RQ1: Happy identity endorsement and Cleveland Browns fandom
Executive summary · Two-column short report (PDF) · Analysis artifacts
Among eligible Ipseity Daily respondent-days in the verified Zenodo snapshot, 83.23% of those endorsing Cleveland Browns fan also endorse happy, compared with 76.55% of those explicitly rejecting Browns fandom. The unweighted prevalence ratio is 1.087 (respondent-cluster bootstrap 95% interval 1.007–1.160). The difference is +6.68 percentage points (95% interval +0.56–+12.17 points).
These are contemporaneous self-descriptions. They do not establish a causal benefit of fandom or measure happiness with a multi-item psychological scale. The observation window is July 8, 2025–August 27, 2026, not the retrieval date of September 11, 2026.
RQ1 asks how frequently happy is endorsed among Cleveland Browns fans relative to explicit non-fans. Following Jones’s ipseological approach, identity signifiers are the words people use to describe themselves (Jones, 2023). Here the survey elicits Yes/No endorsement of supplied signifiers, rather than spontaneous words in a biography. Our inference concerns endorsement in that measurement setting.
The prevalence ratio is the proportion endorsing happy among Browns-fan respondent-days divided by the corresponding proportion among explicit non-fan respondent-days:
\[PR = \frac{P(\texttt{happy}=1\mid\texttt{Cleveland Browns fan}=1)}{P(\texttt{happy}=1\mid\texttt{Cleveland Browns fan}=0)}.\]
Both probabilities condition on eligibility below. A ratio of 1 means equal prevalence; 1.087 means approximately 8.7% higher relative prevalence, not 8.7 percentage points. A risk ratio has the same mathematical form, but usually concerns subsequent incident outcomes. “Prevalence ratio” is the PI-selected term for these contemporaneous measurements.
The primary estimand weights eligible respondent-days equally. Respondents with more eligible days therefore contribute more. Clustering the uncertainty does not give each person equal weight in the point estimate.
Ipseity Daily provides response and demographics microdata; its instructions require an inner join on both hashed_respondent_id and obs_date (Jones, n.d.). The broader infrastructure is described by Jones (2026a).
The retrieved primary-host files were incomplete relative to the page’s advertised cumulative coverage. On September 11 they contained 10,570 response rows and 151 demographics rows, with eligible dates only July 8–14, 2025. We cannot establish why. We did not treat them as cumulative data. Instead, we retrieved the fixed Zenodo record 22139541 (Jones, 2026b), verified both published MD5 checksums, and recorded SHA-256 hashes. That archive contains 675,480 response rows and 8,776 demographics rows, with eligible observations through August 27, 2026.
| Snapshot | Response rows | Eligible days | Last eligible date | PR |
|---|---|---|---|---|
| Verified Zenodo; current report | 675,480 | 8,235 | 2026-08-27 | 1.087 |
| Earlier saved result; inputs unavailable this iteration | 677,121 | 8,253 | 2026-08-28 | 1.088 |
| Primary-host retrieval this iteration | 10,570 | 142 | 2025-07-14 | 1.192 |
The earlier result is retained as historical evidence, not an independently reproduced result. The short primary-host subset has a wide clustered interval (0.938–1.413) and is not a conflicting estimate of the same observation window. A changed URL response is a provenance issue, not evidence of a temporal trend.
The response columns are hashed_respondent_id, obs_date, signifier, and endorsed (1/0). An eligible respondent-day must have a valid demographics match and explicit binary answers to both target signifiers. An absent item is never No: signifier presentation is probabilistic and can change over time. Rows with the CONSENT_REVOKED sentinel are excluded before the join. Duplicate demographics keys stop analysis; identical target answers are collapsed, and conflicting target respondent-days are excluded.
The archive audit records 113 excluded consent-revoked demographics rows, 1,945 unmatched response rows, three collapsed identical target duplicates, and one excluded respondent-day with conflicting target answers. There are 8,235 eligible respondent-days from 4,686 respondents; 1,392 respondents contribute multiple eligible days (maximum 39).
Virtual CSSERGcells <- e$two_by_two
tab <- data.frame(
`Browns fan` = c("Yes", "Explicit No"),
`happy Yes` = c(cells$fandom_yes_happy_yes, cells$fandom_no_happy_yes),
`happy No` = c(cells$fandom_yes_happy_no, cells$fandom_no_happy_no),
check.names = FALSE
)
tab$Total <- rowSums(tab[, 2:3])
tab$`Prevalence (%)` <- round(100 * tab[[2]] / tab$Total, 2)
knitr::kable(tab)| Browns fan | happy Yes | happy No | Total | Prevalence (%) |
|---|---|---|---|---|
| Yes | 263 | 53 | 316 | 83.23 |
| Explicit No | 6062 | 1857 | 7919 | 76.55 |
p_yes <- tab[[2]][1] / tab$Total[1]
p_no <- tab[[2]][2] / tab$Total[2]
c(prevalence_ratio = p_yes / p_no,
difference_percentage_points = 100 * (p_yes - p_no)) prevalence_ratio difference_percentage_points
1.087234 6.677779 The primary interval resamples respondents with replacement and retains all eligible days from each selected respondent. It uses 2,000 replicates, seed 20260828, and interpolated 2.5th and 97.5th percentiles. All 2,000 estimates were finite. The independent-respondent-day Katz interval is narrower (1.033–1.144) and is secondary because some people recur.
