The decision process behind inclement-weather school closings: a case-study in Maryland, USA

Authors


Abstract

School closings due to inclement weather cause inconvenience for parents, employees, and other affected parties, and such closings have been shown to lower scores on standardized tests. Delaying or closing school is usually a last-minute decision based on current and predicted weather trends. The decision process can also be stressful, as an incorrect decision results in community scorn.

Little formal research has been conducted on the procedures used to make an inclement weather closing or delay decision. This research sought to identify the procedures school administrators follow, including the types of weather data used. Statistical analysis of actual closing dates was also carried out to examine if non-meteorological factors, such as day of the week, play a role. Maryland, a mid-Atlantic state in the eastern United States, was chosen for study because there are only 24 school districts but they span a widely-varied physical and socioeconomic landscape. School closing data were collected and assembled and interviews across school districts of varying size were carried out during the autumn and winter of 2010–2011.

Overall, three major findings emerged from this study. First, the decision-making process was remarkably similar for districts of all sizes and locations, and largely the same for all types of weather. Second, school transportation directors are well-literate, high-level weather consumers in spite of little or no formal meteorological training. Finally, non-meteorological factors play almost no role in the decision making process, though a slight predilection towards Friday closures was observed. Copyright © 2012 Royal Meteorological Society

1. Introduction

Weather-related school closings disrupt learning, daily routines, and schedules for parents, teachers, and administrators. Many times schools are closed for good reason, an intense snowstorm, for example, but school administrators sometimes err, either by closing schools on a day without significant severe weather or by opening schools only to dismiss students immediately due to poor weather conditions. A more complete understanding of this process could reduce both unnecessary closings and openings that should not have occurred.

Prior educational research has shown the benefits of a longer school year (Card and Krueger, 1992; Pischke, 2003). Thus, it is reasonable to assume that weather-related school closings disrupt learning, but only recently has the research community begun to test this hypothesis. Marcotte (2007) found that standardized test scores decreased significantly in heavy snow years in Maryland because of lost instructional time. While schools are generally required to make up days lost to inclement weather or other causes, the dates of standardized tests are consistent regardless of how many days have been lost to inclement weather. Sims (2008) also chronicled the importance of the number of days of instructional time, by noting that some schools in Wisconsin moved start dates earlier to increase test scores. Canadian educational consultant Paul Bennett (Bennett, 2010) argued in a white paper that excessive closings due to inclement weather have a negative impact on schools in Nova Scotia (2010). A question not yet addressed by the literature is whether make-up days confer the same benefits as regularly scheduled school days.

School closings almost certainly have other negative effects, but evidence of these is largely anecdotal. Researchers in this area have noted this fact (e.g., Johnson et al., 2008). It is believed that school closings negatively affect low-income students who rely on free lunch programmes (American Academy of Pediatrics and Trust for America's Health, 2007). While parents are inconvenienced by closings, a recent study also found that surprisingly few (10%) of parents reported ‘serious inconvenience’. However, this study was for a small rural community accustomed to school closings in the winter (Johnson et al., 2008), and the results may not be applicable elsewhere. Cohn (1996) found that school closings do not seem to increase crime. This may be because inclement weather keeps people apart: it would interesting to study whether or not unnecessary closings (those made on days when severe weather does not materialize) have such an effect. While it appears that school closings have negative effects, supporting evidence is largely circumstantial.

Even less is known about how and why school administrators decide to close schools. It is important to understand and evaluate the decision-making process because a good process leads to better decisions (as suggested by Dean and Sharfman (1996) and Harrison (1999)). Insights into the school-closing decision-making process come from a variety of secondary sources such as the news media. For example, administrators evaluate current road conditions (Aney, 2008; Russ, 2008), and decisions must be made early in the morning (Aney, 2008; Fairfax County, 2008). Weather forecasts play a role (Fairfax County, 2008; Russ, 2008), but it is unclear how significant this role is. Additional or specific weather information may help. Eisenhardt (1989), in a study of computer firms, found that good decision makers often sampled more information than poor decision makers. Montz et al. (2011) echoes these sentiments, finding that about three quarters of North Carolina school officials use between four and six information sources to make a school closure decision.

