Systematic and statistical studies have been conducted in order to develop an understanding of the parametric dependences of both the global and thermal energy confinement times at low aspect ratio in high power National Spherical Torus Experiment discharges. The global and thermal confinement times of both L- and H-mode discharges can exceed values given by H-mode scalings developed for conventional aspect ratio. Results of systematic scans in the H-mode indicate that the confinement times exhibit a nearly linear dependence on plasma current and a power degradation weaker than that observed at conventional aspect ratio. In addition, the dependence on the toroidal magnetic field is stronger than that seen in conventional aspect ratio tokamaks. This latter trend is also evident in statistical analyses of the available dataset. These statistical studies also indicate a weaker parametric dependence on plasma current than found in the systematic scans, due to correlations among the predictor variables. Regressions based on engineering variables, when transformed to dimensionless physics variables, indicate that the dependence of BτE on βt can range from being negative to null. Regressions based directly on the dimensionless physics variables are inexact because of large correlations among these variables. Scatter in the confinement data, at otherwise fixed operating parameters, is found to be due to variations in ELM activity, low frequency density fluctuations and plasma shaping.