Scientific uncertainty is inherent in the scientific process. Often there is consensus in the scientific community about a specific topic or theory. However, progress in science can involve overturning consensus with new discoveries that are inconsistent with previous theories. Therefore, scientific certainty can be rare in some disciplines. Scientific certainty is only achieved when multiple trials of an experiment or a project give repeatable results.
Uncertainty in the scientific enterprise includes statistical uncertainty, where there is a measured range of variability. An example of statistical uncertainty is the accuracy with which the speed of sound in seawater can be determined. Any effort to measure sound speed has uncertainties in the distance over which the sound traveled, the temperature, pressure, and salinity of the water, and the measurement of the travel time. In early experiments, such as that by Colladon and Sturm (link to History section), it was difficult to determine all of these parameters with high accuracy. Their measurement, approximately 1435 m/s, therefore had significant statistical uncertainty. Modern experiments in which sound speed is measured in the laboratory allow much more accurate control of the variables, resulting in statistical uncertainties as small as 0.05 m/s. Therefore, when this experiment is repeated many times, the values vary by about 0.05 m/s.
Unlike the statistical uncertainty described above, where there is only one true value, biological systems have an uncertainty associated with the range of natural variability. For example, humans' best hearing levels vary between individuals and, in a single individual, over time. The best hearing level is statistically uncertain. Research on human hearing levels provides a range of hearing levels.
True uncertainty is where there is a question regarding the cause, result, or probability of an effect. A scientist understands that at any time, someone else may prove them wrong or refine what they have done. Depending on the science at hand, there can be a range of uncertainty:
- No uncertainty for specific facts, such as the sun produces radiant heat energy
- Intermediate levels of uncertainty for information with statistical variability and known probabilities, such as specific levels of hearing sensitivity
- High levels of uncertainty for cause/effect relationships of complex phenomena with many possible causes, such as marine mammal strandings