Finally, school closings vary geographically. While peer-reviewed research is lacking in this area, examples of uneven closings abound in the media. For example, in January 1994, several schools near Syracuse, New York, opened and then immediately closed due to heavy snow, drawing ire from parents and the media, while other schools did not open at all. A simple Internet search found many other recent examples where some schools chose to stay open and others closed (Eastern Oregon, February 2008; Philadelphia, February 2008; Cincinnati, March 2008). The present research will help elucidate some factors and reasons that may explain uneven closings.

As mentioned above, a longer school year is beneficial and there appear to be at least some negative consequences of weather-related school closings, making them a topic worthy of study. However, the process behind these closings is largely unknown outside of anecdotal or news media accounts. While it is obvious that weather information must be factored into the process, it is unclear what role it plays or if existing weather forecasts are meeting user needs.

The purpose of this study is to identify the inclement weather closing procedures school administrators have followed during the past decade to ascertain commonalties or differences in the methods employed for determining a closure. In particular, three major questions are investigated: (1) how school administrators acquire, interpret, and us weather data; (2) whether school administrators have other non-meteorological criteria that influence their school closing decision (e.g., number of prior closings or day of the week), and, (3) what specific weather conditions cause the most school closures and delays. Given the lack of information about school closing decisions in the literature, this project is exploratory in nature.

2. Data and methodology

School closing data were collected and interviews were conducted with decision makers in Maryland, a mid-Atlantic U.S. state. Maryland offers a unique physical and socioeconomic landscape for examining the complexities of weather and societal dynamics within a compact area. Despite a relatively small land area, the geography of the state is highly variable. Eastern portions of the state are practically flat while the far west is quite rugged with six peaks above 866 m. Elevation and distance from the ocean have significant effects on the weather, especially snow, dense fog occurrence and tropical cyclone impacts. For instance, coastal regions average about 0.3 m of snow annually, while mountainous regions regularly experience eight times that amount. The cultural geography of Maryland is also very diverse. The eastern portion of the state is largely agricultural, the central portion features dense development in and around Washington and Baltimore, and the western region features farming and settlements in valleys separated by sparsely populated ridges. Maryland is also advantageous in that the number of school districts is small (n = 24) and the school districts follow county boundaries, thus simplifying data collection and spatial aggregation (Figure 1).

Figure 1.

Map of Maryland climate divisions (numbered) and school districts/counties (named). Climate division 8 was excluded from the analysis due to a lack of data

2.1. School closings, delays, and early dismissals

Information about Maryland school district closings, delays and early dismissals was acquired from the school district transportation services or superintendent offices for the academic years August 2000 through June 2010. Assembling data was challenging as each district had to be contacted individually and the office where the data were stored varied from district to district. Non-responding districts were contacted at least three times in an attempt to collect data from as many sources as possible.

In order to compare the school dismissals and corresponding weather conditions directly, data retained for analysis needed to include exact dismissal dates and the weather-related rationale (e.g., snow). School districts with only annual dismissal data and/or days closed for non-weather related events (e.g., electrical fire) were not examined. Of the 24 Maryland school districts, 20 districts had acceptable data for at least part of the study period (Table 1), but only 9 of these (Dorchester, Frederick, Harford, Howard, Montgomery, Queen Anne's, Washington, Wicomico and Worcester) provided closing information for the entire study period as districts are not required to retain such data. Four districts do not archive school dismissal data or were unable/unwilling to provide the data (Baltimore Co., Garrett, Prince George's and Somerset).

Table 1. Maryland school district closings, delays, and early dismissals for academic years August 2000 through June 2010 sorted by climate division
Climate divisionDistrictData availability (years)Closed (%) (n = 690)Opening delay (%) (n = 584)Early dismissal (%) (n = 94)
  1. Percentages are based on the total number of dismissal days (n) for available district data in their respective category (closed, opening delay, and early dismissal). The closed category includes those days the district was partially closed.

