A Methodology to Predict Specific Communication Themes from Overall Communication Volume for Individuals and Teams
Abstract
We focus on a means to code voice communications and derive communication measures because communication plays such a critical role in military decision making and mission accomplishment. Voice communication has proved labor intensive to code manually and, beyond simple counts of utterances, has proved relatively intractable to automate coding even for powerful computers. The methodology we describe has the potential to alleviate a significant portion of the current coding burden. It only assumes there is a technology to count the number of utterances per trial. The process involves randomly selecting a subsample from a larger data set, manually coding the subsample using standard manual coding procedures to produce a small set of communication measures, constructing regression models using corrected part-whole correlations to predict each communication measure from the number of utterances, and applying the models to predict communication measures for the remaining part of the data set. This methodology was tested using data from a recent study. Results revealed acceptable corrected part-whole correlations and subsequent regression models. Moreover, predicted communication scores from the subsample based regression models showed similar communication patterns found for scores derived from the whole sample. Implications of these finds are discussed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2006
- Accession Number
- ADA463297
Entities
People
- Elliot E. Entin
- Shawn A. Weil
Organizations
- Aptima (United States)