic views than backlinks are popular sites that present healthcare content through a polarizing opinion on a controversial healthcare topic. Conversely, websites that have more backlinks than search traffic are sites that are heavily referenced, but not actively searched for online. These sites present healthcare content from medical institutions or accredited thought leaders.
Ideally, the two should align. People should search for sites that have the most backlinks because they are deemed the most credible by the Google search engine overlords. But the two seem to move in opposite directions – at least for the most popular healthcare websites – and it belies a curious behavior among those searching for healthcare content.
The way they search for content does not correlate with the quality of content available.
Many websites, like WebMD or Medscape, format their articles into easily digestible articles. They present content in small paragraphs with short sentences that distill medical information into the simplest terms.
But simplifying healthcare usually entails adding a bias. After all, a bias is a heuristic, a shortcut way of thinking. So naturally biases will appear when healthcare content is simplified.
Take drug interactions among opioids as an example. When most people search for these terms, they use simplified language like ‘mixing’ instead of ‘interactions’. But just a slight change of wording leads to a massive shift in the links that appear in the search query.
When using the word ‘mixing’, we see more sites advertising addiction centers. When using the word ‘interaction’, we see more academically oriented sites, even direct links to clinical studies. These are the articles with a high number of backlinks. Interestingly, they are the sites frequently referenced by the more popular sites that appear when using the search term ‘mixing’.
It’s an example where the way people search affects the actual results. Since our words reflect our thoughts, it’s natural to assume people searching online for healthcare content will gravitate toward content that is predisposed to the way they already think.
It’s how search algorithms are designed to work. But it’s not how credibility is garnered for healthcare content. Sure, there’s some correlation between website credibility and what appears in a search query. It’s detailed in the way Google formats its search algorithm. But there’s a stronger correlation between how websites mirror search phrases and overall search traffic.
To its credit, Google has tried to modify its algorithm to reflect the quality of content instead of just buzz-worthy search terms, but there’s a limit to how effective this can be. That limit represents the bias that innately appears in the search algorithms.
Google uses words and word-sequences to gauge how credible a site is. Those sequences are then logically arranged in a way that conveys information. And Google makes judgment calls based on those arrangements.
But every judgment, no matter how sophisticated, comes with some variability. This is regardless of the quality of the underlying search algorithm. The variability in this case is the differences in the search rankings between the most popular and the most credible sites that arise from using various search terms.
Behavioral economists call this noise, or unwanted variability that comes in any system that makes judgments. Noise is the reason Google’s search algorithms prefer websites that match how people search for healthcare content instead of websites that convey highly credible, objective information about that content.
For each individual search, there may be little noise or variability in the search results. But over time, that variability compounds to where wild fluctuations in search results create massive disparities between links that are popular and those that are credible.
This variability explains why search traffic and backlinks do not match up for high-volume healthcare websites. But more than anything else, it explains why misinformation is so pervasive online. We encourage it, search over search. People searching for healthcare content aren’t focused on the most credible sites. They’re focused on the best ranking sites, which are determined by how they are searching.
They use terms laden with bias, so naturally the results will carry that bias into what appears in the query. It all starts with a slight shift in the way we think that manifests in the way we search for healthcare content. This introduces an initially subtle but growing variability that turns healthcare content from a source of information into a veritable transaction – and turns facts into preconceived value judgments.
