Andrew Gelman has a long blog post discussing 3 metaprinciples of statistics. http://www.stat.columbia.edu/~cook/movabletype/archives/2010/01/bayesian_statis.html
“First, the information principle, which is that the key to a good statistical method is not its underlying philosophy or mathematical reasoning, but rather what information the method allows us to use. Good methods make use of more information.”
I can go a bit further here and invoke a principle of mine: data drives out analysis. The availability of better data actually measuring something means we no longer need complicated methods to estimate it. As a result, the complicated analysis has to (and can!) move further out to the edge of the scientific envelope.
“My second meta-principle of statistics is the methodological attribution problem, which is that the many useful contributions of a good statistical consultant, or collaborator, will often be attributed to the statistician's methods or philosophy rather than to the artful efforts of the statistician...”
In other words, if you are successful in solving an applied problem, you are likely to attribute this success partly to you and partly to your method – sort of like fixing something around the house with a Dremel tool. Your major contribution may well have been knowing what needed to be fixed in the first place and you might have fixed it just as well with an electric drill.
“My third meta-principle is that different applications demand different philosophies.”
A good principle that needs no further illustration beyond noting that some statisticians work on medicines that might heal/kill people and I work in marketing research.