Partially filled rows

• A form was abandoned partway through and the partial entry was saved.
• Records were created with the minimum needed at the time, with a plan to finish them later.
• Rows from a source with fewer fields were combined with rows from a richer one.

• Analyses that need complete rows silently shrink to a subset that may not resemble the whole.
• Outreach fails for records that lack the one field the channel needs.
• Sparse rows are hard to match, so they are often undetected duplicates of fuller ones.

• Count the filled fields in each row and look at how the counts are distributed.
• Check which fields tend to be missing together. That often traces back to one source or one version of a form.
• Look for rows with only an identifier and nothing else.

• Define which fields are required for each use. A row can be complete enough for email and useless for mail.
• Check whether a sparse row matches a fuller record for the same entity before treating it as a separate one.
• Deciding to drop incomplete rows is a judgment about the analysis, not a cleaning step. Record how many were dropped and why.

• True blanksA cell is really empty, and the file does not say whether the value is unknown, not applicable, or was never asked for.
• Near-duplicate recordsTwo rows describe the same real thing but differ enough that an exact comparison misses them.
• Disguised missing valuesA cell is not empty, but what it holds means there is no value: N/A, a dash, 999, or the word none.

TermsCompleteness, Missing value, Record, Data quality