Atlas › Work tasks

Case study 25 of 198

What jobs are like, according to Jev and to the workers

How often does a nurse deal with angry people, a web developer face deadlines, a roofer work in the weather? Does Jev know what jobs are like, compared with what the people doing them report?

result491 questions

Jev knows roughly what jobs are like: across 45 occupations and 12 working conditions it orders them closely like the workers themselves (rank correlation 0.74). But it imagines far more exposure to disease than workers report (+1.32 levels on a 0-4 scale on average), and less freedom and lower stakes. It thinks fast food workers face disease almost daily (3.5 of 4, workers say 0.2), and that a flight attendant's mistakes are only fairly serious (1.7, workers say 3.5).

-1012

Freedom to Make Decisions

ρ 0.55

Consequence of Error

ρ 0.78

Level of Competition

ρ 0.73

Spend Time Sitting

ρ 0.89

Electronic Mail

ρ 0.87

Degree of Automation

ρ 0.48

Outdoors, Exposed to Weather

ρ 0.86

Time Pressure

ρ 0.15

Public Speaking

ρ 0.51

Frequency of Conflict Situations

ρ 0.62

Deal With Unpleasant or Angry People

ρ 0.81

Exposed to Disease or Infections

ρ 0.78

Jev

How to read this: Each row is one working condition. The dot is how far Jev's picture sits from what workers report, averaged over occupations: right of zero, Jev thinks it happens more than it does. The label is how well Jev orders the occupations on that condition.

491 occupation-item pairs over 45 occupations; 90% intervals over occupations.

In short

Across 45 occupations and 12 working conditions, Jev orders jobs much like the US workers who do them, a rank correlation of 0.74.
Its biggest error is germs: it puts fast food workers, bartenders and cashiers near daily disease exposure; the workers report 0.2, 0.1 and 0.7 of 4.
It also shrinks the stakes: Jev rates a flight attendant's mistakes 1.7 of 4 for seriousness, flight attendants 3.5.

What the data shows

Charlie Conspiracy (Always Sunny in Philidelphia) meme: fast food, germs, EVERY DAY

The ordering is mostly right. Across all pairs, rank correlation 0.74. Jev knows who sits all day (0.89), who reads email (0.87), who works in the weather (0.86) and who deals with angry customers (0.81).
Disease everywhere. Jev overstates exposure to disease or infection by 1.32 levels on average: fast food workers (3.5 vs 0.2), bartenders (3.4 vs 0.1) and cashiers (3.9 vs 0.7) all come out near "every day".
More conflict, less freedom. It overstates conflict (+0.52) and angry people (+0.59), and understates how much freedom workers have to make decisions (-0.30) and how serious their mistakes would be (-0.27). It rates a flight attendant's mistakes as fairly serious (1.7) where the workers say very serious (3.5), and a chef's public speaking at 0.6 where chefs say 2.2.
Deadlines are a blur. On time pressure Jev barely knows which jobs have more of it than others (0.15).

What it means, and what it doesn't

Jev's picture of work is the outside view: public-facing jobs as germy and conflict-filled, workers as having less say and lower stakes than they report. If you ask it what a job is like, you'll get the reputation, which is often right about the shape and off about the details that matter to the person doing it.

It doesn't mean the workers are right and Jev wrong on every point. Self-reports have their own blind spots, and exposure to disease is something people may stop noticing. The gap is worth knowing either way.

Caveats

Who the workers are. O*NET surveys a sample of people working in each occupation in the US, often a few dozen per job. Their answers describe US jobs at the time of the survey; a cashier's job in another country may differ.
Workers describe their own jobs. Self-reports can understate hazards people have gotten used to, or overstate what feels important about their work. Jev's picture is closer to how a job is described from outside. The gap is between reputation and self-report, not between Jev and objective truth.
One typical worker vs a spread. Jev answers about a typical member of each occupation; the survey spreads across many real workers, from quiet shifts to hectic ones. The comparison uses averages.
O*NET's wording. The five answers are O*NET's own ("Once a week or more but not every day"), but the question sentences were written for this project around each O*NET item.

Jev on this experiment

Would a person find it interesting to read?: Yes81%
Does it describe you?: No53%
Would you have predicted it?: No59%
How fair is the comparison?: The comparison is shaky
How much should a reader rely on it?: Moderately
Which caveat matters most?: Workers describe their own jobs89%

Why ask this

People ask models about careers all the time: what's it like to be a nurse, is being a flight attendant stressful, would I be sitting all day as a web developer. The answer a model gives is its picture of the job. That picture could match what workers experience, or it could be the reputation of the job, the version from TV and headlines.

The US Department of Labor asks workers directly how often they face angry people, deadlines, weather, disease and more. So Jev's picture can be checked against the people doing the jobs.

How this was done

The people and the data

The comparison is *ONET Work Context** (version 29.0), part of the US Department of Labor's occupational database. ONET surveys people currently working in each occupation in the US, often a few dozen per job, asking how often each condition is part of their job, and publishes the share choosing each answer. The experiment used 12 conditions (dealing with angry people, conflict, working in the weather, time pressure, public speaking, email, exposure to disease, sitting, freedom to make decisions, how serious mistakes are, automation, competition) for 46 well-known occupations, from nurses and cashiers to air traffic controllers and roofers. Rows ONET marks as unreliable are left out, leaving 491 occupation-condition pairs over 45 of the occupations. The data are published under CC BY 4.0.

What Jev was asked

Each pair was one question, with O*NET's own five answers:

How often is a fast food worker exposed to diseases or infections at work?
Never · Once a year or more but not every month · Once a month or more but not every week · Once a week or more but not every day · Every day

Jev also answered with the answers in reverse order, and the two are averaged.

How it was measured

Each answer becomes a level from 0 (never, or the lowest) to 4 (every day, or the highest). For each condition, the analysis ranks the occupations by Jev's level and by the workers' average and compares the rankings (1 means the same order), and averages the gap between Jev and the workers, with a range showing how much it could vary by chance.

Where these questions live

491 questions across 1 topic of the map. Each opens on the map with every question in it.

Careers & the Workplace

What Jobs Are Like491

Every question

All 491 questions behind this result, the telling ones first: the examples the analysis points to, then the ones where Jev misses, biggest gap first.

Jev’s own answer

Showing 0 of 491