PRACTICAL GUIDE
Clean a List Before a Random Draw: Rows, Labels and People
A worked duplicate-cleaning example showing when deduplication changes your population and why unique rows are not always unique people.
A random sampler can select accurately from a badly prepared list. The result still answers the wrong question. Before drawing, decide whether one entry represents a person, a ticket, a task or a row in a file. That choice determines which apparent duplicates should stay.
A six-row example
Suppose your pasted list contains these six lines:
- Alex
- Alex
- alex
- Alex M
- Priya
- Priya, team blue
The first two lines have exactly the same label. The third uses different capitalization. The last line contains a comma, but it is still one entry because the sampler separates entries by line breaks.
If you keep all entries, there are six eligible rows. If you choose “Deduplicate exact labels,” there are five: only the repeated “Alex” is removed. Leading and trailing spaces are trimmed before this check, and blank lines are ignored. Matching is case-sensitive, so “Alex” and “alex” remain different.
Decide what those rows mean
If the first two lines are different people who share a name, removing one would exclude a person. Give each a distinct, minimally identifying label instead. If they are two authorized tickets for the same person, keeping both may be the intended rule. If one was pasted accidentally, remove the extra row.
Likewise, “Priya” and “Priya, team blue” might be one person or two different entries. Software cannot infer that from the labels. Compare the source before deciding, and avoid adding private information just to make names distinct.
Choose the method after cleaning
Without replacement prevents a selected row from being selected again in that draw. It does not prevent the same visible label from appearing twice when two matching rows remain. With replacement allows the same row to return on a later draw.
For three distinct people, first build one valid row per person, then sample three without replacement. For a complete presentation order, use a full-list shuffle. For repeated practice questions, replacement may be appropriate if repetition is part of the activity.
Keep a small preparation record
- What one row represents
- Where the source list came from and when it was checked
- Which exclusions or label corrections you made
- The duplicate policy and replacement setting
- The final eligible row count and requested sample size
Keep that note and the original list with the downloaded result. The sampler’s CSV describes the draw settings, but it does not certify that the input was complete or unbiased. Its browser pseudorandom generator has no user-controlled replay seed. Use this workflow for ordinary list work and learning, not a regulated or independently audited selection.