Local Exchange Research
Methodology
Who we hear from
Local Exchange works with owners of independent and neighborhood businesses across Chicagoland. Participants come to us through Business on the Block, community partners, and owners who reach out directly. Our participants are not a random sample of all businesses, and we say so whenever we report findings.
How the research is collected
Each participating owner first tells their story in an open conversation. Afterward, the owner and a Local Exchange researcher complete a short structured instrument together, one section at a time:
- Business profile
- Current business conditions
- Business barriers
- Access to help
- Information and trust
- Business networks
- Priority needs
- Review with the owner
The barrier matrix
Owners rate twenty barrier domains — from access to capital to transportation and parking — on a standard five-point scale:
- 0 — Not a problem
- 1 — Minor problem
- 2 — Moderate problem
- 3 — Major problem
- 4 — Critical / immediate problem
The owner then chooses their own top three priorities. We do not assume the most severe barrier is the most important one; the owner decides, and then confirms the priorities before anything is recorded as verified.
Consent
Consent is granular and recorded separately for research use, follow-up contact, anonymous quotes, attributed quotes, referrals to partners, and media participation. Appearing in a Business on the Block episode does not mean an owner has joined the research, and joining the research does not mean an owner can be quoted by name. Owners can read exactly what each choice means on how we use research information.
Three layers of data
- Raw — what the owner said, preserved as recorded.
- Coded — standardized variables and research codes from a published codebook.
- Interpretation — researcher memos, qualitative themes, and aggregate analysis.
Raw material is never overwritten. Every interview is reviewed and verified by a person before it is included in research findings.
Follow-up
With the owner's permission, we check back at roughly 30 days, 90 days, and six months to learn whether contact occurred, whether assistance occurred, how useful it was, what remains unresolved, and whether business conditions changed.
How we measure
| Measure | Definition | Note |
|---|---|---|
| Connection rate | Verified connections ÷ needs for which a connection was sought | A referral alone is not counted as a connection. |
| Assistance rate | Connections resulting in substantive assistance ÷ verified connections | A verified connection is not automatically substantive assistance. |
| Resolution rate | Needs fully or mostly resolved ÷ needs with sufficient outcome follow-up | Always reported alongside the follow-up completion rate. |
| Median time to connection | Days from need identified to verified connection | The median is used because a few long cases can distort an average. |
Rates are always published with their numerator and denominator — for example, 68% — 34 of 50 — so readers can judge how much weight a number can bear.
We collect the components of navigation burden (organizations contacted, handoffs, days spent seeking help, unsuccessful attempts, and documentation burden) individually. We have not created a composite score, and will not until its reliability and validity have been tested.
What our findings do not claim
- Improvement after a referral does not prove the referral caused it. We do not make causal claims from this data.
- We do not score business owners, predict business success, or decide who deserves assistance.
- We do not infer discrimination or intent without evidence from the owner's own account.
The role of AI
AI tools may help draft interview summaries, suggest codes from our approved codebook, or flag possible gaps between what an owner said and what was recorded. Every AI output is labeled as a draft, kept separate from verified data, and reviewed by a person. AI never changes what an owner said, never assigns final codes, and never publishes findings.
Privacy
Each business receives a persistent research ID that contains no location information. Research datasets shared for analysis exclude names, contact details, exact addresses, and social handles. Private platform records are never shown on this website.