The Role of the Statistician: Scientist or Shoe Clerk (1974) [pdf]

>"The statistician who wants to be more than a shoe clerk must, therefore, be willing to risk unpopularity and even unemployment to achieve a study with a good scientific plan."

Not just statisticians, although they have many good ideas regarding the planning. Unfortunately, a lot of studies that get funded can never actually achieve the stated goals, but no one wants to admit it because it would ruin careers. The worst is that naive grad students get suckered into this mess, are given an impossible to accomplish task, and then have incentive to believe somehow the significant p-value means their theory is true.