Within the framework of $\mathtt{FREYA}$ event-by-event fission simulations, we study the sensitivity of various neutron observables to the input yield function $Y(A,Z,\mathrm{TKE})$ on which the fission event sampling is based. We first perform a statistical analysis of the available fission data in order to determine the distribution of possible yield functions $Y(A,Z,\mathrm{TKE})$ and we then construct a sample of 15 000 such yield functions. For each of these, $\mathtt{FREYA}$ is used to generate one million fission events, leading to a corresponding ensemble of fission observables, including the neutron multiplicity distribution and its factorial moments, the neutron energy spectrum, and the neutron-neutron angular correlation. This procedure allows us to study the sensitivity of those neutron observables to the uncertainty in the experimental data. Particular attention is given to the pronounced anticorrelation between the mean neutron multiplicity $\overline{\ensuremath{\nu}}$ and the mean total fragment kinetic energy $\overline{\mathrm{TKE}}$. Because the former observable is very well determined experimentally, it is possible to exploit that inherent anticorrelation to derive a significantly stricter tolerance on $\overline{\mathrm{TKE}}$. In addition, we study the sensitivity to the various $\mathtt{FREYA}$ parameters and we introduce a method for determining a $A$-dependent $x$ parameter, $x(A)$, based on the measured $A$-dependent neutron multiplicity, $\overline{\ensuremath{\nu}}(A)$.