With the objective of applying Laser-Assisted Atom Probe Tomography to compositional analysis within nanoscale InGaAs devices, experimental conditions that may provide an accurate composition estimate were sought by extensively studying an InGaAs blanket film.Overall, the determined Arsenic concentration was found to exhibit an electric field dependent deficiency, which was more pronounced at low field conditions, whereas, the determined group III site-fraction remained more or less independent of analysis conditions and in close agreement with the nominal value. In this study, we investigate and discuss the mechanisms that could potentially contribute to As underestimation. Given the field dependence observed, the phenomena occurring between low and high field conditions are compared. At low field, the tendency of As to field evaporate in significant amounts as multiply charged cluster ions (Asn i+ with n as large as 9 and i=1,2,3) is shown to be a significant source of compositional inaccuracy. These clusters may lead to peak overlap in the mass spectrum (e.g. the peak at 150 Da may represent As4 2+ or As2 + or both), thereby creating an uncertainty in the quantification. Emitted clusters may also dissociate with the likelihood of neutral generation and multi-hit losses being non-negligible. Experimental studies and density functional theory calculations are presented to characterize cluster stability and its contribution to measurement uncertainty. Under high field conditions, although fewer clusters are detected and the composition appears more accurate, the emergence of two additional mechanisms, i.e., multi-hits and DC evaporation, may degrade the data quality. The challenges in evaluating the impact of all these loss mechanisms are examined in detail. Finally, we show that for InGaAs under UV illumination, due to the lasertip interaction, the resulting asymmetric electric field distribution across the apex introduces local concentration variations.