Interactivity Theory: Analyzing Human Environments Using Linear Prediction Filters

Analyses of complex interactive human environments pose analytic difficulties for commonly used methods. Linear prediction filters were selected as a methodology that could realistically reflect the characteristics of an interactive environment and conform to appropriate scientific criteria empiricism, replication, prediction, and parsimony. Filters were compared to related linear models e.g., ANOVA, path analysis. Scientific criteria were used to identify weaknesses in interactive applications of traditional linear methods. A multichannel linear prediction filter was derived and integrated with a measurement model, producing time-series factor analysis. The model was applied to data showing a long term cyclical relationship between promotion rates in the U.S. Army and survey measures of company effectiveness.