Accurate measurement of the pore size distribution (PSD) in coals is crucial for guiding subsequent coalbed methane (CBM) engineering practice. Currently, mercury intrusion porosimetry (MIP) measurement has been widely used as a PSD testing method due to its effectiveness and convenience. Nevertheless, it is worth noting that the elevated pressure during the MIP experiments can lead to matrix compressibility, potentially causing inaccurate estimations of PSD in coals. Therefore, correction methods are used to modify the PSD in the high-pressure segment to improve the accuracy of MIP data. This study proposed a novel method with higher accuracy and convenience for calculating the matrix compressibility coefficient compared to the traditional calculation methods. Firstly, the matrix compressibility coefficients of six coal samples were calculated by using low-temperature nitrogen adsorption (LTNA) data. Subsequently, by utilizing the mathematical correlation between Kc (the compressibility coefficient of the coal matrix) and Ro,max (the maximum vitrinite reflectance) from prior research, a novel statistical method was designed to determine the matrix compressibility coefficient of the samples. Finally, the statistical matrix compressibility coefficient determination method was used to examine the fractal characteristics of the actual PSD. The results indicate that when the pressure exceeds 24 MPa, the volume obtained from mercury intrusion exceeds the pore volume measurement. The Kc calculated using the traditional correction method is in the range of 0.876–1.184 × 10−10 m2/N, while the Kc values of our proposed statistical correction method range from 0.898 × 10−10 to 1.233 × 10−10 m2/N, with a comparison error rate of ~0.11–5.25%. The MIP data greater than 24 MPa were effectively corrected using the statistical correction method, thus reducing the mercury intrusion volume error by 91.75–96.40%. Additionally, the corrected pore fractal dimension (D2) values fall within the range of 2.792 to 2.975, which are closer to the actual values than the pore fractal dimension range of 3.186 to 3.339.