User Story

The following decription is based on an actual processing of around 900 answers to a paper based survey.

The survey is made up of two separate schemas to be answered. Each participant answer the two sets, and the two sets must be correlated using some id.

The answers was collected over a 5 week period.

The post processing of the schemas was conducted in the following way:

• Using a scan of unused schemas an original "answer mask" was created
• schemas was scanned using an HP 2840 AIO scanner. Half was scanned individually, half was scanned in batches.
• scans was roughly aligned with the originals
• scans was processed by program that detects marks indicating answers
• scans was processed in sets, where a set is a set of schemas that was answered in the same session
• Program output was collected into a single excel file for import into Stata.

Various parts of these steps was automated, but still a large number of the steps was executed manually.