This paper presents a robust method for determining the calibration coefficients in polynomial calibration equations, and discusses the corresponding calibration uncertainties. An attenuator method that takes into account all measurements with and without an attenuator screen was used to restrict the impact of the absolute calibration of the light source. The originally proposed procedure attempts to simultaneously determine all unknowns nonlinearly using polynomial curve fitting. The newly proposed method divides the task into two simpler parts. For example, in the case of a quadratic calibration equation, the first part becomes a quadratic equation solely for the transmittance of attenuator, which has an analytical solution using three or four sets of measurements. Additionally, it is straightforward to determine the median value and the standard deviation of the transmittance from the solutions using all combinations of measured data points. In conjunction, the second part becomes a linear fit, with the ratio of the zeroth-order to first-order calibration coefficients as the intercept and the ratio of the second-order to first-order calibration coefficients as the slope. These ratios are unaffected by the absolute calibration of the light source and are then used in the calibration equation to calculate the first-order calibration coefficient. How the new method works is straightforward to visualize, which makes its results easier to verify. This is demonstrated using measurements from the Joint Polar Satellite System (JPSS) J1 Visible Infrared Imaging Radiometer Suite (VIIRS) reflective solar bands (RSB) pre-launch testing.