What does your file physically look like?

How we handle this

• Shape is detected from your file. Column types, repeat keys and date columns decide whether you have a cross-section, a panel or a duration.
• The wrong shape is refused, not fudged. If your data can't support the analysis you picked, you get told which analysis it can support.
• Repeated measurements are respected. Multiple rows per subject are modelled as such instead of being averaged into a single misleading number.