Many biological systems, from flocks of birds, swarms of locusts, shoals of fish, to crowds of humans show collective behaviours which are emergent properties that manifest only at the level of a group. In each of these cases, local, individual-level interactions give rise to highly coordinated and synchronized emergent patterns that are observable at the group level. Theoretical, computational, and empirical research in this field have been used to address questions about various aspects of collective behaviour, from characterizing its structural and dynamical properties to deciphering how the individual behaviour produces emergent group patterns. However, most of these studies examine the group-level properties in terms of their mean values and do not attempt to characterize the variability present in the collective properties which arise due to the stochastic fluctuations. In real groups, stochastic fluctuations in the group-level properties arise due to the probabilistic interactions between finite number of individuals. As a result, this intrinsic noise present in collective systems can produce non-intuitive and non-trivial behaviours. Real groups in finite space will likely have some interaction with the boundary, and the observed collective dynamics may also include effects of interaction with the boundary. To gain comprehensive understanding of collective dynamics in real groups, examining the effect of interaction with the boundary is necessary. However, whether boundary interactions confound analyses of inferred interaction rules have not been investigated rigorously. In the current study, we examine how the parameters of the boundary conditions affect the simulated collective dynamics. We performed stochastic simulations of two data-inspired spatial models developed by Jhawar et al. that have contrasting collective properties: (i) where individuals show pairwise interactions and collective order is driven stochastically (ii) where individuals show ternary interactions and collective order is driven deterministically. We characterize intrinsic noise in the data generated from the simulations and examine the susceptibility of the results to the presence of boundary. In these data-inspired spatial models, we show that the essential features of the group-level dynamics obtained from the simulated time-series do not change because of boundary interactions. Furthermore, our inference of local interactions also remains unaffected by the boundary conditions – at least within the model framework and the context we investigated, which were parameterized to the experimental conditions of our laboratory.