How are counts and percentages calculated?
Overview
Byleron counts coding assignments: one code applied to one selected passage, region or range. A passage with three codes is one selection and three assignments. A parent code's count adds the assignments of its subcodes. Shares are shares of all coding assignments in the scope. Coverage is the share of a document's characters that carry at least one code, with overlapping passages counted once; it is measured for text only. Density is coding assignments per 1,000 characters of text. Intercoder agreement is calculated per document, from whether each document holds each code. Co-occurrence scores are always calculated from documents.
What to do
- Worked example: an interview of 1,000 characters has passage P1 (characters 100 to 300), coded with Subcode A1 and Subcode A2, and passage P2 (characters 250 to 400), coded with A1. Both subcodes belong to Parent A.
- Recorded assignments: 3 (P1 with A1, P1 with A2, P2 with A1). Subcode A1 has 2, Subcode A2 has 1, and Parent A has 3, although only two different passages were coded.
- Shares: A1 is 2 of 3 (67 %), A2 is 1 of 3 (33 %), and Parent A is 3 of 3 (100 %). The parent already contains its subcodes, so do not add 100 % to 67 % and 33 %.
- Coverage: the coded characters run from 100 to 400, so 300 characters, or 30 % of the interview. The overlap of P1 and P2 is counted once.
- Density: 3 assignments in 1,000 characters is 3.0 per 1,000 characters.
- Intercoder agreement: two coders are compared on whether each document holds each code. It does not show whether they marked the same passage or the same boundaries.
- Co-occurrence: distinguish sources meeting the selected rule from repeated coding-assignment pairs. Same document uses standard source-based similarity coefficients; Overlap and Nearby use relationship coverage and explicitly named source ratios. See the Co-occurrence Help article for a worked example.