| Abstract: | King, Keohane, and Verba (1994, pp. 27-28) argue that we should "maximize the validity of our measurement," "ensure that data-collection methods are reliable," and make all data and analysis replicable. In an effort to improve the measurement of the events data collected by the Intranational Political Interactions (IPI) project, this extension of the project produces two new valid and reliable interval-like scales. Following Azar (1982), Goldstein (1992), and Moore and Lindstrom (1996), I produce interval-like scales of cooperative and hostile political actions based on a group of experts' judgements. The collective scaling procedure produces data suitable for use in OLS regression models as well as a standardized interval-like scale that more accurately represents the true scores of event types. The paper discusses the procedures taken to derive the new measures, proceeds to argue why the new measures are improvements over existing measures, and reports the findings of statistical analytic comparisons. The statistical comparisons demonstrate that the new scales make a difference in various statistical models using different temporal units of aggregation. |
