By Jan Hanspach
In his recent post on this blog Joern Fischer complained about the loss of ecological knowledge and intuition, while the application of more and more complex statistical techniques is getting more and more important. Plainly spoken, he felt that good ecology was being replaced by fancy methods. Here are some points why I disagree:
1) Statistics and its applications are progressing. Some methods are newly developed (generalized linear mixed effects models, fourth corner analysis) and others are much easier implemented due to higher computing power (e.g. multivariate and Bayesian methods). It’s only recently that ecologists have the means and the liberty to analyze data and to answer research questions where a simple t-test would have failed.
2) For young scientists it can spoil their career not to be fluent in programming languages and advanced statistics. Frankly, studies using fancier methods are easier to get published than the simple ones.
3) I agree that more complex statistics need more time to be understood, but I believe that many students would benefit greatly if stats was taught more properly (if taught at all), and if students were supervised more intensively (if supervised at all).
4) The problem we face is not only that statistical techniques are becoming increasingly diverse and complex. It’s all of science that is following that trend. More scientists are researching, more publications are coming up and more knowledge is gathered all the time. How do we handle all this? Why does nobody complain that we are doing less well ecology because we are busy with reading papers, visiting conferences, piling up rejection letters, reading blogs or whatever fills most of the day of an average ecologist?
One way to handle increasing complexity is to become more and more specialized – and indeed, that might mean to become a quantitative ecologist. Another solution might be to have a shallower understanding of things (more and more trivial questions?) which is fostered by a trend to ask bigger questions (What is the influence of global environmental change?) in a broader context (interdisciplinary research).
I believe that we need both good ecological knowledge and expertise in advanced statistical methods. Statistics, simple or complex, are a standardized way to draw conclusions from ecological data. It’s crucially important for every ecologist to be literate in the world of applied statistics (you don’t need to know all details), since it shapes how we formulate hypotheses, design our studies and handle data – and consequently how we do science.