Recent research (Strunk and Reardon forthcoming) applies Partial Independence Item Response (PIIR) models to teacher bargaining agreements in California to calculate the latent restrictiveness of these contracts. Further research (Strunk and Grissom 2010; Strunk forthcoming) tests the external validity of these estimates. Given that much research on collective bargaining focuses either on high-profile provisions (Moe 2009) or categories of provisions (Koski and Horng 2007), it is important also to assess the internal validity of these estimates: are PIIR estimates of contract restrictiveness robust to analyses on different subsets of provisions? We introduce a new dataset derived from "all" provisions in "all" active CBAs in the state of Washington, and apply the PIIR methodology to calculate estimates of contract restrictiveness using the full range of provisions and various subsets of provisions. We find that estimates calculated from a subset of high-profile provisions are moderately highly correlated with estimates calculated from the full range of provisions, as are estimates calculated from several categories of provisions: Association rights, evaluation procedures, teacher benefit and leave policies, hiring and transfer provisions, and teacher workload agreements. However, estimates calculated using only grievance and layoff policies produce different results, suggesting that estimates using these contract sub-sections may capture another dimension of bargaining and lead to inappropriate inferences on the relationship between union strength and outcomes of substantive interest. Protocol for collecting and coding collective bargaining agreements (CBAs) is appended. (Contains 5 tables and 19 footnotes.)