They follow strict procedures and use objective evidence to identify health violations. Thus, their project appears to be as immune to human error as any regulatory project could hope to be. However, hundreds of thousands of potential violations go unreported each year, putting people at additional risk for foodborne illness.
While no inspection process (or inspector) is perfect — are there some common factors that could compromise the quality of inspectors’ work?
Maria Ibanez, assistant professor of business at Kellogg, hoped that behavioral science might shed some light on the question. Ibanez’s research focuses on how work scheduling affects decision-making, so he naturally wondered whether the order in which inspectors make their site visits affects how many violations they report.
“I think it’s very important to understand how people are affected by the sequence of tasks they have to complete at work. This gives us an opportunity to improve performance by being smarter about planning,” he says.
She too Michael Tofel of Harvard Business School analyzed thousands of health inspection reports from restaurants, schools, hospital cafeterias, and other places where food is handled and found that the order in which inspectors conducted their work biased the quality of the work in two ways.
First, they found that if an inspection did not go relatively poorly, with the inspector reporting a relatively high number of violations at one site, that inspector would then conduct more than usual inspection at his next site and subsequently report relatively more violations. However, they also found that if the timing of an inspection threatened to extend an inspector’s typical workday, that inspector would report fewer violations than usual.
Ibanez interviewed and even shadowed several inspectors throughout their day in order to better understand how they do their jobs.
“They are very dedicated and careful about following procedures,” he says. “But in the end, they’re still human. Their behavior will be influenced by their emotions and psychology.”
Biases in Food Safety Inspections
Ibanez and Toffel received more than 12,000 inspection reports from Hazel Analytics, which aggregates them from local governments across the US. This detailed data set allows researchers to track the reporting behavior of individual inspectors as they move from site to site. For each inspector and location, they were able to calculate a baseline percentage of violations reported, as well as note any significant deviations from that baseline.
“This data is extremely rich—we can observe both the inspectors and the facilities they visit over time,” says Ibanez. This richness allowed the researchers to rule out alternative explanations for their findings.
Their analyzes revealed that inspectors were operating under two different cognitive biases.
Previous Surveyors Website Topics
First, inspectors’ attitudes about the site they are currently inspecting are influenced by what happened at the previous (unrelated) inspection site. Inspectors, the researchers found, bring additional scrutiny when that past experience was unusually negative. In cases like this, when an inspector came from a site with a significant number of violations above their previous rate, the same inspector increased the number of violations reported at the next site by an average of 8 percent. They call this “results”.
However, the reverse was not true. There was no effect on violations reported when the previous inspection went unusually well.
“Negative events are generally more intense, dominant, and have stronger effects than positive events,” explains Ibanez. “If an inspector is shocked by a negative trend at one site, they will remember that experience and struggle at the next place they visit.”
Food Safety Inspectors Daily Program Topics
However, the second type of bias behavior the researchers observed reduced reporting. There were two components to this finding: conducting an inspection that could delay the end of an inspector’s workday reduced the inspector’s level of control, as did increased inspector fatigue during the day as a result of inspections they already had carry out. Both of these factors resulted in fewer violations being reported, which the researchers call the “day schedule effect.”
Inspectors try not to start an inspection without completing it on the same visit, as they would need to spend extra time and money to set up and travel to the site a second time. So, to finish what they start, inspectors can exercise some discretion over their working hours.
“They are free to complete one day’s work a little earlier than usual and another day a little later, depending on how long the inspections they did that day take,” says Ibanez. Still, inspectors try to maintain a cut-off time they know they won’t exceed from day to day.
“If you usually go home at a certain time, you’ll try to stick to that shift end time,” Ibanez explains. “But because it doesn’t make sense to stop an inspection once it’s started, it also means you might be in a bit of a hurry.”
Researchers find that if inspectors start an inspection at a time that would mean they wouldn’t finish before the normal smoking cessation time, they complete the inspection 4 percent faster than usual—and 5 percent fewer violations are found.
The second part of the day schedule effect occurs as inspectors become fatigued during busy days. The researchers found that each additional hour an inspector conducts inspections during a day results in 3.7 percent fewer citations per inspection that day.
Putting bias to work
Rather than claiming that food safety inspectors are failing at the job, Ibanez believes her findings suggest ways to help inspectors perform even better than they already do.
For example, the outcome effect—which prompts inspectors to exercise more control, which is generally good for public health—could be leveraged by realigning inspector schedules. Inspectors generally have some discretion in their scheduling, and Ibanez hopes these results will help them and their managers decide how to follow up on site visits.
“What if every time you finished a really bad inspection, you went to a primary school afterwards?” says Ibanez. “If certain experiences are going to increase an inspector’s detection rate, we should send them to places where that increased scrutiny will be more beneficial to society.”
Meanwhile, the effect of the daily schedule could be mitigated by “scheduling inspections earlier in the day so that the last inspection is always expected to finish before a fixed time of their choosing,” says Ibanez. And perhaps limiting the number of inspections an inspector can do each day to reduce fatigue.
Either way, Ibanez believes her research findings about real-world inspector biases can only help public policy.
“These are biases that people don’t notice,” he says. “Understanding them can help make individual decisions.”
And the results extend beyond this one type of work.
“Our work is not only about food safety, but other types of operations as well,” he says. “When we use data to make so many decisions, understanding the behavior of the people who generate that data is something we can use to improve our analysis – and therefore our decisions.”