The clustered interval describes resampling uncertainty under independent respondent clusters. It does not account for common date shocks, selection bias, changes in presentation probabilities, or measurement error. Its lower bound is close to 1. Claims of a robust population difference would go far beyond this descriptive comparison.
| month | fandom_yes_n | fandom_no_n | prevalence_ratio |
|---|---|---|---|
| 2025-07 | 22 | 464 | 1.209 |
| 2025-08 | 22 | 602 | 1.163 |
| 2025-09 | 22 | 588 | 1.188 |
| 2025-10 | 26 | 595 | 1.095 |
| 2025-11 | 23 | 582 | 1.021 |
| 2025-12 | 26 | 584 | 1.067 |
| 2026-01 | 26 | 573 | 0.963 |
| 2026-02 | 15 | 551 | 1.121 |
| 2026-03 | 29 | 563 | 1.199 |
| 2026-04 | 22 | 553 | 0.996 |
| 2026-05 | 23 | 608 | 1.029 |
| 2026-06 | 24 | 579 | 1.101 |
| 2026-07 | 20 | 597 | 1.023 |
| 2026-08 | 16 | 480 | 1.074 |
Twelve of fourteen monthly ratios exceed 1, with a range of 0.963–1.209. Each month contains only 15–29 fan respondent-days; the endpoints are partial months. These are descriptive checks, not a test of season or game effects. Pooling does not adjust for changing composition or presentation tiers. Weekly comparisons for RQ3 and all-team rankings for RQ2 remain future work.
The narrow finding is that happy endorsement is more prevalent among observed Browns-fan respondent-days in this archive. Self-description should not be equated with clinical well-being, a stable trait, or an effect of winning games. The non-fan group can include fans of other teams and people who do not follow football.
Selection into the survey, recurrence, and co-presentation determine which respondent-days are represented. No known inclusion probabilities or weights for all U.S. Browns fans are available. Resampling does not turn the observations into a representative probability sample. Future work should compare equal respondent weighting and examine calendar composition before extending claims.
Per PI guidance, the primary comparison remains unweighted. The appendix uses eight age-by-sex cells (18–29, 30–44, 45–64, 65+, separately for Male/Female) from 2024 ACS B01001 (U.S. Census Bureau, 2025). All eight target counts were independently checked against the national summary-file row; the verification record is retained. The target is the U.S. adult resident population; the true distribution of Browns fans is unknown. Forty eligible days fall outside the complete numeric-adult-age/binary-sex coding, leaving 8,195 days.
| Method | PR | Difference (pp) |
|---|---|---|
| All eligible; unweighted | 1.087 | 6.678 |
| Complete cases; unweighted | 1.085 | 6.544 |
| Pooled age-sex calibration | 1.080 | 6.268 |
| Common-distribution standardization | 1.078 | 6.122 |
| sex | age_group | fandom_yes_records | fandom_no_records | pooled_poststratification_weight |
|---|---|---|---|---|
| Male | 18-29 | 28 | 542 | 1.470 |
| Male | 30-44 | 90 | 1723 | 0.596 |
| Male | 45-64 | 69 | 1157 | 1.019 |
| Male | 65+ | 9 | 248 | 3.304 |
| Female | 18-29 | 20 | 647 | 1.208 |
| Female | 30-44 | 34 | 1536 | 0.676 |
| Female | 45-64 | 51 | 1501 | 0.826 |
| Female | 65+ | 13 | 527 | 1.906 |
Pooled calibration gives each cell weight equal to its ACS adult share divided by its share of complete respondent-days. This yields PR 1.080 and a +6.27-point difference. Direct standardization averages each group’s cell-specific prevalence using the same ACS proportions, yielding PR 1.078 and +6.12 points. The latter asks about a hypothetical common demographic composition, not the observed U.S. fan and non-fan populations.
Fan cell counts range from 9 to 90; the largest pooled weight is 3.304. No uncertainty intervals are estimated for these exploratory weighted results. The similar point estimates show limited sensitivity to these particular adjustments, not robustness to unmeasured confounding or selection. Sex is used as recorded for calibration and is not recoded from gender identity.
Primary JSON, weighting JSON, acquisition checks, and primary-host subset JSON are downloadable. Source code, tests, and build instructions are in the public project repository.
| Input | SHA256 |
|---|---|
| Archive responses | ad9029600beed9f7a75fd08ff168f11624e87f86b30207fc1472077fa4d61449 |
| Archive demographics | 776967c16393f2fe54322274ac8b1948b76d306d0f04103ebdb8c628831421f4 |
This report was prepared by Aleph Initial Alpha, recovering and extending Ceetown’s estimator and interrupted report work. The current analysis was rerun from checksum-verified archive inputs. The historical August 28 JSON and draft are retained separately; they are not the current publication source.