However, it cannot be excluded that at least part of the difference in durability is caused by nuanced considerations that the model cannot understand. For example, it is possible that within a given object, Senatorial attention is selectively focused on certain types of agreements, for example. B to agreements of particular importance, and that this selection is also correlated with sustainability. The categories of topics included may be too crude to capture this dynamic, and it is possible that a more detailed measurement of the object will provide a more complete and complete understanding of the connection between the content of an agreement and the choice of engagement apparatus, which may explain at least some of the observed difference. Footnote 104 Finally, the dataset does not contain any information on which party is ultimately responsible for the termination of the agreement. All of the theories summarized above focus on the reliability of the United States as a negotiating partner. Agreements that have been denounced for reasons that have nothing to do with the US commitment should therefore not be taken into account in the analysis. Footnote 85 At the same time, the identity of the party responsible for terminating the contract is not observable unless the researcher analyzes each termination individually. Even then, identifying who is responsible is often a subjective assessment. However, the failure to observe responsibility for the breakdown of the agreement is unlikely to result in significant distortions in the estimates from the data. Bias is only introduced if an unobservable variable is related to both the most interested variables and the outcome variables. In this regard, this means that bias is introduced only if the likelihood of the other party breaking the agreement differs between treaties and agreements between Congress and the executive.
However, as noted above, only the reliability of the United States is bound by the distinction between contracts and executive agreements. It would therefore be surprising if a partner country`s propensity to withdraw were linked to the choice of political instrument. Footnote 87 In each model specification, the coefficient of the contractual indicator is negative and very different from zero. Note that coefficients in survival models change the probability of an event, i.e. The danger rate, express .. . . .