The synthetic values 2, 2, 3, 3, 5 have two equally common values: 2 and 3. Reporting only one can make a tie look like a unique winner.
Start by counting frequencies explicitly. A small table with value and count makes the tie visible before a function chooses how to represent it. Then decide whether your question needs every tied value, one selected representative or a different summary altogether.
In Python’s statistics module, mode and multimode have different behaviour when several values are equally frequent. Check the documented return rule before interpreting a single output as evidence that no tie exists.
For a survey example, suppose “morning” and “evening” each receive six votes. Selecting morning because it appears first in the file is an ordering rule, not a stronger preference among respondents. A report should either show the tie or explain an independently agreed selection procedure.
Be careful with very precise measurements. If almost every recorded value occurs once, a list of modes may add little insight. Grouping values can help answer a different question, but the group boundaries must be stated.
The key check is simple: ask what the function returns when frequencies tie, then test it with a tiny dataset whose counts you can verify by eye.
Image: an editorial illustration, not a documentary photograph.