1Somerset0N/AN/AN/A
1Wicomico104.814.66.4
1Worcester104.713.76.4
2Caroline74.16.80.0
2Dorchester106.116.39.6
2Talbot54.20.00.0
3Calvert21.90.72.1
3Charles63.71.45.3
3St. Mary's54.22.73.2
4Anne Arundel85.20.00.0
4Prince George's0N/AN/AN/A
5Kent62.97.52.1
5Queen Anne's105.24.36.4
6Baltimore0N/AN/AN/A
6Baltimore city22.30.01.1
6Carroll74.53.96.4
6Cecil95.63.80.0
6Frederick108.85.014.9
6Harford109.15.79.6
6Howard106.84.16.4
6Montgomery106.43.85.3
7Allegany11.50.00.0
7Washington108.05.714.9
8Garrett0N/AN/AN/A

Such large differences in data availability among school districts necessitated spatial aggregation to ensure a more balanced expected frequency distribution and to enable a broader regional analysis. School districts were combined based on climate division because divisions are based on topographical, hydrological and meteorological characteristics of each state and represent a climatologically homogeneous area with spatial and spectral coherence (Guttman and Quayle, 1996). In addition, Maryland's climate division boundaries are coincident with the boundaries of the counties/school districts (Figure 1). Furthermore, each climate division contains two or more districts, except division eight. This division, which only contains one school district, Garrett, was excluded from further analysis due to a lack of data.

The school district closing dataset contained largely categorical information on the dismissal day timing and reason for closure. A chi-square (χ2) goodness-of-fit analysis was conducted on the school closing data to determine if a significant difference existed in the set of observed frequencies from the theoretical or expected distribution. A large chi-square value indicates the goodness-of-fit between the observed and expected frequency distributions is weak and that the null hypothesis (i.e., no differences in the distributions) should be rejected. In this case, the alternative hypothesis is accepted (p < 0.05), the observed frequency distribution is not random. The school data were divided into two major groups, all-day closures and partial closures (either an opening delay or early dismissal). School holidays and the academic year were considered when comparing the monthly frequencies by reducing the expected closure frequencies for the months of April, June, August, November and December by 1–2 weeks each. July was not used in the analysis. In particular, data were examined for any certain months, days of the week, or weather closure type (e.g., ice) with significantly higher or lower closure frequencies among the climate divisions. It was hypothesized that schools were more likely to close during the winter (December to February) due to snow-related conditions, the most common weather closure reason, and on Mondays and Fridays to allow for longer weekends and recovery time.

2.2. School administrator interviews

The authors conducted interviews to collect data about the school closing decision-making process, implementation procedures and use of weather information. Contact information about potential participants was obtained from school web sites or simply by contacting the main school switchboard. Those willing to participate engaged in a telephone interview of approximately 20 min. A script with the following eight general questions guided the interview but the discussion was allowed to wander to related topics that stemmed from the responses.

  • What is your title and experience in current position and overall in education?

  • Who makes the decision to close schools? What people assist in this decision and what are their roles?

  • What steps are involved in this decision process? What time is it performed? How is this information disseminated internally and externally?

  • What specific weather information is used? What sources does this information come from? Does this vary with the type of inclement weather? What criteria are most important?

  • What non-meteorological information is used? How is it obtained?

  • Are there any other criteria involved in the decision making process (e.g., number of prior closings, state mandated test dates, time to previous/next scheduled off days)?

  • Are the current weather sources/types adequate for making the decision? If not, what other sources/types/data would help?

  • Do you have any other comments (e.g., specific examples)?

Ten interviews were conducted during the winter of 2010–2011. In order to limit questioner bias, all interviews were conducted by the same author (Call). At least one interview was done in each of the seven climate divisions under study, and both large and small school districts participated. In all cases the interview was with conducted with the ‘Director of Transportation’ or similar official who makes a school closing recommendation, but in a few instances another official involved in the decision-making process participated. Although notes were taken, all interviews were also recorded and later transcribed.

3. Closing results

The closing and delay/dismissal dataset contains 1464 entries. Approximately 47.1% of the total entries (n = 690) correspond to days when an entire district or a portion of a district was closed. Closing events occurred on 149 distinct days, which makes sense as weather phenomena significant enough to close schools generally affect multiple districts. Harford County has the most school closing entries (9.1%), followed by Frederick (8.8%) and Washington (8.0%). All three districts lie in interior northern Maryland and have a complete 10 years of record (Table 1). The remaining 52.9% of total entries consist of the following: 1 or 2 h opening delays (n = 584); 1, 2, or 3 h early dismissals (n = 94); and a shortened school day with unknown timing on opening delays and/or early dismissals (n = 96) that were not considered in the analysis. As with school district closing days, Frederick and Washington districts have the highest percentage of early dismissals. In contrast, delayed school openings are more typical in the southeastern coastal districts of Worcester, Wicomico and Dorchester.

