Previous Page

|

Next Page

Analyzing a Full Factorial Design

The ADX Interface provides a large repertoire of graphical tools and statistical methods for analyzing designed experiments. However, only a small subset of these techniques is needed for any given experiment.

The analysis strategy followed here for the Tablet Formulation design is as follows:

• Explore each response by using a main-effects plot, an interaction plot, an -way effect plot, and a factorial plot with the Explore Data window.
• Select and investigate a model with the Choose a Response window.
• Determine which combination of factor levels will simultaneously provide the desired responses by using the prediction profiler in the Response Optimization window.

Note that for a design with sufficient observations to estimate the variance of the experimental error (underlying noise), ADX computes -statistics with corresponding -values for a formal assessment of statistical significance. This will always be the case for factorial designs, unless you set the master model to include every possible effect up to the highest-order interaction.

Exploring the Response Data

Fitting a Model

Determining the Predictive Model