Skip to main content
Family riding a tandem bicycle: School leadership is a team effort!
School Data Leadership Association · Analysis & Interpretation

Asking a good
data question

Turn a concern into a question someone can answer—and connect the answer to a decision.

Start with a concern. Give it boundaries.

The concern

“Attendance seems worse. What’s going on?”

Worth investigating, but not yet clear about who, when, or what “worse” means.

A question we can investigate

“For each grade at Pacific Crest, how does the percentage of possible attendance days recorded as present in the first 20 instructional days compare with the same period last year?”

This describes a comparison. A second investigation would be needed to understand the reasons.

Build a question brief

Edit the example. The brief updates as you type. Use process descriptions and role names only; keep student information in approved systems. Your edits stay in this page until you reload or close it.

Name the population and any inclusion rules.
Use a measure you can define and calculate consistently.
Use instructional dates or a clearly bounded reporting period.
Use a comparison with the same definition and period length.
Name roles rather than individuals.
Describe the next decision, not a conclusion already reached.
State the source, calculation, and checks. Do not enter student information.
Name a limit and the follow-up it calls for.
SDLA · Question brief

A question worth answering

Discussion draft · Confirm definitions and feasibility with the data owner.

Audience
Decision this could inform
Evidence and checks
Limits and follow-up

Before you hand it to a colleague

Does “attendance” mean the same thing to both of us?

Write the numerator, denominator, eligible population, and time period. A percentage of days present is different from a percentage of students meeting an absence threshold. Do not substitute one for the other.

Is the comparison fair enough for this decision?

Confirm calendar length, enrollment eligibility, coding, and correction status for both periods. The students in a grade this year are not necessarily the same students as last year. Record differences that could affect interpretation.

Am I asking for a description or a cause?

“Which grades changed?” is descriptive. “Did our outreach cause the change?” needs a design that can examine other explanations. A before-and-after difference alone does not establish a cause.

What if the data cannot answer it yet?

Record the missing definition, source, or access decision. Ask the owner whether an existing source can answer the question, whether its scope needs to change, or whether more information is needed. Do not fill a gap with an assumption presented as a fact.