If my reading is in any way accurate (my own version of sample preparation bias) there are nearly 200 varieties of bias. What does this say about our decisions made so far?
Cognitive bias tricks people into spending money now rather than saving for the unknown stranger they will morph into by the time they retire (or go on that around the world holiday or get a new kitchen).
Confirmation bias is the one we data analysts are confronted with more often than we would like. People look for data to confirm what they already think rather than understanding what the data is showing.
Sample Preparation Bias is caused when data analysts, business colleagues, or anyone really chooses what data should be collected for use in analytics. Truly limiting the relevance of the analysis and severely reducing how fit-for-purpose the data is for evaluation.
I’m one of many who’ve been trying to get perspective on managing bias. We work in industries where competent decision making is fundamental. The effect a bad decision can have on unsuspecting people is enormous.
My mind is distracted about bias.
I know a lot of people thought that big data and powerful analytical tools would remove bias. Put all the data in one big unstructured environment, apply a probing algorithm and hey presto problem solved.
Not quite so fast.
This all reminds me of a TED talk I watched last year. Tricia Wang is an Ethnographer (I had to look that one up) which means she undertakes systematic studies of people and cultures.
Most notably in 2009 Tricia worked for Nokia studying the use of technology in low income populations. In particular (as outlined in her TED talk) she advised Nokia to jump on the smartphone bandwagon. They declined, and the rest is history.
Tricia introduced Quantification Bias to my vocabulary. “The unconscious belief of valuing the measurable over the immeasurable.”
Tricia discussed how useful Big Data is with quantitative analytics over contained systems. But when it comes to dynamic and unpredictable systems where new factors can be introduced at any time Big Data is much less useful.
Tricia calls the later Thick Data (which I take as rich with nourishing density). Pointing out that to analyse this kind of data you need people. To think. To provide further information. To apply their experience. And to understand the human perspective.
I agree with this distinction.
I also agree with Tricia’s closing statement that to make better decisions we need better data (all types of data), better algorithms, and better output.
So now I’m thinking about how to hang this all together? What will it look like as a business model? And will Ethnography be the new career de jour?
Data – Mel.