Chicago City Council, 2003–2026

Technical appendix

Author

Boris Shor

Updated

October 18, 2026

This appendix documents the pooled analysis in the main report: 132 members, 693 contested roll calls and 32,980 observed binary responses, 2003–2026. The earlier analyses are preserved separately.

Source, identity and construction

The input is the corrected Chicago journal package downloaded October 16 from PR #32, with source coverage January 16, 2003–June 17, 2026. It contains 26,215 roll calls, 727 with both recorded yeas and nays. This is a journal-only delivery; the overlapping recent ELMS source is not added, preventing double ingestion. The earlier recent-only report uses a different source/window and is not simply appended to this matrix.

The preserved delivery SHA-256 is 1be2b79dd4481c00c1fbfd1b1ee78f307e4f50eb37c2670a73268eff6a19c08e. It was downloaded via rclone, checked for ZIP integrity and preserved separately from the prior delivery. Every file inside the ZIP matches the committed package manifest on repair branch commit 12e7271.

The published replay inputs were also downloaded and checked. The row-level author validator passes, and an independent local replay reproduces all 20 delivered files byte-for-byte.

Dated roster identities resolve people before constructing the vote matrix. This separates the Moore, Burnett, Mell and Colon/Colom successions and Milly Santiago from Miguel Santiago. Michael Chandler and Michael D. Chandler are an explicit same-person variant. Internal analysis keys do not allocate permanent IDs. Conflicted or missing registry IDs remain blank in metadata rather than defining a fictional person. Rows lacking a resolved roster identity, and explicitly flagged out-of-term rows, are excluded.

The upstream package restores 1,028 previously dropped responses across 1,019 ballots. Whole-file checks reproduce its response totals and find no duplicate roster-person response within an event. Thirty-nine events retain printed-versus-counted tally discrepancies; unresolved responses are not invented to close them. Our seven source-verified omissions and the two Scott/Solis misassignments are now corrected upstream, so no local vote additions or identity overrides are applied. The raw delivery remains unchanged.

Compared with the preceding prepared matrix, the new input adds 47 observed cells, changes none of the 32,933 already observed responses and removes none. The member and retained-event universes are unchanged. This is an input-conservation check, not a source-accuracy estimate.

Source roll-call key	Member	Upstream-restored response
2020:2020_06_17_1:9	Emma M. Mitts	Y
2017:2017_04_19_VI_VII(1)_2:66	George A. Cardenas	Y
2021:2021_10_27_1:0	George A. Cardenas	Y
2020:2020_06_17_1:4	George A. Cardenas	Y
2020:2020_01_15 (1)_1:21	George A. Cardenas	Y
2021:2021_07_21_VI_VII (2)_1:10	George A. Cardenas	Y
2021:2021_11_17_1:28	George A. Cardenas	Y

The culling rule retains members with at least four observed contested responses, and events with at least one observed yea and one observed nay. Both are rechecked until stable, yielding 132 members and 693 events. No new minimum-attendance threshold is imposed. Structural nonmembership and missing-not-listed responses remain unknown; neither is an observed absence or a nay.

Identical duplicate person/event responses would be reduced to one; conflicting responses would be excluded with a ledger. There are no remaining duplicate cells in the roster-resolved analysis input after the explicit identity corrections. Source event IDs are retained as package-scoped keys, not claimed as implemented permanent RC identifiers.

Model and estimation

The one-dimensional probit model is

$\Pr(y_{ij}=1)=\Phi(x_i\beta_j-\alpha_j).$

Each member has a constant pooled position xᵢ; each item has a discrimination βⱼ and difficulty αⱼ. Default pscl::ideal() priors have member variance 1 and item-parameter variance 25. Output is normalized, and posterior signs are aligned to a vote-derived singular-vector reference with Raymond Lopez positive. That convention does not fix his score or supply an external ideological calibration.

Exactly two chains use party starting values: Democratic −1, Republican +1 and unknown or conflicting/multiple designations 0. The completed party research supplies 113 Democratic starts, one Republican and 18 zero starts (15 Unknown and three with multiple/changing designations). Starting values enter neither the likelihood nor the priors.

