Intermediate interview questions often describe a plausible output that is still wrong. A total can double because a lookup key repeats, and a union can keep the expected row count while shifting values into the wrong fields. Explain the mechanism before proposing a repair. A duplicate row can also be legitimate, so don't remove it without identifying the intended grain.
These questions ask you to trace matching combinations, field alignment and report shapes. Use the input definitions to decide what each output record means. For parsing, distinguish validation from extraction. For summaries, state the grouping fields as well as the aggregation. A short answer that names the actual cause is more useful than a long list of tools you might try. The choices use original fictional examples.