Scientific Test Analysis & Design Techniques (STADT)

Duration of courses

4 Days



About the Course

Scientific Test Analysis & Design Techniques (STADT) course provides test practitioners with the ability to apply a mathematical approach to the analysis and design of tests. The best analysis and design tools from combinatorial testing and Design of Experiments (DOE), also known as Multi-Variable Testing (MVT), are presented. It covers the key terminology and various options to analyze and design tests, showing why orthogonal designs are the most effective and efficient testing technique.

This course covers many examples in the world of test and evaluation and gives participants practice at test design and analysis. DOE is shown to be the science of data collection as it applies to testing and that it must be in the toolkit of anyone who wants to be competitive in the global market of test and evaluation.

Course Outline

This course covers the analysis activities that must precede a test, including the first line of defense against variation and Measurement System Analysis (MSA). Testing strategies, such as modeling, screening, and verification/validation (confirmation) testing are presented. The 12-step approach to analyzing and designing a test provides a framework for adequately considering all aspects of the test.

Basic graphical and regression analysis of the resulting test data are also covered:

  • Two-level and 3-level designs will be presented for both screening and modeling. Response surface designs will be shown to be more efficient than factorial designs when generating non-linear mathematical models of performance.
  • Nearly Orthogonal Latin Hypercube Designs (NOLHDs) will be shown to be very valuable when testing many factors at 3 or more levels, such as when dealing with high fidelity computer simulation models.
  • Simple Rules of Thumb are provided for sample size and design selection, along with determining the significance and power of a test.
  • Interpreting regression output and the coding of factors levels will facilitate the analysis of data not collected under a DOE strategy and provide a means of analyzing data coming from multiple test scenarios.
  • High Throughput Testing (HTT) provides a combinatorial testing approach that is extremely useful in verification/validation testing when there are many factors, both qualitative and quantitative, each with a differing number of levels. HTT is shown to provide superb test coverage at much lower cost than traditional designs. Latin Hypercube Sampling (LHS) and Descriptive Sampling (DS) are shown to be very useful space-filling designs when only a limited number of tests can be conducted.