Scientific research in a number of fields is in a state of crisis due to the discovery that many published results are non-reproducible, and applied statistics has been assigned a substantial share of the blame. Proposed solutions range from requiring independent statistical review of results for major journals to abolishing the use of certain methods entirely.
Lennox argues that the problem does not lie with statistical methods, but rather from misleading training for non-statisticians. The talk is intended to establish that statistics is not just a set of numerical procedures, but rather a distinctive way of thinking about and solving problems. Real-world examples demonstrate the pitfalls of "procedural" statistics, and that non-statisticians can be successful by approaching statistical challenges in the same way that they do problems in their field of expertise and by leveraging the statistical expertise available at the laboratory as necessary.