In this paper we introduce a new PYTHON package, the PULSAR SIGNAL SIMULATOR, or PSRSIGSIM, which is designed to simulate a pulsar signal from emission at the pulsar, through the interstellar medium, to observation by a radio telescope, and digitization in a standard data format. We use the PSRSIGSIM to simulate observations of three millisecond pulsars, PSRs J1744-1134, B1855+09, and B1953+29, to explore the covariances between frequency-dependent parameters, such as variations in the dispersion measure (DM), pulse profile evolution with frequency, and pulse scatter broadening. We show that the PSRSIGSIM can produce realistic simulated data and can accurately recover the parameters injected into the data. We also find that while there are covariances when fitting DM variations and frequency-dependent parameters, they have little effect on timing precision. Our simulations also show that time-variable scattering delays decrease the accuracy and increase the variability of the recovered DM and frequency-dependent parameters. Despite this, our simulations also show that the time-variable scattering delays have little impact on the root mean square of the timing residuals. This suggests that the variability seen in recovered DM, when time-variable scattering delays are present, is due to a covariance between the two parameters, with the DM modeling out the additional scattering delays.