Event History Modelling of Spatial Point Patterns: Issues Regarding Interval definition, Censoring and Explanatory variables

Abstract

<p>This thesis attempts to further the research by Odland and Ellis (1992) in applying event history models to the analysis of spatial point patterns (i.e., event patterns). Its empirical focus is the event pattern derived from the adoption of an agricultural innovation, the Harvestore, in southern Ontario, Canada from 1963 to 1986.</p> <p>Event history analysis involves the use ofdiscrete-state, continuous-time stochastic models to investigate a temporal longitudinal record on discrete variables (event history data). Event history models are primarily concerned with durations of time between events and the effects of intertemporal time dependencies on future event occurrences.</p> <p>Many of the methods used in event history analysis do not preclude the use of other non-negative interval measurements in place of standard temporal intervals to investigate a series of events. In particular, spatial intervals (durations) of distances between points (events) may also be accommodated by event history models.</p> <p>This thesis is methodological in nature, and extends the previous research of Odland and Ellis (1992) by using a wider range of parametric models to explore duration dependence, investigating the role of spatial censoring, and using a more extensive set of explanatory variables. In addition, simulation experiments and graphical tests are used to evaluate the empirical event pattern against one generated from Complete Spatial Randomness.</p> <p>Results indicate that the event pattern formed by the Harvestore adopter farms is clustered (i.e., is described by positive duration dependency). Also, the sales agent is found to be a significant factor in the distribution of Harvestore adopters. In addition, contrasting results obtained from the analysis using censored data versus uncensored data (traditional nearest neighbours) underscores the importance of considering edge effects when using nearest neighbour durations. It appears that an event history approach is a valuable methodology that provides insight into spatial point patterns and processes.</p>