“BAD” Days Are “Bad” Only Superstitiously?! An Examination of the Relationship between Mortality and

  • 研討會日期 : 2002-12-10
  • 時間 : 15:00
  • 主講人 : 黃維喬教授
  • 地點 : B棟110室
  • 演講者簡介 : 黃維喬教授為Ph.D. in Economics, University of California, Santa Barbara (1984)。 現為Professor,Western Michigan University。 其主要研究領域為Labor Economics/Applied Microeconomics、Applied Econometrics及Economic Development。
  • 演講摘要 : There is a view that the 4th and the 13th days of a month are bad and therefore more deaths are likely to occur on these days. The main objective of this study is to examine the temporal pattern of mortality so as to evaluate if mortality is higher than expected on the 4th and the 13th days of a month as is superstitiously believed by certain groups of a society. Based on a sample of sufficiently large longitudinal mortality data from Taiwan, we failed to find statistical evidence that corroborates the presence of significantly higher than average mortality rates on these dates,. However, we observed consistently significant deviations between observed and expected mortality rates on eight other days (the 1st, 2nd, 5th, 7th, 8th, 11th, 15th and 21st) in a month. We also found significantly more than average deaths on Tuesdays and Mondays; and in the months of January, February, March and December. Although the differences between observed and expected mortality rates in these months could be attributed to a seasonal effect, we were not able to single out why deaths on Tuesdays and Mondays in Taiwan are higher than expected. Statistical tests on the association between attributes (such as the cause of death, gender, occupation, place of death, age group, marital status etc) and death patterns on the 4th and 13th days of a month were also conducted using contingency table analysis. The tests did not show significant association between most of the covariates and the reference days, except for gender and employment status and the death pattern on the 13th day of a month. While certain mortality patterns exhibited in the Taiwanese data were also observed in the State of Michigan (USA) data, in most cases the Michigan mortality data file reveals that several of the variations in the mortality pattern across monthly dates and the covariates of the subjects are just random noises. A series of Canonical discriminant functions were used to explore how certain covariates are related with the likelihood of individual deaths on these calendar days. The results indicate that age, date of birth, cause and place of death are factors that pose maximum separation between groups of people who die on calendar days with higher than expected deaths and those who die on calendar days with average deaths. Compared with economically inactive and elderly (65+) persons, results from the Logistic regression estimates indicate that economically active persons involved in “manual-labor” activities, and children and teenagers alike have higher propensity of death in months characterized by more than average mortality rates. The likelihood to die on calendar dates with higher than expected deaths is lower among those that have family; and those who die in hospitals and health centers as opposed to those without a family at the time of their death, and those who die in places other than hospitals.