"hard" vs. "soft" sciences --- Trying to label certain sciences as "hard" or "soft" comes with its own set of issues and those terms can also be confused as making value judgments about the relative merits of certain disciplines. A more value-neutral way of distinguishing between disciplines is to assess the degree to which they have to apply statistical analyses to their datasets. This is often a function of how "messy" the data is, i.e., how many other confounding factors might be influencing the data. For example, when particle physicists use particle accelerators to generate their data they are considering the interaction of two particles in a highly controlled environment, whereas an ecologist studying a forest has to consider (or at least be aware of) all the interactions taking place that might have an effect on their data, and only by applying careful statistical analyses in the latter case, can the data collected be determined as more than likely giving the ecologist an answer to their question, so basically teasing the truth out of a very 'noisy' system from a data approach. So "hard" sciences in this sense rely less on statistics because they don't need to, whereas "soft" sciences use statistics to help make sense of messier datasets, and this distinction is certainly more of a spectrum than a hard line, and in no way should used to argue that some scientific disciplines are "better" or "more rigorous" than others. It's more about making sure the right tools are applied to the right questions and types of data that scientists use to do their research.

Source author: Ryan Haupt ryan.haupt@nysf.com