Early dismissals in public schools on potential severe weather days

Survey responses from 76 public school districts in Mississippi and Alabama were analyzed along with county-level demographics, tornado-path data, and tornado warning data to investigate the emerging trend of early dismissals. No two districts had identical responses, but the concern of safety was consistently mentioned. The highest number of reported dismissals would have resulted in a loss of >2 % of total annual class time. The rates of dismissals reported by districts were compared with tornado warning days to identify potential over-dismissal, which was found in 51 % of districts. Rates of dismissal and over-dismissal were analyzed with variables from survey responses, tornado warning data, tornado-path data, and county census data to determine variables that influence differences among districts. Various stepwise models indicate that districts with higher poverty levels could be less likely to dismiss. Permutation tests and stepwise models show that early dismissal and over-dismissal are more likely when a district: (1) has a higher number of killer tornadoes in the last 5 years, (2) has a higher number of recent injury-causing tornadoes, (3) has a higher number of total killer tornadoes since 1950, (4) is in a smaller county, (5) has dismissal decisions made by a team, (6) is a county district rather than a city district, and (7) has a lower level of poverty. The inverse relationship with poverty suggests that schools in poorer areas may be holding students in school because they consider them to be safer there than at home, or it could be due to a lack of resources at the school. All of these are based on a district’s location and plans, things that can be determined before a potential dismissal. These indicators can be used to highlight districts most likely to dismiss and possibly reduce rates of over-dismissal by alerting districts to their predisposed tendencies. Copyright Springer Science+Business Media Dordrecht 2014