The Think Smart Series
Teach Yourself to Think Better!
Success is shaped by luck, resources, skills – and the way we think. By thinking better we can perform better.
Our ‘Think Smart’ series offers a set of stepping stones to understand how the world really works.
Each bite-sized video examines and explains a single aspect of looking more carefully, thinking more clearly and crossing the technical, cultural and social barriers that divide us.
Knowledge is Strength!
Full Video Directory
Each Smart Bite addresses a specific flaw or technique.
- Where to look
- Know your limits!
- Harnessing their power
- Limiting their harms
- When to and when not to
- Everyday Bubbles – examples
- Reinforcing Bubbles
- Group Think
- Mathematics behind bubbles
- How to be aware and ‘break out’ of your own bubbles
- Red Teams & Devil’s Advocates
- Truth Tables and Comparing Hypothesis
- Forecasting & Hindcasting
- Inventing causes
- Understanding the problem
- Data incest
- Shaped by location: where you look shapes what you see
Find the right levers to change the world
'Women Can't Park`
We tend to search, interpret, favour and recall information that confirms our beliefs, and ignore or reject or forget that which counters them.
Why Nothing is Important
'Men are Tall! As a Woman I must be short...`
When a Duck is not a Duck.
Feminists and Footballers
Don't Look There, Look Here!
What is Confirmation Bias
Confirmation Bias is is a type of cognitive bias where previously existing beliefs causes unjust favourism. When researching a conjecture we must be aware of our own confirmation bias – are we actively seeking evidence which supports our existing beleifs or ignoring contradictory evidence? After sampling, we also need to consider the affects of confirmation bias on qyality assurance and concept extraction.
Confirmation Bias: All Videos
Introduction to Confirmation Bias
Confirmation Bias Worked Example
Initial Reaction as a Bias Indicator
How does Confirmation Bias Affect Research?
Biased Source Pools
Confirmation bias over-values sources which agree with prior beliefs. Contradicting sources are over-looked and disregarded. By only considering evidence from a singular viewpoint, you compare your evidence against emptiness, rather than counter-evidence. This mono-tonic information collation fails to include the whole picture; for researchers this skews search results, something that efficient researcher wants to avoid.
Mitigating Confirmation Bias: The 4-Way Method
Suppose we wanted to research the claim ‘Women Can’t Park,’ and found lots of newspaper reports, images, videos and scholary articles which support the controversal statement. Despite collating a wide variety of supporting evidence, we cannot deem the statement true. So far we have only considered evidence of women parking badly. These should be compared against examples of women parking well, and men parking badly. Then both should be compared against the background population. In this case, with very small minorities aside, men.
Only once we collate evidence from all four quartiles can we conclude that women are worse than parking than men or not. Of course, this doesn’t mean that all women are worse than all men (or visa versa), it just indicates the existence of a trend (more on that in a future video!) Although simple, this 4-way approach is a quick tool to combat our own personal, confirmation bias.
Initial Reaction to Claim
Take the claim ‘Women can’t park’ and imagine two hypothetical individuals John and Jannet with opposing beliefs. Jannet disagrees strongly, taking the statement as a personal attack, whereas John agrees with the statement and doesn’t take offence. This initial response is a large indication of confirmation bias. (Watch our initial reactions video for more detail.)
Stage 1: Reconnaissance
Think back to our two individuals, Jannet and John, and how their perception of information surrounding the claim might vary. For example, suppose John had a personal encounter with a woman hitting his car whilst parking. This might have tainted his perception of all woman drivers, and he can use his own experience as supportive evidence. Unless John has told Jannet directly about the specific incident, Jannet would not know of John’s experience. Similar, suppose Jannet reads an article which provides contradictory statistics. Since the magazine targets a female demographic, John does not know of the article’s existence
Stage 2: Accessibility
Next we question their respective accessibility. Suppose Jannet’s magazine was only available through an online subscription. Even if John knew about it’s existence, the magazine would be accessible to him without payment. Similarly, Jannet might not have access to all of John’s contacts, meaning their testimonials are accessible to her.
Stage 3: Survey Size
Now we consider their respective expectations of sufficiency. Suppose both Jannet and John employed the 4-way approach and achieved the following table. John is more likely to accept this collection as sufficient since it supports his beliefs, wheres Jannet is more likely to reject the table on the basis that, since it doesn’t align with her own beliefs, she does not deem it a representative selection of sources, and feels a need to collate more evidence.
Stage 4: Representation
With all this considered, we would expect Jannet and John to select differing source pools when collating evidence.
Stage 5: Individual Scrutiny
Prior conceptions do not only affect which sources are selected, but also which sources are accepted. We would expect Jannet to view sources which affirm the claim with more scrutiny than John as they oppose her own beliefs. This could cause Jannet to reject a source which John accepts, thus making their respective source pools even more different.
With every aspect of evidence collation and source scrutiny directly affected by confirmation bias, we can see how despite being presented with the exact same claim, two individuals may be lead to opposing conclusions. This really highlights the importance of both being aware of, and tackling the effects of confirmation bias when conducting your own research.