School administrators identified the meteorological reason for a school closure, or other disruption, with the following common descriptors: cold; flood; fog; heat; hurricane; ice; mixed (includes snow, ice, and/or sleet conditions); snow; storm (thunderstorm or severe weather event); tornado; wind, or unknown weather-related school closure. Schools closures are predominantly from winter-related precipitation (snow, ice or mixed) events, accounting for 90.6% of all closed days across all divisions (Table 2) and 81.9% of all early dismissals (Table 3). Delays are often caused by winter precipitation, but fog is also quite common, accounting for about half of all school delays (51.0%). Fog is particularly problematic in climate divisions 1, 2, and 5. These divisions are all located on the Delmarva Peninsula between Chesapeake Bay and the Atlantic Ocean, which makes them prone to advection-type fog conditions (Table 4). Although radiation (or valley) fog is prevalent in western and interior Maryland, fog does not cause opening delays in those areas.

Table 2. School closures by meteorological cause and climate division expressed as a percentage of total number of all-day closures rounded to the nearest tenth (n = 690)
ReasonClimate divisionTotal
1234567
Cold0.00.00.00.00.00.10.00.1
Flood0.10.00.00.00.00.10.00.3
Fog0.10.40.00.00.00.00.00.6
Heat0.00.00.00.00.01.00.01.0
Hurricane0.60.90.90.10.32.60.15.5
Ice0.31.60.70.00.64.20.78.1
Mixed0.31.70.60.00.13.81.98.4
Snow8.09.67.15.17.130.66.774.1
Storm0.00.00.00.00.00.40.00.4
Tornado0.00.10.30.00.00.00.00.4
Unknown0.00.00.30.00.00.00.00.3
Wind0.00.00.00.00.00.70.00.7
Total9.414.39.95.28.143.69.4∼100.0
Table 3. School early dismissals by meteorological cause and climate division expressed as a percentage of total number of early dismissal closures rounded to the nearest tenth (n = 94)
ReasonClimate divisionTotal
1234567
Cold0.00.00.00.00.00.10.00.0
Flood0.00.00.00.00.00.01.11.1
Fog1.10.00.00.00.00.00.01.1
Heat2.10.00.00.00.04.30.06.4
Hurricane0.00.00.00.02.11.12.15.3
Ice1.12.11.10.00.01.11.16.4
Mixed2.10.01.10.01.113.82.120.2
Snow7.47.45.30.05.321.38.555.3
Storm0.00.00.00.00.00.00.00.0
Tornado0.00.00.00.00.01.10.01.1
Unknown0.00.03.20.00.00.00.03.2
Wind0.00.00.00.00.00.00.00.0
Total13.89.610.60.08.542.614.9∼100.0
Table 4. School delays by meteorological cause and climate division expressed as a percentage of total number of delays rounded to the nearest tenth (n = 584)
ReasonClimate divisionTotal
1234567
Cold0.00.00.00.00.00.00.00.0
Flood0.00.10.10.00.00.30.10.7
Fog17.718.60.00.014.50.10.051.0
Heat0.00.00.00.00.00.00.00.0
Hurricane0.00.00.00.00.00.10.00.1
Ice2.51.30.30.02.16.91.214.3
Mixed1.60.60.70.00.93.11.08.0
Snow2.51.91.60.04.411.72.124.2
Storm0.00.00.00.00.00.10.00.1
Tornado0.00.00.00.00.00.00.00.0
Unknown0.00.01.20.00.00.00.01.2
Wind0.00.00.00.00.00.00.10.1
Total24.422.64.10.021.922.54.6∼100.0

Chi-square analysis reveals the day of the week is an important factor for some school district all-day closures (Table 5), but not for shortened school days. Based on all climate divisions, schools are closed significantly (p < 0.01) more frequently on Fridays (based on higher residuals or differences between the observed and expected frequencies) than any other day of the week. However, these results vary when disaggregated. Friday closures are more likely for climate divisions 5, 6 and 7 (the central interior), with only climate division 6 being statistically significant (p < 0.01). Closures are least likely on Wednesdays for climate division 1 and Thursdays for climate division 6, significant at the 90 and 99% confidence level, respectively. In contrast, partially closed school days do not exhibit a propensity for more or less delays/dismissals on any given week day for any climate division (not reported). Collectively, the districts exhibit a slight indication of Thursdays being more favourable for partial day closures and Mondays and Fridays less so, but these results are not statistically significant (p = 0.16).

