The apparent spin direction of a spiral galaxy changes based on the location of the observer relative to the galaxy, and therefore it can be assumed that the spin directions of spiral galaxies as observed from Earth are randomly distributed. Here I use large datasets of spiral galaxies from SDSS and Pan-STARRS to show that the distribution of the spin directions of spiral galaxies is not symmetric, and changes based on the direction of observation. The parity violation in spin directions exhibits multipoles, and has a best fit to a quadrupole with probability of less than 6σ to have such distribution by chance. Comparison of SDSS and Pan-STARRS shows that the two sky surveys exhibit nearly identical patterns of the distribution of spiral galaxies, even when the datasets are completely orthogonal. Analysis of galaxies with spectra shows that the parity violation increases with the redshift in all tested directions of observation. The mean redshift of the galaxies is ~0.12, making the dataset far larger than any supercluster or other known astrophysical structure. The data analysis process is fully automatic, and is based on deterministic model-driven algorithms that follow defined rules, and are not sensitive to the size of the galaxies in the dataset. It does not involve manual analysis, and therefore the analysis is not subjected to a possible human bias. It also does not involve machine learning or deep neural networks that rely on complex unintuitive rules that can capture instrumental or atmospheric biases.