Household auto ownership serves as a critical input to tour/activity-based travel demand models. So, it is important to build a behaviorally realistic choice model to analyze auto ownership choices. Most households decide how many vehicles to own after discussing and negotiating with all household members. So, the underlying decision making mechanism is group decision making unlike individual-level choices where an individual chooses an alternative that maximizes his/her own utility. So, auto ownership choice models should account for intra-household interactions that occur as part of group decision making. However, while there have been considerable efforts, both in practice and research, to account for intra-household interactions in short term activity participation and travel scheduling decisions, similar efforts are lacking in modeling long & medium term household level decisions such as auto ownership. One of the main reasons for this gap in literature is that the utility functions that account for heterogeneous group decision making mechanisms involve multiplicative sub-utility components. Each of these sub-utility functions has an observed component and an additive unobserved component (which is the error term). So, the probability expression in the resulting likelihood function does not have an elegant closed form like in the case of the multinomial logit model. So, in this study, we will develop a traditional auto ownership choice model which takes the form of ordered response model and compare its performance against a model which accounts for heterogeneous decision making mechanism in which each individual in a household first determines his/her preferred choice and then the household arrives at a compromise through negotiation. The study findings will highlight the importance of accounting for group decision making in household-level long term choices.