This paper presents a software toolbox used for the behavioral analysis of data collected during interventions with children with ASD. The software employs Lattice Computing (LC)-based data representation to compute behavioral states by processing facial landmark data recorded during the real-world interventions. The correlation of the computed behavioral states at various time instances with the corresponding robot actions, can allow specialists to study the causal effects of the various robot interaction modalities to child behavior. The results from applying the real-world data to the proposed software toolbox demonstrate the capability that the software offers specialists to assess their robot-assisted interventions, with the aim of improving the design of robot-based interventions in order to achieve better therapeutic results.