Each chain uses 100,000 post-burn-in iterations plus 10,000 burn-in, thinned by 100; seeds are 20261015 and 20261016. No additional sensitivity fit or label-free start was run.

Diagnostics

Measure	Value
Chains	2
Post-burn-in iterations per chain	100,000
Burn-in iterations	10,000
Thinning	100
Maximum R-hat	1.0113
Minimum ESS	113
Correlation of chain means	0.999895

Root independently recomputed the diagnostics from the saved draws. No member has R-hat above 1.1. Agreement supports convergence within the sampled solution; it is not proof that no other posterior mode exists, that the source is error-free or that a single dimension captures every issue.

Historical and recent councils share many members. The overlap table records people with observed contested responses in adjacent years, not full service rosters:

Year	Next year	Shared observed members
2003	2004	50
2004	2005	50
2005	2006	50
2006	2007	47
2007	2008	50
2008	2009	50
2009	2010	48
2010	2011	46
2011	2012	50
2012	2013	49
2013	2014	50
2014	2015	49
2015	2016	50
2016	2017	50
2017	2018	50
2018	2019	49
2019	2020	50
2020	2021	50
2021	2022	50
2022	2023	49
2023	2024	50
2024	2025	50
2025	2026	50

Several members present only in the outgoing 2003 fragment have five to seven observations and wide intervals. Their apparent endpoint ranks are not precise ideological findings. All displayed council periods use one pooled position per member; changes in council composition are not individual ideological trajectories.

Party, caucus and election validation

Party research combines the existing upstream pilot, the 2003–2007 council review and supervised follow-ups. Four initially unresolved cases were already labeled under the same local ID but a different name. Subsequent research resolves thirteen of the final fifteen gaps, leaving Darcel Beavers and Leonard DeVille Unknown. The report has 126 Democratic-affiliated, one Republican, three multiple/changing designations and two Unknown. These latest labels enrich the report only: the fitted draws and the 113 Democratic, one Republican and eighteen zero initialization values described above are unchanged.

The new evidence includes candidate questionnaires, explicit reporting of party-organizational membership, a corporate PAC’s stated party classification, and reported convention participation. These are not all registration or original party records. Granato and Colom’s convention evidence is from 1996, Rice’s explicit reporting postdates council service, and other source-specific timing limits remain in the evidence ledger; a person-level designation does not imply verified continuous affiliation throughout local service. Cullerton’s evidence is his former campaign manager’s retrospective description, not Cullerton’s own self-identification. Mosley’s report is preserved through a third-party republication. No affiliation is inferred from a donor’s party, relatives, or roll-call behavior.

Evidence types distinguish party records, self-identification, reporting and party-organization alignment. Source dates are not automatically party start/end dates. UIC’s January 25, 2006 report supplies a 2004 snapshot; it does not establish continuous affiliation throughout later service. Fioretti’s Republican evidence postdates his council service. Uncited infoboxes alone remain leads. Research contribution and source justifications.

The 2024 UIC caucus comparisons classify only the 50 people covered by the roster. Historic ward holders and later successors are unclassified. The separately recorded 2027 DSA endorsement comparison uses only the 52 current-period members; older councillors are not coded non-endorsed from their absence in a later announcement. Tables show group medians and differences, with group sizes; no rank correlations or pairwise-ordering shares are added.

The presidential comparison preserves the previously validated NYT precinct aggregation and matches all 50 wards to the dated roster on November 5, 2024. Ward labels are offset from the points. These votes measure residents, not councilmembers. Prior county/city reconciliation is unchanged; the NYT major-party totals differ slightly from the certified canvass and are not silently adjusted.

