How the Workload Census is counted

Who answers

Anyone who works in one of eight roles may answer: school psychologists, SLPs, OTs, PTs, special education teachers, paraprofessionals, behavior analysts and social workers and counselors. Each person answers for their own typical week in the current school year, which runs from July 1 to June 30.

Medians, not averages

Every figure is a median: the middle answer when everyone’s answers are put in order, or halfway between the two middle answers when the count is even. A median is not pulled up by one unusual week. Ten caseloads of 38, 41, 44, 47, 52, 55, 58, 61, 66, 140 have a median of 53.5; their average would be 60.2, because of the one caseload of 140.

Hours beyond contract are worked out for each person first (the hours they work in a typical week minus the hours they are paid for) and the median is taken of those differences. It is not the difference between two medians.

Time entered in hours and minutes is kept in whole minutes and shown in hours, rounded to one decimal place.

Groups of 10 or more

A figure is shown for a role in a state, or for a role nationally, only once 10 or more people have answered that question. A smaller group is not shown at all, in any table, download or comparison, and its size is not shown either. Each question is counted on its own, so a group can show its caseload median before it shows, say, its travel time, because fewer people answer the optional questions.

The date beside a group is the month its first answer arrived. The count beside it is the number of people whose answers the figure uses.

What the Workload Census cannot tell you

• It is not a random sample. People choose to answer, and those with the heaviest or the most unusual weeks may be more likely to. The figures describe the people who answered, not every person in the role.
• Every answer is self-reported. Nobody checks it against a schedule, a contract or a roster.
• A typical week is the answerer’s own judgment. Weeks with evaluations due, the start of the year and the end of the year can look very different.
• Roles are named differently from state to state, and one person may do more than one job. Each person chooses the role closest to their work.
• One response per device per school year is a convenience, not a guarantee. Someone who clears their browser or uses a second device can answer twice. The Workload Census does not try to prevent that by tracking anyone.
• A link may carry a source tag (for example, from a professional association’s newsletter). It is kept with the response only to count how many answers each link brought in, and is never shown beside a figure.
• The Workload Census is not an evaluation of anyone’s work, and no figure describes a school or a district.

How your answers are kept

• No name, school, district, student, email or address is part of a response. Every answer is a number, a time or a choice from a list; there is nowhere to type a name.
• No IP address is stored with a response, and the Workload Census sets no cookie.
• When you first save, your browser makes a random key and keeps it in its own storage. The Observatory stores only a one-way fingerprint of that key, which is how your browser can reopen and change your answers until June 30. A new school year starts a new key, so two years of answers from one device cannot be linked. After the year ends the fingerprints are deleted.
• A response is stored with the date it was saved, never the time, so it cannot be matched to anything else by when it arrived.
• If you leave an email for the full results, it is kept apart from your answers, with nothing that links the two, and used only to send that year’s results.
• Individual answers are never published or shared, with anyone. Only the medians described above leave the database.

Corrections

If a figure looks wrong, write to us. A corrected figure says so beside it.