Uncertainty and confidence

The word 'uncertainty' to most people means that something is not known and/or unlikely, for example, "I'm uncertain if it is going to rain today" or "I'm uncertain if I passed that test".

The word 'uncertainty' to a scientist is a measure of how sure we are of an outcome.  For example: "A single squirt of cheese will be 3cm long, within a 5% level of uncertainty".  That means that we are 95% certain that the length of the squirt of cheese will be 3cm long, but there is a 5% chance that the cheese could be longer or shorter than 3cm.

Confidence is another word which can be confused when crossing between science and popular media.  In popular media if someone says that they are confident that something is going to happen, that means they think there is a very high likelihood that it will.  However, in science, confidence is merely the level of certainty and this can range from being very certain something will happen to being very uncertain, depending upon the level of confidence (usually expressed as a percentage or as a range of results expected within a likelihood).

Graphs in science often indicate the "95% confidence boundaries" for each data point.  These are the range within which 95% of results will fall - sometimes the confidence boundaries are very large, and sometimes very narrow depending upon the data available.  This indicates to other scientists how precise the average, stated figure is.  For example, if the mean (average) squirt of cheese was 5cm long from one type of packaging, and the range of measurements within the 95% confidence boundary was from 4.8cm to 5.2cm, then there it is very likely that the next squirt of cheese from that packaging will also be around 5cm long (presuming an endless supply of cheese).  However, if another type of packaging also has a mean squirt of cheese 5cm long, but the range of measurements within the 95% confidence boundary was from 0.2cm to 9.8cm, it becomes much harder to predict the length of the next squirt of cheese, and it is important to record this degree of variation for any test results.

These different definitions for two commonly-used words (uncertainty and confidence) can cause confusion when translating science to a more general audience and it is this confusion that many refuters of science focus on when making their arguments about the invalidity of results.  All natural systems inherently contain some level of uncertainty when studied by scientists due to their high levels of complexity.  Predictions for the future, such as climatic models studying climate change, produce scenarios with varying levels of certainty, and changes in the parameters which feed into the model, such as the amount of fossil fuels burnt, affects which output scenario becomes more likely as time progresses.

It is important that scientists, policy makers, journalists and members of the public all "talk the same talk" when discussing data to avoid misunderstanding between the different parties.  Scientists will never be "certain" of a fact, there will always be a level of uncertainty due to the complexity of nature and the impossible task of accounting for every possible factor which may affect the outcome.  But that does not mean that they are uncertain enough to warrant being ignored, or the problem left until it becomes unmanagable.