November 1, 2017

Distribution of work per transcriber

Power transcribers

How did engagement change over time?

Weekends are higlighted in blue.

Transcription lengths

Transcription time

Use of [unclear] tags.

About 44.3% of transcriptions used the [unclear] tag at least once.

Special characters

15.2% of transcriptions were marked as "Contains special characters". This covers 40.1%, though, which means that volunteers were not at all consistent in marking which documents had special characters.

Special characters

Judging Transcription Consensus

Most documents have 6 useful, non-blank transcriptions, while a few have 5, but this is as low as it goes, so the data are quite satisfactory.

## # A tibble: 12 x 2
##    elligible_transcriptions     n
##                      <fctr> <int>
##  1                        5   169
##  2                        6  1904
##  3                        7   242
##  4                        8    24
##  5                        9     3
##  6                       11     7
##  7                       12     4
##  8                       13     2
##  9                       14     4
## 10                       15     3
## 11                       16     1
## 12                       17     1

Pairwise-similarity summaries

The levels of transcription similarity/dissimilarity are in line with both our beta results as well as the mid-stream check-in. The vast majority of docs had largely agreeable transcriptions.

Drawings

Each file had 3 votes for the quesiton "is there a drawing?" The vast majority of votes were unanimous, with just 40 docs having conflicting answers.