Try something like that (you can adapt it) right after \begin{document}:

\makeatletter
\fancyhead[C]{\the\AMCid@etud{} / ballot \thepage\\
\AMCIDBoxesABC}
\makeatother

I am still working on figuring out how exactly to do the scoring.

You can use auto=1 or similar, or just read the required data from the "stats table" in the exported ODS file, or from the CSV exported file with "ticked boxes" support.

That does the trick! My TeX fu is almost non-existing, so I am learning it step-by-step, And now knowing what to search for (\AMCIDBoxesABC), I also found the related Wiki entry. It would be a useful addition to the PDF userguide, actually.

I take it that \AMCIDBoxesC is a 6-bit deep error correction field that covers both questionnaire and page number? I ask because I wonder how easy it would be to "sneak in a ballot" by manually filling in one of the empty ID boxes.

Related to that: If AMC reads two scans with the same ID (but different content), do I get an indication of that in the UI, so that the user can then decide within AMC which ballot to keep and which to discard? That would need to be done after an examination of the physical paper, not just the scanned image.

I take it that \AMCIDBoxesC is a 6-bit deep error correction field that covers both questionnaire and page number?

That's it.

how easy it would be to "sneak in a ballot" by manually filling in one of the empty ID boxes.

This error correction code is only here to detect bad scans, and not to counter sneaking: this code is decreasing by 1 for each page...

If AMC reads two scans with the same ID (but different content), do I get an indication of that in the UI

• When using the 'one scan per page' mode:
  No. When you feed AMC with another scan of the same page, the results overwrite the old ones.
  You have to check that the number of results is the same as the number of scans. If not (but this should not be the case), this will take you some time to discover which page has multiple scans.
• When using the 'photocopy allowed' mode,
  you will get as much entries as scans you gave to AMC. So you will be able to check the number of scans for each page, look at them and erease those you want.

• most importantly it prevents dupes, caused by scanning the same ballot accidentally twice
• it also permits me to handle all ballots (=election rounds) in one project, with the page number being used as the ballot number

If I may, I'd like to suggest a feature enhancement here: Somehow notify the user of the file names when the same ID (an ID conflict) is found being caused by images having different filenames. That could be a log entry, or a pop-up, or a button with a pop-up combined with a status line just like "All scans were properly recognized". I believe this also matters for testing students, not just paper ballots, because as rather simple manipulation of the ID bar can allow a student to write a test for another student.

Likewise a log for the files that were hit with the "Remove" button seems wise, showing both file name and the paper/page number.

As to my project: I am still undecided whether I want all 12 ballots in the same project file, or if I should rather have one project for each. The entire process is spread out over approx. 3 days, and elections are done in 2 or 3 batches.

• the page number (and through it the invisible question number) can be used to prevent a ballot paper to "spill over" from one ballot to another; such a spill over could silently change the results of past election
• one central and consolidated statistics (OpenOffice Calc export)
• no need to manage (open and close, backup) many projects

• individual and shorter .tex files, resulting in fewer "damage" when edited incorrectly
• scans stored in separate folders, allowing for somewhat better auditing and verification
• cleaner AMC UI because the "Data capture" tab there is no way to sort or filter the scans by page number (the existing "updated" column is not sortable)

Did I miss anything relevant?

Missing info: There can be up to 5000 ballot papers in total.

the page number can be used to prevent a ballot paper to "spill over" from one ballot to another

Another way to do that is to change the starting value of the correction code associated to student/page for each ballot.

\makeatletter\AMCid@check=50\makeatother

(the existing "updated" column is not sortable)

You can open a feature request: that sounds feasible...

Missing info: There can be up to 5000 ballot papers in total.

Be careful that the max value for the number of students is 4095 (12 binary digits), so you have to change that:

\makeatletter
\def\AMC@NCBetud{13}
\makeatother

On purpose this ballot paper is very narrow (74.25mm x 210mm).

Your scanner allows you to scan this page size easily using a feeder?

Yes, tested with a Fujitsu ScanSnap ix500 - no dropouts so far due to angled paper intake (in admittedly limited testing). Yet there was one case where an orientation circle got very close to the edge. Side remark: I always wondered why all 4 circles are required, wouldn't finding 3 of the 4 be sufficient? But back to topic: From your comment it appears that you would rather recommend A6 instead? That would reduce scanning speed by 1/4 though.

With regard to "AMCid@check=50": Very interesting! Due to the large number of sheets I am leaning towards a separate project for each ballot, yet the "spill over" danger is what keeps me from doing that. Typically each ballot also comes with its own paper color, but there are a) strict limits as to the number of available light colors, and b) it requires human attention.
So, if each ballot has 500 sheets, how exactly would I employ "AMCid@check= reliably? I am asking because this field by itself can only count to 63.

wouldn't finding 3 of the 4 be sufficient?

Theorically yes, but less reliable... and not implemented.

would rather recommend A6 instead?

No. I recommend testing before large scale use. But you tested, so this is OK.

how exactly would I employ "AMCid@check= reliably?

You have to set a different value (choose one from 1 to 60) for each ballot, but only once before (outside) the \onecopy loop. It will be decremented for each page automatically. The point here is that the same student/page couple will have different check (correction code) for different ballots, so that if you feed AMC with a scan from ballot 1 in the ballot 2 project, you will get an unrecognized scan that won't affect the ballot 2 results.

A few days laster, having learned a lot through more tests, I can now state that it would be definitely a useful feature if AMC were "more tolerant" and accepted images with a positive identification of only 3 of the 4 orientation circles. In all of the cases that I came across where a paper was scanned in too much of an angle - and/or one corner was covered by the next paper - everything else (ID box, checkboxes) was perfectly fine.

• 74mm heigth x 210mm width (A4 quartered): Up to 4% scans with one circle missing
• 105mm heigth x 148mm width (A6 turned): Up to 2% scans with one circle missing
• 148mm heigth x 105mm width (A6): Near 0% scans with one circle missing
  The smaller the height, the faster the scanning speed, which in my case is very relevant.

I can now state that it would be definitely a useful feature if AMC were "more tolerant" and accepted images with a positive identification of only 3 of the 4 orientation circles.

I created #197 for that.

I'll be very glad to test or provide test material.

If you can send me the source file and some scans with only 3 corner marks?
And also test revision r1416...

Check your email.

Check your email.

Thanks!