Predicting accident susceptibility: a logistic regression analysis of underground coal mine workers

The history of research into individual liability to accidents is long and troubled (Iverson and Erwin1997). Some researchers studied the distribution and risk of accidents, and concluded that various groups of individuals displayed differences in susceptibility to accidents (Peters1989; Bhattacherjee1991). Other investigators sought to characterize the accident-prone individuals either by comparing the individual characteristics of workers involved and non-involved in accidents, or by evaluating the degree of association between individual characteristics and accident involvement (Shaw and Sichel1971, Hansen1988, Dhalback1991; Dhar et al.1997). However, studies on accident proneness have attracted a significant amount of controversy. Debate has occurred over the effects of various personal factors towards accident occurrences (Shaw and Sichel1971; Guilford1973). Equally important has been the criticism of their causal attributions which researchers make frequently on the basis of bivariate relationships, without controlling for the effect of work environment (Hadden et al.1964; McKenna1983; Hansen1989; Sutherland and Cooper1991). According to Iverson and Erwin1997, accident researchers should broaden the study of individual characteristics to include the work-environmental factors which may affect the risk of injury. The literature on mine accident causation revealed several personal and work-environmental factors affecting the occurrences of injuries to underground coal mine workers (Root1981; Bennett1982; Bennett and Passmore1986; Peters and Schaffer1986; Phiri1989; Kenny1993). The interesting notion about these investigations is that the variables studied were generally common in most of the cases which included: miner’s age, experience, occupation, workplace location and mine. The early research studies on the quantitative analyses of accidents/injuries categorized these variables into suitable groups/classes and concluded about the differences in accident susceptibility among different groups of workers through classification tables and aggregate statistics (Pfleider and Krug1973; Zebetakis and Zalar1982). Later on studies were conducted to assess the relationships of mine and miners’ characteristics with severity of injuries (Bennett1982; Bennett and Passmore1986; Phiri1989; Maiti et al.1999). These studies clearly demonstrated the differences in accident/injury susceptibility among different groups/classes of workers. However, they lacked in two aspects. First, the majority of these studies included only the injured miners into their analysis ignoring the effect of uninjured miners’ population. Second, some studies although they considered the uninjured miners’ distribution, however, they addressed only their bivariate relationships. To address the issues of accident proneness taking into consideration the injured and uninjured workers population, Maiti and Bhattacherjee1999 investigated the risk of Predicting accident susceptibility: a logistic regression analysis of underground coal mine workers