Pergamon 0021-9630(95 )00140-9 J Child Psychiil. Psychiat. Vol. 37. No, 4, pp 135-444, 19% Gl&cvier Science Lid O 199(j Association for Child Psychology and Psychiatry Pntited in Oieal Bnlain. All nghls reserved 002t-Q630/96 115.00 + 0.00 Disaster Related Stresses, Depressive Signs and Symptoms, and Suicidal Ideation Among a Multi-Racial/Ethnic Sample of Adolescents: A Longitudinal Analysis George J. Warheit University of Miami, Cora! Gables, Florida, U.S.A. Rick S. Zimmerman University of Kentucky, Lexington, U.S.A. Elizabeth L. Khoury University of Miami, Coral Gables, Florida, U.S.A. William A. Vega University of California, Berkeley, U.S.A. Andres G. Gil University of Connecticut, Storrs, U.S.A. Longitudinal findings are presented on the relationships between disaster related stresses, depression scores, and suicidal ideation among a multi-racial/ethnic sample of adolescents (A'r: 4,978) all of whom have been exposed to Hurricane Andrew. Regression analysis showed that being female, hurricane generated stresses, low levels of family support, pre- hurricane suicidal ideation, and post-hurricane depression scores were significant predictors of post-hurricane suicidal ideation. Path analysis revealed that being female, low socioeconomic status, pre- and post-hurricane depression, high stress scores, low family support, and pre-hurricane suicidal ideation had significant direct/indirect effects on post-hurricane suicidal ideation. Copyright ir; 1996 Association for Child Psychology and Psychiatry. Keywords: Disaster, stress, depression, suicide Abbreviations: SES: Socioeconomic status; WNH: white non-Hispanic; FACES: family cohesion evaluation scales Introduction An extensive body of research findings report positive relationships between large scale natural disasters and their deleterious physical and mental health sequelae among adults. Summaries of these relationships have been presented by a number of researchers/scholars (Lystad, 1984, 1988; Benedek, 1985; Bolin, 1985; Warheit, 1988; Green, 1991). The literature on disasters and their consequences among children and adolescents is far less extensive; however, it has expanded in recent years and the results have generally paralleled those reported for older people (Burke et al., 1986; Fairley et al., 1986; Belter et al., 1991; Green et al., 1991; Kozlovskaia et al., 1991; Sullivan et al., 1991; Hardin et al., 1992). Although there is a general consensus regarding disasters and their adverse effects on impacted popula- tions, prior research has shared a number of methodo- Requests for reprints to: G. J. Warheit, Department of Sociology, University of Miami, Coral Gables, FL 33124, U.S.A. logical, theoretical, and analytical problems. At a methodological level, one of the most confounding factors has been the absence of pre-disaster base-line data on the populations being studied. This has been true for research which utilized cross-sectional designs; it has also been the case for the few studies which have had access to longitudinal data (Powell & Penick, 1983; Drabek & Key, 1984; Murphy, 1989; Green et al., 1990; Christian, 1992). Previous research on disasters and their psychosocial consequences has also suffered from the absence of clearly defined theoretical models and research hypotheses. With few exceptions (Baum et al., 1981; Fleming et al., 1984; Green, 1982, 1990), most researchers have tended to assume that the disaster being studied produced a number of stressful events and, further, that the physical and mental health problems identified in post-disaster populations were attributable to these events. Additional limitations have included a reliance on univariate and descriptive analyses and many studies did not include controls for important variables such as socioeconomic status (SES) race, ethnicity, and gender. The serendipitous occurrence of Hurricane Andrew between the second and third data collection periods of a 435