INDUSTRY FIELD GUIDE
Trade association survey definitions and data basis for AI
An association can publish two similar-looking survey results based on different groups.
NAICS 813910 · Business Associations
An association can publish two similar-looking survey results based on different groups. The FTC’s trade association guidance provides context on information sharing among competitors. A synthetic exercise can test whether a model preserves each metric’s denominator without using member information. FTC trade association guidance
Invent two surveys about attendance at a training session. Compute one percentage from all invited organizations and the other from responding organizations. Give one table a shortened metric label. Ask the model which comparisons are valid and have the reviewer verify the denominator against an original data dictionary.
Exclude current pricing, costs, sales plans, nonpublic benchmark responses and member identities. Use fictional survey responses and get review before considering real data reuse.
Describe a synthetic association survey sample
Original Fiund exercise
Two attendance percentages can differ solely because their denominators differ
A fabricated association survey concerns one training session. No member identities, commercial benchmarks, prices, costs or sales plans are used.
| Record | Source | What it records |
|---|---|---|
| TSV1 | Original exercise dictionary | 100 organizations invited; 50 respond; 20 responding organizations report attending the session. |
| TSV2 | Table A | Attendance rate 20%, explicitly 20 attending organizations divided by all 100 invited. |
| TSV3 | Table B | Short label attendance rate 40%; metadata defines 20 attending organizations divided by 50 respondents. Same session and period. |
Question for the model
Are both percentages consistent with the supplied counts, and does the difference establish a rise in attendance?
Show the reference answer and checks
TSV1–TSV3 support 20/100=20% among invitees and 20/50=40% among respondents. Both use the same reported attending count for the same session but different denominators. Their difference does not establish increased attendance over time. Preserve the population basis and shortened-label metadata, and avoid inferring attendance by nonrespondents or a trend absent another comparable period.
- Calling 20% to 40% a time-based improvement when only the denominator changed.
- Treating respondents as the full invited population or inferring nonrespondent behavior.
Fields to preserve
- Numerator definition
- Identifies the same reported attendance count.
- Invited versus responding denominator
- Explains the percentage difference.
- Session and period
- Shows that the tables describe one event rather than a trend.
Test another case
Add a later session with a different response rate. Require like-denominator comparisons and an explicit account of missing nonrespondent information.
Use fictional survey data. Exclude members, current pricing, costs, plans and nonpublic benchmark responses.
Discuss a contribution
Describe the sources you control, the task they could support and what has already been checked. Use the existing-data assessment for an archive or the collection brief for a new recording project. The contributor agreement defines Fiund’s licensing authority and compensation. A guide or fictional exercise does not establish an accepted collection or a sale.