Questions Create the Dataset

A survey does not discover information automatically. The questions decide what information can exist in the response sheet.

If a question is vague, leading, double-barreled, or missing a useful answer option, the problem is already in the dataset before anyone opens Sheets.

Every question creates a field

Suppose a Form asks:

How satisfied are you with the school Wi-Fi and Chromebook battery life?

A learner answers Not satisfied.

What does that answer mean? Wi-Fi? Battery? Both? One question tried to measure two different things.

A better design separates the variables:

1. How satisfied are you with school Wi-Fi reliability?
2. How satisfied are you with Chromebook battery life during a school day?

Now the response sheet contains two interpretable fields.

Match question type to the information

Common Forms question types support different evidence:

• multiple choice — one option from a defined set;
• checkboxes — zero or more options when multiple selections make sense;
• linear scale — ordered rating, but only if endpoints are clearly defined;
• short answer — flexible text, harder to summarize consistently;
• dropdown — compact selection from a defined list.

Do not choose a type because it looks interesting. Choose it because it matches the information you need.

Worked repair: leading language

Weak:

Don't you agree that our awesome esports room needs more computers?

The wording pressures the respondent and frames one answer as desirable.

More neutral:

Which change would most improve the esports room for your current use?

Then provide appropriate choices plus an Other option if the design needs it.

Run a Question Design Audit

Repair these fictional questions:

1. “Do you like technology?”
2. “How good are the fast computers and internet?”
3. “Why is the current schedule bad?”
4. “Enter your full name, home address, phone number, and favorite school app.”

For each, identify the problem before writing a revision. Question 4 may not need to be revised—it may need fields removed.

Use this table:

Failure mode: adding “Other” without a plan

Free-text answers can capture something the designer did not predict, but they are harder to categorize and can invite information you did not intend to collect. Use free text when its value outweighs the added interpretation/privacy cost.

Evidence

Submit the Question Design Audit and one original question for a non-sensitive classroom survey. State the field name/type you expect the question to create in the response sheet.

If you cannot describe the resulting field, the question is not ready for data collection.