Eye2theLongRun

Considering some physics recently I was reminded of the well-established fact that to date at least, all measurement is ever an imprecise business. Measurement is an imperfect act insofar as if the object is to measure one specific phenomenon and that alone – undistorted, without bias, devoid of contaminating adjacent effects, such untouched forms of purity are unattainable,

It is simply the case that the very act of measurement generates “unwanted” intrusions.

Summarised and simplified we have at least:

Systematic Error

• Consistent and predictable bias in measurements.
• Often caused by faulty equipment, calibration errors, or environmental conditions.

An example: A scale that always reads 0.5 kg too high.

• Unpredictable and varies from one measurement to another.
• Caused by human limitations, small fluctuations in experimental conditions, or instrument sensitivity.
• Example: Slight hand tremor affecting stopwatch timing.

Random Error

These errors produce what we conveniently term “noise”. Phenomena which in and of themselves are distinctly separate from and are not that which we seek to measure – but do exist and their presence is captured nonetheless.

Most Significantly

We may be testing or probing to find the presence of (say) a given disease for example. Even where no such disease or symptom thereof exists at all (i.e. patient fit and well – no disease), nonetheless we will measure  something. Noise will register in our equipment or otherwise make itself felt in our data.

The trouble is in a great many (most if not all) cases we cannot reliably distinguish the “signal” from the “noise”. We may therefore mistake “absence” for “presence”. No disease but a non-zero reading. We simply do not know with great reliability where noise stops and the phenomena which we are interested in  begins.

This is not always simply because of our sloppy ways. The uncertainty principle suggests for instance that even at the most disaggregated of scales we cannot establish both speed of movement and location. One or the other perhaps but both most certainly not. The situation is difficult at other scales as well.

Conclusion

It does strike me therefore that “if we look long enough and hard enough we will find something” but that “something” will not necessarily be anything other than the noise occasioned by measurement.

This effect is worth pondering – notably in medicine and behavioural sciences. In the case of the latter, it may account for the fact that in early days researchers uncovered perhaps five or so cognitive biases. On a recent count Wikipedia claimed to identify 175 or so cognitive biases. Really?