Customized In-House Training

Professionals at Stat Tenacity have provided training at small startups and large Fortune 500 companies on topics such as:

• Predictive Modeling

• Experimental Design

• Summarizing Data

• Comparisons of Multiple Groups

• Analysis of Dose-Response Data

We can provide predictive modeling and statistical consulting for you, or we can give you the tools to do it yourself. Our individualized statistical training classes are customized for your scientists and statisticians, based on your particular needs. You’ll get in-person, hands-on instruction in beginning, intermediate, or advanced statistical techniques.

Training is available on-site or remotely. Classes range from half-day to two-day sessions, depending on your needs.

Predictive Modeling Course

As data has become more readily available, organizations need to harness predictive information to their benefit. Uncovering predictive patterns requires the appropriate tools and processes to ensure that the patterns are relevant and valid for making predictions on other “out of sample” or future data.

This four-part course introduces the foundations of model building, as well as several traditional and modern predictive models. It is specifically designed for statisticians and practitioners who want to extend their expertise in this area. The course illustrates the predictive modeling process using examples from real data.

Our general predictive modeling training consists of four half-day segments (listed below). These modules can be tailored to meet your company's specific needs.

Part 1: Foundations and basics of regression (4 hrs)

Part 2: Modern regression techniques and pitfalls to avoid (4 hrs)

Part 3: Modern classification techniques (4 hrs)

Part 4: Practical challenges with data (4 hrs)

Mix-and-Match Training Modules

Each of these training modules has an optional hands-on component that uses real data. We can provide the hands-on portion using one or more of the following software packages: JMP®, R, GraphPad Prism®, and Minitab®.

Summarizing Data

Statistics plays a critical role in many parts of the discovery process, but like other fields, it has vocabulary that can be difficult to understand and methods that can be complex for researchers to use. This session explains statistical concepts and tools for addressing common research questions. We’ll address questions such as: What statistics should I use to summarize my data? Should I report the standard deviation or standard error? What is the meaning of a p-value? Does my molecule have a statistically significant effect?

Comparing Two Groups

Statistically comparing two groups of samples is a core research tool. Often, this kind of comparison is performed using a t-test, which can be generated in many software packages. While software has made it easy to compare two groups, you need to know the primary assumptions behind the test to ensure that the results can be trusted for making decisions.

This session provides recommendations on graphical and numerical summaries of two groups of data. We also provide an intuitive explanation of a t-test, the assumptions behind the test, and the resulting p-value. Although a p-value indicates statistical significance, the results may not be scientifically significant. The seminar will contrast this subtle difference. Under the framework of a t-test, we can also derive a confidence interval about the difference in group means. This summary is often underused, and we’ll explain why it should be a fundamental component of any report.

Comparing Multiple Groups

Comparing more than two groups is another core research tool that’s more complex than comparing two groups. Statistically comparing more than two groups of continuous data samples is best performed using an Analysis of Variance (ANOVA) test, which can be generated in many software packages. While software has made it easy to compare multiple groups, you need to understand the primary assumptions behind the comparisons to ensure that the results can be trusted for making decisions.

In this session, we provide recommendations of graphical and numerical summaries for multiple groups of data. Under the framework of an ANOVA, we are also interested in performing multiple comparisons tests of group mean differences. We’ll discuss which multiple comparison tests are appropriate for a given experimental design, which aren’t, and the differences between using multiple comparison tests and multiple t-tests. We also discuss the benefits and drawbacks of a nonparametric ANOVA.

Analysis of Repeated Measures and Complex Experimental Designs

Many experiments include more than one factor that contribute to the variability of the response. Failing to account for sources of variability due to repeated measurements, blocking factors, and multiple factors will result in an incorrect and less powerful analysis. While software has made it somewhat easier to perform complex statistical analyses, you need to understand certain primary assumptions to ensure that the results can be trusted for making decisions.

This session covers the following topics:

• Analysis of repeated measurements (i.e. body weights measured over time)

• Statistical analysis of experiments with blocks

• Statistical analysis of experiments with two qualitative experimental design factors (i.e. dose group and subject gender)

• Analysis of repeated measurements with two qualitative experimental design factors

• Nonparametric test alternatives

• Analysis of Dose-Response Data

This module is designed for researchers who work with dose-response data. It covers the statistical methods and approaches for fitting curves to dose-response data, the underlying assumptions of those methods, determining starting values, and the impact of fixing parameters on the final results. The session also reviews common functions associated with one-site saturation binding, receptor binding, and competitive binding. We’ll discuss the interpretation and presentation of results, visualization of data, and common pitfalls when fitting curves to this type of data.

An Introduction to Experimental Design

When you want to understand many factors, you may be tempted to perform one experiment for each factor. Studying multiple factors individually (one factor at a time) will likely miss crucial relationships between factors that affect the attributes. Performing one experiment for each factor also uses more resources and takes more time than studying several factors in the same experiment.

Statistical experimental design provides valid answers by using resources efficiently. In this module, you’ll learn fundamental experimental design concepts such as experimental units and types of factors that can affect the attributes. You’ll also examine types of experimental designs that are used to answer the question, Which factors affect the response?