Motivated Reasoning: What It Is And How To Avoid It In Data Analysis
Posted by Thomas Maydon
March 02, 2017
Use versus abuse of statistics can often be characterised by the analytical approach adopted to the problem at hand. In this blog post, which is part of a series on Logical Fallacies to avoid in Data Analysis, I’ll be focusing on defining the motivated reasoning logical fallacy and how to avoid it in data analysis.
What is Motivated Reasoning?
If statistics are being produced or selected to prove a point – we call this motivated reasoning (regardless whether the point is true or not). Motivated reasoning is the process of finding evidence to support a pre-held belief.
The video below goes into some depth about this cognitive bias and gives some good examples.
Motivated reasoning is wide-spread in society. We see it regularly in politics where supporters of one party will dismiss any information that contradicts their belief (a phenomenon to help them avoid cognitive dissonance – i.e. the holding of two contradictory ideas).
Similarly in law, advocates and lawyers for both the prosecution and the defence will be utilising motivated reasoning to prove guilt or innocence/doubt respectively. The judge or jury, conversely, have to use deductive reasoning to build the evidence up to one of these two conclusions. The challenge always is how they adequately and fairly assess the evidence that had been presented by the advocates/lawyers with the intention by either the prosecution or defence and how do they as humans eliminate any motivational bias in their own assessment of the evidence.
Businesses are regularly in a process of motivational reasoning. Holding of opinions and beliefs is part of the human cognitive process and as such data scientists may be stuck with conducting a piece of diagnostic analysis where they themselves have a preconceived idea of the result they might find.
How to avoid Motivated Reasoning in Data Analysis
Be fully aware that biases may creep into your analysis (metacognition: “think about thinking”)
Before you begin your analysis try and do the following:
Note what metrics/analysis you wish to calculate/do and why
Note what you’ll need to find to reach a conclusion
Ask yourself whether the analysis will result in a fair assessment
Ask yourself “If I had an opposing view would I conduct the same analysis?”
Is there anywhere else I can look that will help me to the truth? If so can I predict what I will see?
After you have conducted your analysis answer the following:
Am I presenting a full version of the truth?
Prevent any motivated reasoning by conducting blinded analysis.
If possible, “blind” your data-sets, labels or trials ensuring that the subjects, those conducting the trials and, if possible, your analysts can conduct the research without bias.
Deductive reasoning is the process of reasoning from one or more statements to a logical conclusion.
Although deductive reasoning is a cleaner form of analysis, errors can also slip in here. The human tendency to avoid cognitive dissonance (the holding of two conflicting ideas) means that it is human tendency to only include data and analysis that fits a clear narrative and conclusion. Awareness of these cognitive biases is essential in ensuring the presentation of a fair narrative.
Some practices linked to motivated reasoning
Publication bias or the file-drawer effect – the practice of some researchers to only publicise positive studies
Cherry-picking – only choosing examples and data that promotes your intended conclusion
Confirmation bias – the act of only recognising or recalling information that supports a pre-existing belief
Falling into the habit of conducting motivated reasoning, even subconsciously so, is a common problem. The best place to start is with awareness of the problem. As the physicist Richard Feynman once said, “The first principle [of science] is that you must not fool yourself — and you are the easiest person to fool.”
About the Author, Thomas Maydon
Thomas Maydon is the Head of Credit Solutions at Principa. With over 13 years of experience in the Southern African, West African and Middle Eastern retail credit markets, Tom has primarily been involved in consulting, analytics, credit bureau and predictive modelling services. He has experience in all aspects of the credit life cycle (in multiple industries) including intelligent prospecting, originations, strategy simulation, affordability analysis, behavioural modelling, pricing analysis, collections processes, and provisions (including Basel II) and profitability calculations.