Table 5. Chi-square (χ2) results and associated probability values (p-value) examining day of the week school closed by climate division
Climate divisionX2p-value
  1. P-values significant at the 90% (*), 95% (**) and 99% (***) confidence levels are denoted by asterisks.

18.460.08*
23.480.48
32.150.71
42.890.57
56.680.15
619.480.00***
77.540.11
All divisions29.050.00***

In addition to day of the week, the month the closure occurred in was also examined using a chi-square analysis to determine if closures are more or less prevalent during certain months of the year. As expected, the chi-square results (χ2 = 1243.5, p < 0.01) using all divisions show all-day closures are most common during the winter from December through February, with February having over four times more closures than the expected frequency. December and January are twice more likely to be closed, whereas all other non-winter months are much lower. Additional analysis of the partially school closed days largely confirms these results, except that January and February equally have the highest partial closure frequency and are twice as likely to have a late start or early dismissal (χ2 = 311.5, p < 0.01). A separation of the partial school dataset into early dismissal and delayed start did not significantly alter these findings.

4. Interview results

Ten subjects volunteered to respond to questions about school closing decision-making, including the use of weather information. At least one participant was located in each climate division, except region 8 (Garrett). At least three large, medium-sized, and small districts were included in the sample. A large district was defined one of the eight districts with the greatest enrollment (minimum 200 000 students), a medium district was in the next group of eight (61 000–200 000) and a small district had fewer than 61 000 students. Results in this section will reference districts as being in one of those three groups to protect respondent confidentiality.

The first section of the interview focused on background information about the respondents. Nearly all respondents carried the title ‘Director of Transportation’ (DoT) or a synonymous title. All participants were responsible for administration of the bus system. In smaller districts DoTs often supervised other operational tasks as well. Most respondents had 5–10 years of experience in their current position. Participants reported a wide variety of life experiences though most had at least some experience as either a teacher or an employee of a transportation-related company.

4.1. Decision-making process

The decision-making process was similar in all sizes of districts. Beginning around 0300 h LST (often later in smaller districts), DoTs and their employees drive on roads in the county to check the conditions. In the small districts, the director may follow a prescribed route or check ‘trouble spots’ themselves. In larger districts the director often does minimal driving; instead, they remain in the office and receive reports from staff. Directors also collect information from other people with first hand knowledge of road and weather conditions. All directors call police agencies (state and/or local), road departments (state or county), and neighbouring school districts. Several DoTs in small districts also spoke with local residents such as retired teachers or farmers located in strategic spots (e.g., crest of a hill, near a river likely to fog over). Few speak directly with meteorologists. Three have contracts with private forecasting companies that allow them to consult with a meteorologist: one participant sometimes calls the National Weather Service (NWS).

Following the information-gathering phase, most DoTs issue a recommendation to senior administrators, who then make a final decision. In 7 of the 10 districts surveyed, the recommendation goes to the superintendent, while in 2 others (1 large, 1 medium) it is sent to the Chief Operating Officer (COO), a high-level administrator who often oversees most daily operations such as facilities, finance, food service, and transportation (COO duties vary from district to district: smaller ones may not have a COO at all). In one of the smallest districts interviewed, the transportation director (who also oversees other operations) makes the decision themselves. (It is worth noting that this director has more than 15 years of experience in that position, the longest of any person interviewed.) Timing of the recommendation varies with district size. In the largest districts, the DoT makes the recommendation around 0330 or 0400 h LST while in medium-sized and smaller districts, the recommendation is made closer to 0430 h or even as late as 0500 h LST (two cases). Larger districts have earlier deadlines because commute times are longer and they have more cross-district bussing to specialized schools.

