ANOVA is one of the most popular statistical techniques in the social and behavioral sciences.
Unfortunately, the ANOVA framework has some serious drawbacks. First, it does not take the
expectations of the researchers fully into account. For example, researchers may expect that a new
treatment will work better than an old treatment, or that one task is much more difficult than
another, but ANOVA will ignore these expectations. Second, the more variables (and interaction
terms) we include in the analysis, the more difficult it becomes to extract the information the
researchers are looking for. Often, researchers are looking for overall (average) effects of their
treatments or manipulations, or they are looking for the specific (conditional) effects within welldefined
subpopulations. Often, these effects can not be obtained from a regular ANOVA analysis.
The goal of this project is to develop a new statistical procedure that looks and feels like the ANOVA
framework (so that researchers find it natural to use it), but where the expectations of the
researchers are explicitly taken into ccount, and where it will be much more easier to provide the
answers to the original research questions. In addition, we will create free, open-source and userfriendly
software so that these new statistical tools can be used by the whole research community.