Sunday, March 11, 2007

What Are the Steps to Obtain Unbiased Data?


What Are the Steps to Obtain Unbiased Data?


Unbiased data is possibly extremely difficult to collect. Yet according to Evans & Lindsay one must consider data as always subject to error and that while it may be considered valid data if it measures what was originally intended then how reliable it is often is dependent upon the quality of measuring instruments. As is apparent the human instrument is often knowledge rich and quality poor. Considering oneself as a flawed instrument would be a good first step to improving an awareness of information bias.


An example of my own information bias. I have my facts wrong here.

For example, a measuring instrument such as the oxygen management systems and pressurization systems on the Helios commercial flight which crashed a few years ago were ineffectively operating and producing the wrong results on oxygen content and pressurization reading "normal" when cirumstances were far from the case. Relying on faulty instruments data measurement impacted upon pilot performance resulting in great fatal errors.
The truth is that the instruments and all the warnings in the cockpit were providing 100% correct information, unfortunately the pilots were either, incapacitated or not capable of interpreting them correctly or had ignored them because they did not trust them, as has happened in many aircraft accidents. The accident report says so too. I just wanted to correct the analysis as I have also spent almost 30 years flying for the airlines.
Feroze Khan, Vice President at Stratford University
Evans & Lindsay further recommend that the validity and reliabilty of a company's ability to collect data be regularly audited externally or with cross-functional teams or through standardized input or automatic data templates or formats. This would ensure consistent framing to confirm or disconfirm whether data internally or externally generated actually results in improved corporate performance.


Many businesses according to their research appear to mismanage the representational sharing of data resulting in the effects of silos in the organisational structure which further undermines growth and leadership performance.


This suggests that the practice of making decisions collectively often requires the construction of a decision tree to determine the variables and as one step towards overcoming bias. While trees attempt to integrate subjectivity to minimize it this does not eliminate it. The difficulty of making decisions based on good data is aligned in the decision tree as an attempt to gain the input of as many decision-makers as are necessary for the possible nature of the problem or character to be easily discussed. Obviously an open information environment is necessary to maintain an interest in eliminating bias.


Myers-Briggs was employed in one learning program I took part in in 2003 related to adult education. I fail to recall my particular score but I am aware that the purpose of such tests are to allow individuals to explore the various learner activated or instructional strengths of character which are lacking in one's own result. When one is able to energize a sense of empathy, understanding and the evolutionary divergences in character strengths which easily rely on each other then one begins the steps necessary to perhaps a greater understanding and value for the input and collaboration of individuals with varying learning and instructional strengths.


Growing an awareness of the Vroom-Yetton model should assist in developing a greater awareness of collective data bias. Particularly reassuring is the categorization of decision processes which combined with the four outcomes (decision quality, decision commitment, efficiency and team development) imply a reliance upon mixed scanning even in the construction of the tree. As one seeks to practice exploring options particularly to yes/no questions one is perhaps attempting to concurrently measure the degree of acceptance of a possible decision or its acceptability and at the same time increasingly structuring the process to maximize collaborative input.


In such cases where one seeks to resolve a group decision through concensus one must hope to build it by increasing the probability that a group decision may reduce the level of information bias and in such cases confirm the adage that, "two (or more) minds are better than one." While some may be able to do this independently which implies two-mindedness, even learning about the actual sources of bias themselves is enlightening.


Thus, "no man is an island" also comes to mind. Search for the self. Awareness of the source(s) of bias is an obvious first step to future prevention.


Evans, J.R. & Lindsay, W.M. (2002) The Management and Control of Quality, Fifth Edition, South Western, Thomson Learning, USA.

No comments: