There exist discrepancies in measurements of the number and evolution of galaxy pairs. The pair fraction appears to be sensitive to both the criteria used to select pair fraction and the methods used to analyze survey data. This paper explores the connection between stellar mass estimation and the pair fraction of galaxies making use of STEEL, the Statistical sEmi-Emprical modeL. Previous results have found the pair fraction is sensitive to choices made when selecting what qualifies as a pair, for example luminosity or stellar mass selections. We find that different estimations of stellar mass such as photometric choice mass-to-light ratio or IMF that effect the stellar mass function also significantly affect the derived galaxy pair fraction. By making use of the galaxy halo connection we investigate these systematic affects on the pair fraction. We constrain the galaxy halo connection using the stellar-mass-halo-mass relationship for two observed stellar mass functions, and the Illustris TNG stellar mass function. Furthermore, we also create a suite of toy models where the stellar-mass-halo-mass relationship is manually changed. For each stellar-mass-halo-mass relation the pair fraction, and its evolution, are generated. We find that enhancements to the number density of high mass galaxies cause steepening of the stellar-mass-halo mass relation, resulting in a reduction of the pair fraction. We argue this is a considerable cause of bias that must be accounted for when comparing pair fractions.