BIAS & the Quality of Data


To separate good information from bad stick to the ABCs of forensic investigators: Assume nothing; Believe, no one; and Check everything. For people issues there is also a D: Dollars are usually a signpost to the root of an issue


Evidence - called here data - as discussed in Bias Aware Thinking, is made up of an assortment of objects, information, testimony, indeed any sort of clue that supports or challenges a claim, theory, or argument. Data, of course, can be fraudulent, erroneous, incomplete, or the whole truth of the matter. Guidance to determine the quality of this data can often be found by an awareness of bias.

 

When a self-interested group has made use of some data and the findings appear to be self-serving then the conclusions would generally be examined for bias. It is usual in those cases that supportive aspects of the matter were considered and other facts were either ignored or downplayed. But when one agrees with a certain conclusion the existence of biases in the data supporting it will rarely be questioned. As data quality is a major piece in solving The Bias Puzzle the merit of all data should established as far as possible or practical. This is no easy matter as the quality data can be misinterpreted when being used. So the unconditional rule is to examine data skeptically, whether one agrees with its conclusions or not.

 

Perhaps the best way to evaluate the quality of data is to solicit input from diverse sources; especially from those whose brains function differently from one's own (see Step #10 in Bias Management).  That is why most legal systems have juries of more than one person and information is presented by a prosecution and a defense each with (in many jurisdictions) their own experts.  

 



An analogy of how data can be biased is shown in sliding tile puzzle graphics below. There are eight facts represented by numbers 1 to 8; but the logical arrangement, or conclusion, from those facts can be different - only one of which would represent a truth. The bias would be in the determination of the appropriate arrangement - "sequential line-by-line" or "circular".


both arrangements of the tiles seem logical but only one can represent the truth

(Facts can be arranged to conceal biases)