Abstract
The purpose of this study was to create and test two hypothesized models for swing in jazz performance. In order to estimate the hypothesized fully latent variable Swing model, eighth note samples (N = 815) from five improvised solos by Chris Potter were analyzed. Three first order factors were identified and examined that were hypothesized to define swing: eighth note duration, beat placement, and note dynamics. These three factors were measured with accuracy beyond a thousandth of a millisecond. Nine second order factors were identified and grouped into soloist-controlled variables and non-soloist controlled variables: metrical beat placement, melodic character, intervals, articulation, range, underlying harmony, tempo, bass beat placement, and drummer beat placement. The methodology of the measurement system and data collection procedures prescribed in this research study was found to contain limitations that prevented the administration of a proper statistical analysis for an estimation of the proposed models. Results of the simultaneous multiple regression analysis revealed that metrical beat placement, melodic character preceding, melodic character succeeding, interval preceding, interval succeeding, articulation, range, underlying harmony, and tempo combined to account for 6.7% of the variance in eighth note duration (N = 394), with interval preceding and interval succeeding having a statistically significant effect. Metrical beat placement, melodic character preceding, melodic character succeeding, interval preceding, interval succeeding, articulation, range, underlying harmony, and tempo combined to account for 22.3% of the variance in note dynamics (N = 231) with tempo having a statistically significant effect on note dynamics. The omnibus test for beat placement (N = 99) was found to be statistically insignificant.