There's an old saying in statistics: "Get there before anyone collects any data." Don't just rush headlong into gathering data and expect to get useful insights by "analyzing" said data. With the training I provide, you'll be better able to define precisely the who/what/where/when of the events you are counting and the measurements you are making.

Often under-emphasized in introductory statistics classes, graphing your data (and graphing it well) is probably the most important step in an analysis. Good graphs should provide insights into relationships in your data. Sadly, the familiar go-to tools (e.g. Microsoft Excel spreadsheets) are quite good at making bad graphs and only fair at making good ones. With some new methods and tools I'll introduce, you'll be able to understand and show relationships in your data more clearly.

After graphing your data, this is probably the next most important step in an analysis. It's always important to look at both graphs and summary statistics

Finally we come to statistical tests---the (in)famous p-values. These are not as important as your training may have led you to believe, but sometimes they are useful.

Modeling is the bread-and-butter of data analysis. It is a vast topic which we cannot cover in a single training class, but I can help you understand the basics, when certain models might suitable, and point out some of the pitfalls to be mindful of.

Moving from the numerical results of an analysis to the operational meaning of those results requires that you understand the "provenance" of the data (how it came to be in your hands), how terms were defined, the analytical methods used, and the assumptions built into those methods. This all ties back nicely to issues discussed in the early planning stage.

R is a computer program for statistical analysis and graphing. It has become the gold standard in scientific research, in many industries, and increasingly in public health. R is free to use, by anyone for any purpose---increasingly important with agency budgets tightening and licenses for some commericial statistics programs costing thousands of dollars. While there are point-and-click interfaces to R, its real value lies in the ability to write simple programs--"writing code"--that you can execute again whenever needed, perhaps with new data. This makes it easy to automate recurring tasks, such as weekly disease surveillance reports, quarterly program assessments, or what have you. It also documents your analytical decisions. Imagine trying to answer a colleague in a neighboring jurisdiction, your chief, or your county executive (or even your future self, six months from now!) when they ask how you defined a particular adverse event, or whether you excluded the data from from the week of the big ice storm. Could you answer confidently? If you have the code, you could.

REDCap stands for Research Electronic Data Capture. It was developed in 2004 at Vanderbilt University. REDCap is used by medical researchers all over the world, to safely record, store, and manage protected health information. It is also available to governmental and non-profit organizations. REDCap provides a scaffold on which you can build a data management system for any project, suited to your needs, without having to start from scratch each time. Except perhaps for budgets and for tiny, strictly one-off things, REDCap is almost always better than a spreadsheet for any data collection activity.

The software sits on a server--either yours or Vanderbilt's. Nothing needs to be installed on desktop computers; all interaction is via your web browser. Your staff can enter data, and projects can also, if you choose, contain public surveys. Everything that happens with the data and with the project design is automatically documented, providing a robust audit trail.

I have extensive experience building and operating REDCap projects and can provide training. (If you host your own REDCap instance, technical support for the software must reside with your agency.) Read more about REDCap here.