Skip to lesson
Exit
Mixed Methods & Quant Literacy1 / 3

2 min lesson

Where the bias hides

Answer this: "You launch an in-app survey via a banner and 2,500 developers respond. A colleague wants to report the satisfaction average as representing all Cursor users. What's the main risk?"

Step 1 of 3

Where the bias hideserrors that survive a clean-looking dataset

Order effects

Earlier questions prime later answers

Asking about bugs first depresses later satisfaction

Randomize item and option order where you can

Acquiescence

People tend to agree with statements

Inflates every “do you agree…” item

Mix positively and negatively keyed items

Non-response

Who skips the survey isn't random

Frustrated power users may opt out entirely

Check response rate by segment, not just overall

Self-selection

Volunteers differ from the population

An in-app prompt over-samples active, happy users

Weight or caveat; don't generalize to churned users

Learn more

Full explanation

Sampling and how confident you get

Sampling and how confident you getwho, how many, how sure

Sample size is a precision dial, not a magic threshold. Bigger samples narrow the margin of error with diminishing returns and they never fix a biased frame. Decide who you need first, then how many.

Reasoning about n
Who
Define the target population and the frame you'll actually reach (in-app vs. emailed vs. all seats).
How many
For a single proportion, ~400 responses gives roughly a ±5% margin at 95% confidence.
Segments
You need that n per subgroup you want to compare, not in total.
Bias beats size
A representative 300 beats a skewed 3,000; coverage error doesn't shrink with n.

These figures are standard survey rules of thumb, not Cursor-specific numbers.