There are many different scientific and practical areas of focus that rely on the collection of quantitative data. Collecting quantitative data is, for instance, of central importance in research-based fields such as chemistry, physics, and even some branches of linguistics. It is also essential for testing and other purposes in engineering, computer science, and other data-intensive fields and projects aimed at producing an end product. The specific methods used for collecting quantitative data vary drastically across projects, but there are some principles of data collection that can be widely, if not universally, applied. It is, for instance, important to take all means possible to eliminate human and experimental error, to collect and analyze all data rather than only that which fits one's theories, and to run an experiment or test multiple times to check for errors.
Though minimal error is occasionally acceptable, in some cases it can lead to substantial inaccuracy or even to the failure of a project. Whenever possible when collecting quantitative data, then, one should determine the degree to which error can be tolerated. The techniques and devices used for collecting quantitative data should be able to do so within this tolerable range of error. If they cannot, it is probably necessary to refine the data collection method or to come up with an entirely new one.
When collecting quantitative data it is often tempting to record and use only the results that correspond to prior experiments or to theoretical expectations. This is especially true when only a few of the collected numbers differ significantly from expected results. These outliers, however, can be extremely important and should not be ignored, especially if they recur in subsequent experiments. Unexpected results can indicate problems with the experimental procedure or materials or may even suggest that the existing theories on the topic of experimentation or testing are incorrect. The process of collecting quantitative data can only be effective and objective when the researcher collects and reports all data.
Running multiple independent trials is an excellent way to minimize error when collecting quantitative data. Doing so can reveal issues such as device calibration, human error, or the effects of unexpected and uncontrolled variables. When possible, distinct groups of people should run the tests or experiments aimed at collecting specific quantitative data. The two groups can compare all methods and variables if they collect different results, thereby allowing them to isolate the particular errors that arose during the process of collecting quantitative data.
