Importance of using rigorous statistical methods to analyze low energy laser experimental data: Part two
Related Experiment Video
Updated: Sep 2, 2026
Dependence of Laser-induced Breakdown Spectroscopy Results on Pulse Energies and Timing Parameters Using Soil Simulants
Published on: September 23, 2013
Background And Objective:
Numerous authors have reported successful alteration of peripheral nerve action potential characteristics through application of low energy laser irradiation (LELI). The statistical analysis that accompanies many of these reports frequently does not account for the special nature of the data generated in typical LELI experiments. The objective of this study was to evaluate the application of repeated measures linear regression techniques to the analysis of this type of data. Issues of analyzing raw versus normalized data, proper accounting for correlation between measurements, and discrete time point hypothesis testing were addressed.
Study Design/Materials And Methods:
The data analyzed in this work were generated from an experiment in which in vitro frog sciatic nerves were irradiated with a helium-neon laser using a variety of treatment protocols. Compound action potential (CAP) amplitude, latency, depolarization rate, and repolarization rate were recorded at 1-minute intervals for 135 minutes for each nerve. Laser-induced changes in CAP parameters were analyzed using various repeated measures linear regression models.
Results:
The findings of statistical significance were highly dependent on the rigor of the regression model applied. Application of the same regression model to raw and normalized data produced different findings of significance. Determination of significant contrasts was highly dependent on how well the regression model accounted for the correlation between repeated measurements made on the same nerve. In general, models that failed to account adequately for this correlation produced more findings of significant contrasts than increasingly rigorous models. Finally, discrete time point hypothesis testing on normalized data can suggest improper statistical conclusions if the proper correlation structure is not applied to the data set.
Conclusion:
Linear regression analysis offers advantages over discrete time point hypothesis testing in the analysis of highly correlated serial data of this type. Trends in the behavior of the measured parameters are evident, rigorous accounting for correlation between measurements is facilitated, and hypothesis testing is highly flexible.
