Equity concerns of urban planners and policy-makers cannot be addressed unless disability effects on daily activities are disentangled. The findings, however, strongly depend on how disability is incorporated into the model. Two MDCEV models for analyzing disability effects on daily activity type and duration are discussed and compared in this paper. In the "classic" approach, an independent dummy variable is used to distinguish disability. While, in the "separate" approach, the dataset is divided into disabled and nondisabled groups, and then a separate model is calibrated for disabled group. The two approaches result in different coefficients and elasticity values, evidencing that model specification matters for policy assessments. Three transferability metrics are adopted to evidence that the separate approach outperforms the classic approach in explaining travel patterns of persons with disabilities. Finally, three policies that have been practiced across the globe to prevent social exclusion of disabled people are discussed in terms of the effects of model specification on the policy assessment outcomes. This assessment offers managerial insights for policy-makers to develop appropriate infrastructure and accessibility strategies for disabled people.