Spectrally encoded Mueller matrix measurement has the advantages of high measurement speed, compact structure, low loss, and no moving parts. Furthermore, all elements in the Mueller matrix can be obtained with only a single measurement. The main principle of this method is to use a set of phase retarders with a specific thickness ratio to modulate the Mueller matrix elements to the frequency channel of the spectrum; then, the Mueller matrix is demodulated through the Fourier transform of the spectrum. However, the thickness or phase error of the retarders causes a large error in the demodulated Mueller matrix elements. In this work, we theoretically obtained the general expression of light intensity with phase error and then calculated the phase error using a single sample. This method can avoid the influence of the different initial phases of different samples and improve the calculation accuracy of the phase error. We calculated the influence of phase errors of the retarders via simulation and experimentally verified the feasibility of the error calculation and calibration methods.