What Is Empirical Evidence?
Empiricism is the basic practice of science. Science can be described as empirical because it relies on direct experience or observation in order to describe or explain phenomena. In other words, a scientific or empirical approach is inductive, and bases its explanations upon that which can be directly observed in a replicable or repeatable manner.
The requirement concerning empirical observations being potentially replicable is key. This is what differentiates science from mystical or religious traditions. Science does not rely on dreams, or visions, or faith in the authority of sacred texts or spiritual beings as a basis for knowledge. That's why science does not recognize the existence of the human soul, for example, since it cannot be observed. Similarly, science can neither prove nor disprove the existence of God, since God is generally believed to be transcendent (i.e. as existing beyond nature, and therefore potential observation).
This is, once again, because empiricism is skeptical of anything which cannot be replicably observed. Empirical evidence is confined to only those phenomena which more than one person can observe. If someone claims that the tallest mountain in the world is in the Himalayas, for example, it must be possible for others to go there and confirm this contention. And if a taller mountain is found elsewhere, opinions must be revised based upon this new evidence.
Empiricism, however, is not equivalent to modern, Western science. This suggests a contrast between modern science and science as such. After all, all cultures rely on direct experience to a large degree in developing their technologies and patterns of subsistence. What makes modern Western science unique is that it has systematized its empiricism, especially through measurement and formal experimentation.
Another central empirical concept which all science makes use is the concept of the hypothesis, and the method of testing hypotheses against the empirical evidence. An hypothesis is a statement about relationships which can possibly be shown to be /untrue/. Hypotheses are basically if/then statements, which are intended to be predictive (eg. if Theory X is correct, then we would expect to observe Y).
Note that the definition does not say "relationships which can possibly be shown to be true, or proven correct." This is because the scientific method proceeds through the /falsification/ or /testing/ of hypotheses against empirically observable data. (In the above example, you may /not/ observe Y, as predicted, which would falsify the tentative hypothesis "X").
When you find an hypothesis which appears to stand against empirical testing, you haven't proven your hypothesis is correct. Rather, what you have done is shown that it works, or accomplishes your goal of explaining or understanding a particular phenomena. At least for the time being. Due to its inherent skepticism, empiricism always assumes that more adequate or more complete explanations may be found in the future.
In other words, as the old saying goes, "science never proves anything." The perpetual revisability of scientific opinion in light of new hypotheses tested against the empirical data is its greatest strength as a means of understanding the world around us.
This also makes science quite unique as a form of argumentation. The legal profession, for example, also makes use of empirical evidence, as forensics does generally. But the aim is quite different.
The difference is precisely the same as the contrast between philosophy and sophistry as argumentative styles in ancient Greek thought. Like philosophy, science is a search for truth or knowledge. Or at least for /provisional/ truths. The legal profession is based on sophistry, since neither the prosecution nor the defense are ultimately interested in truth itself-each is interested in winning the argument on behalf of their client (either the defendant or the state). Empirical evidence in the later case is not an end in itself (as it is for science and philosophy), but merely a means to an end (winning).