Regardless of when the recommendation is sent, the superintendent almost always makes a final decision within half an hour of receiving it. At this point, notifications are sent out to parents, employees and the media. In large districts communications staff handle this responsibility, while in smaller districts the transportation staff are often responsible for this. Most districts telephone the media and use automated phone-dialling systems to notify staff and parents. Many also post the information on the school web site. A few use additional services such as ‘School's Out’, a notification service started in Frederick County, Maryland. Whenever a school is closed or disrupted for any reason this service posts the information on their web site and e-mails subscribers. It also offers a paid subscription option for those who would like text messages sent to their cellular telephones.

School transportation directors stressed that their decisions were solely based on safety and not ‘outside’ criteria such as the number of dates closed already, state-mandated test dates, or other non-safety related items. When pressed further, a few admitted that these issues may occasionally influence whether or not the superintendent or COO adopts the recommendation of the DoT. Outside criteria are most likely to be invoked in a ‘borderline’ situation where the best closing decision is unclear. Two districts (located in different areas of the state) also mentioned that they are more cautious if it the first storm of the season out of fears that motorists would be less adept at winter driving. A few districts also remarked that they may have to close if there is not enough time to clear buses and shovel walkways, most likely in a case where heavy snow falls overnight. Schools may also need to close if schools are being used as shelters or there are widespread power problems (probably following a hurricane or ice storm).

4.2. Weather information usage

Weather information, naturally, plays a significant role in the decision-making process. The 10 DoTs surveyed were savvy weather ‘omnivores’ who sampled a wide range of weather information from multiple sources. This section will detail the types and sources of the information used.

Three of the 10 DoTs subscribed to private forecasting services. Even so, these directors sampled as much information as the other participants. Most directors reported use of various weather websites such as Accuweather.com, Weather.com, and Weather.gov. Many also monitor The Weather Channel (cable station focused on United States weather) and local television and radio stations.

Most directors examine online radar and temperature data. Nearly all view online forecasts from multiple sources. Often, they aggregate the forecasts to try to establish a consensus forecast. Directors located on the Delmarva Peninsula, where fog sometimes causes school closings or delays, also look at visibility information and dew points. One director in this region even uses dew point depression to determine fog potential.

As mentioned earlier, many directors also incorporate non-meteorological information such as reports from adjacent counties, police and road agencies, and various individuals. One western district reported the use of webcams to see conditions in other portions of the county.

Most transportation directors were highly satisfied with the weather information they received. This was true for those that subscribed to private services as well as those that relied on the National Weather Service (NWS). When asked what other information they would like, one director on the Eastern Shore requested a fog forecast. Another requested hour-by-hour snow accumulation amounts. Both of these districts are medium-sized districts that do not subscribe to private forecasting services.

Several directors also provided more general feedback about forecasts. A director in the western region complained that the NWS, The Weather Channel, and other meteorologists use too much ‘doom and gloom’ which ‘crazes’ the public. This director was annoyed about complaints from parents when school was held on days with winter storm warnings because this director feels that their district can handle the amount of snow that is associated with such a warning. A director in a large district remarked that societal pressure to close has increased over the past dozen years or so. This director remarked that not closing (when schools should be closed) is more likely to generate complaints than closing unnecessarily, though both errors generate complaints. They also added that sometimes employees are upset when school is open on a day that they expected schools to close. Finally, another director noted that forecasting improvements have led to the public asking whether or not they plan to close several days in advance, much earlier inquiries than in the past; this director often responds that they do not yet know.

4.3. Zones

Maryland schools are organized by county, and weather conditions can vary considerably within a county-sized area. Four counties in the state have more than 450 m in elevation variation and an additional five counties have between 250 and 450 m in variation. These variations can result in dramatic weather differences within a county. As a result, several counties have adopted ‘zones’ where certain portions of the county are closed or delayed for school while others remain open on a normal schedule. While most transportation directors have considered zones, few actually use a zone system. Based on the present dataset, partial district closures using a zoning system are most prevalent in Frederick, Washington and Harford districts. However, the partial district closures constitute less than 3% of all closures. Logistical issues are generally cited in arguing against zones. Large districts are likely to have cross-county bussing to magnet or other specialized schools. Very small districts may only have one or two high schools at all. Several DoTs acknowledged the frustration that parents in fair weather areas must feel when the entire district is closed, but they feel that zoned-closings would unnecessarily complicate matters.

