Clarify the regression fitting procedure and reconcile the correlation discrepancy for Zhang et al. (2005) PNAS Figure 1c

Determine whether the exponential curve fit for Figure 1c in Zhang et al. (2005) Proceedings of the National Academy of Sciences was obtained using the Kvalseth (1985) and GraphPad procedure described in Appendix B or by a different method, and reconcile the discrepancy between the published correlation value (0.999) and the recomputed correlation (~0.954), including whether a typographical or numerical insertion error occurred.

Background

In re-evaluating past results, the authors re-fitted the exponential decay in Zhang et al. (2005) PNAS Figure 1c and obtained a correlation lower than the published value. They note a likely typographical error in a parameter and uncertainty about the original fitting method, as the data analysis was conducted elsewhere and documentation is unavailable.

References

Srinivasan is not sure if this was the procedure used in the study, because he was not involved in the data analysis part of this paper. The reason for the discrepancy between the published value of 0.999 and the newly computed value of 0.954 is either (a) an error of numerical insertion, or (b) the original calculation of the regression was performed using a different method, information about which is no longer available.

— Comment on Miscalibration of the Honeybee Odometer (2408.11520 - Srinivasan et al., 2024) in Appendix B, section 'Zhang et al. 2005 [23] Figure 1c'