1,200 respondents, each counted once in every column · journal: g_paths.tsv
a column is one item; a block is one answer, sized by how many gave it; a ribbon joins two answers on consecutive items, sized by how many gave both; hover anything for exact counts
The largest flows, each counting the people who gave one pair of answers on two neighboring items:
share
people
flow
32.0%
384
q1_consent Yes → q3_party Democrat
29.3%
352
q1_consent Yes → q3_party Republican
19.1%
229
q1_consent Yes → q3_party Independent
16.1%
193
q3_party Democrat → q5_voted Yes
15.2%
183
q3_party Democrat → q5_voted No
14.8%
177
q3_party Republican → q5_voted Yes
13.7%
164
q3_party Republican → q5_voted No
12.3%
148
q5_voted Yes → q8_approve Strongly disapprove
The most common full paths, reading q1_consent → q3_party → q5_voted → q8_approve.
share
people
path
7.7%
92
Yes → Republican → No → Somewhat approve
7.4%
89
Yes → Democrat → No → Somewhat disapprove
7.3%
88
Yes → Democrat → Yes → Somewhat disapprove
7.0%
84
Yes → Democrat → Yes → Strongly disapprove
6.8%
82
Yes → Republican → Yes → Somewhat approve
6.3%
76
Yes → Democrat → No → Strongly disapprove
6.2%
74
Yes → Republican → Yes → other answers
4.7%
56
Yes → Republican → No → other answers
3.2%
39
No → no answer recorded → no answer recorded → no answer recorded
2.9%
35
Yes → Independent → Yes → Strongly disapprove
How to read this map, step by step
Start at the left. The first column is q1_consent. The column stacks all 1,200 respondents: each colored block is one answer, its label gives the share who gave it, and the top block here is Yes at 96.8%. Light grey blocks collect everyone with no answer recorded on that item.
Follow a ribbon to the right. A ribbon connects an answer on one item to an answer on the next, and its thickness is the number of people who gave both. Hovering a ribbon shows the exact count. Where a block fans out into several ribbons, the sample is splitting; where ribbons converge on one block, paths are merging.
Blocks labelled other answers pool everything past the 3 most common. Nothing is dropped: every respondent sits in exactly one block of every column, so each column adds back to 1,200.
Each line of the table under the figure is one complete path from the first item to the last, with the number and share of respondents who took exactly that path. The ribbons above join two items at a time; the table lines run the whole way across.
Reading rule: counts are unweighted respondents in scope. A ribbon is the number of people who gave that pair of answers on two consecutive items; a table line is the number who took that complete path. no answer recorded includes people the routing never showed the item, which item data cannot separate from a decline. Drawn by surveymap from g_paths.tsv; the journal has a weighted count column when the scan was weighted.