5. Discussion and conclusions

School closures are most common during the winter months (December to February), particularly February, due to winter weather conditions (snow, ice, sleet). These months coincide with the Maryland snow season. However, the February peak may be an artefact of extraordinarily snowy winters during the last 2 years of the dataset. During the 2009–2010 school year, most districts closed for at least 10 days in February alone after two closely-timed storms brought over 75 cm of snow to the Mid-Atlantic region. Although snow and sleet related conditions account for the vast majority of closures and early dismissals across the state, school openings in coastal districts are more likely to be delayed by fog, especially during the fall.

Several themes emerged from the interviews. First, the process varied little from district to district and event to event. While districts on the Eastern Shore are concerned about fog and districts in southern Maryland close for smaller amounts of snow compared to their counterparts elsewhere, the decision-making process is similar regardless of event and geographic location. Some minor differences in procedure did occur between larger and smaller districts. In larger districts the decision process was carried out earlier, and these districts had separate communications staff to notify interested parties of the closing/delay decision. Another significant finding of this research was that DoTs are highly literate weather consumers. Their knowledge comes from experience as none were former science teachers or described any sort of meteorological training. Finally, the third significant finding as that non-meteorological factors appear to play little to no role in the decision making process at the DoT level. Directors of Transportation are focused on student (and to a lesser extent, employee) safety and this directs their closing and delay recommendations. That said, the use of interviews may have resulted in participants providing ‘desired’ answers instead of correct ones, so non-meteorological factors may have a subconscious or even silent, illicit influence. Additionally, it was found that Fridays are more likely to have a closure than any other day of the week, so it is possible that non-meteorological factors (e.g., longer weekend) may come into play when the superintendent takes action based on the director recommendation.

The authors believe the findings of this study could be applied to other places with comparable weather and similar school organization. Procedures varied little from district to district, and anecdotal evidence suggests that DoTs in other cold-weather states follow similar procedures. However, different procedures are more likely to occur in places where closings due to inclement weather are rare or where average snowfall totals are much higher. Comparisons of procedures between different countries, may also yield some benefits.

The depth of findings and analysis were sharply limited by incomplete closing data. Although most districts provided at least some data, many districts had only a few years of data and/or provided annual data that lacked specific dates and reasons. This issue, when coupled with ‘data holes’ created by non-responding districts, made it difficult to carry out more than a rudimentary statistical analysis of closing patterns and reasons. Future researchers may wish to conduct the research in a state where school closing records are maintained by a central authority, such as a board of education. More detailed, consistent, records would be invaluable for examining items such as decision error rates for different events or districts.

Another limitation was the use of interviews. Participants were asked to summarize general procedure from multiple decisions. As a result, finer details, such as the web sites participants use, tended to be left out, and ‘desired’ instead of accurate answers may have been provided. The authors tried to minimize these potential errors by conducting interviews during the closing season and by conducting 10 interviews. Nonetheless, a project where researchers shadowed decision makers through their process would provide more specific details and add bulk to our conclusions (see Klein et al., 1989) for a suggested methodology). This would also allow researchers to examine how accurately DoTs interpret and use weather information.

Much of the motivation for this study was an interest in helping meteorologists better communicate weather information to school DoTs and others who make inclement weather closing decisions. The DoTs provided little in the way of guidance for the meteorological community. They seem content with current meteorological observations and forecasts, and the few that requested more specific data (such as hour by hour snow accumulation amounts) may benefit from contracts with the private sector. From the perspective of meeting user needs, no DoT provided explicit criteria used in deciding to close school, such as specific snow accumulation amounts. Rather, it seems that the school closing decision is holistic in nature, combining current weather observations, trends and forecasts with actual ‘ground truth’ conditions such as accumulation on pavement. Instead of closing (or opening) school simply based on snow amounts, DoTs also consider storm timing, visibility, temperature, condition of walkways and other factors. This echoes other findings that suggest that the impact of snow storms is affected by many other items besides accumulation amounts (Call, 2005).

Finally, it is also worth noting that the interview participants were highly-versed in interpreting weather products despite a lack of formal training. A deeper understanding of how they obtained this knowledge could help the meteorological community better educate the general public, improving inclement weather response, protecting property and saving lives.

Ancillary