Full estimates

Member	Researched designation	Observed votes	Position	95% lower	95% upper	R-hat	ESS
Daniel La Spata	Democratic	333	-2.536	-3.453	-1.759	1.000	206.745
Rossana Rodriguez Sanchez	Democratic	315	-2.408	-3.275	-1.667	1.007	258.032
Jessica Fuentes	Democratic	206	-2.035	-2.780	-1.411	1.010	551.814
Anthony Quezada	Democratic	79	-1.908	-2.974	-1.167	1.000	694.801
Leni Manaa-Hoppenworth	Democratic	205	-1.828	-2.514	-1.281	1.002	307.833
Maria Hadden	Democratic	315	-1.823	-2.436	-1.314	1.000	275.547
William E. Hall	Democratic	205	-1.820	-2.541	-1.244	1.002	418.707
Byron Sigcho-Lopez	Democratic	329	-1.818	-2.459	-1.306	1.003	325.850
John Arena	Democratic	233	-1.792	-2.265	-1.364	1.004	344.912
Angela Clay	Democratic	202	-1.732	-2.335	-1.204	1.002	415.613
Michael D. Rodriguez	Democratic	323	-1.731	-2.258	-1.257	1.001	545.689
Carlos Ramirez-Rosa	Democratic	392	-1.650	-2.085	-1.235	1.003	346.205
Julia M. Ramirez	Democratic	205	-1.590	-2.111	-1.142	1.000	433.236
Bob Fioretti	Multiple / changing designations	142	-1.578	-1.992	-1.194	1.009	465.510
Toni Preckwinkle	Democratic	107	-1.572	-2.087	-1.116	1.002	566.832
Matt Martin	Democratic	331	-1.452	-1.909	-1.074	1.001	409.679
Lamont Robinson	Democratic	201	-1.389	-1.889	-0.960	1.001	404.060
Jeylu B. Gutierrez	Democratic	206	-1.237	-1.638	-0.887	1.002	436.171
Dorothy Tillman	Democratic	52	-1.220	-1.733	-0.813	1.001	603.269
Andre Vasquez	Democratic	330	-1.166	-1.529	-0.854	1.000	266.081
Walter Redmond Burnett	Democratic	57	-1.104	-1.593	-0.701	1.001	520.728
Jeanette B. Taylor	Democratic	254	-1.070	-1.451	-0.727	1.005	386.460
Sandi Jackson	Democratic	65	-1.001	-1.450	-0.604	1.005	410.842
Arenda Troutman	Democratic	54	-0.984	-1.424	-0.599	1.001	520.640
Ricardo Munoz	Democratic	350	-0.745	-0.979	-0.506	1.011	189.370
Susan Sadlowski Garza	Democratic	260	-0.625	-0.958	-0.284	1.000	376.563
Ruth Cruz	Democratic	202	-0.621	-0.848	-0.395	1.000	174.284
Jason C. Ervin	Democratic	570	-0.607	-0.818	-0.393	1.001	118.418
Ronnie L. Mosley	Democratic	188	-0.578	-0.813	-0.340	1.000	184.960
Sophia King	Democratic	242	-0.573	-0.906	-0.250	1.000	506.906
Desmon C. Yancy	Democratic	196	-0.561	-0.795	-0.341	1.002	177.991
Leslie Hairston	Democratic	455	-0.559	-0.786	-0.326	1.000	227.165
Billy Ocasio	Democratic	79	-0.543	-0.886	-0.218	1.000	349.147
Toni L. Foulkes	Democratic	285	-0.459	-0.720	-0.206	1.002	217.909
Anabel Abarca	Democratic	16	-0.440	-1.635	0.506	1.000	1903.686
Joe Moore	Democratic	331	-0.417	-0.652	-0.172	1.004	185.807
Theodore Thomas	Democratic	58	-0.397	-0.781	-0.028	1.000	641.009
Scott Waguespack	Democratic	611	-0.373	-0.564	-0.168	1.000	171.406
Chris Taliaferro	Democratic	475	-0.348	-0.540	-0.154	1.001	143.723
Timmy Knudsen	Democratic	237	-0.323	-0.524	-0.120	1.000	148.275
Walter Burnett Jr.	Democratic	596	-0.321	-0.522	-0.126	1.000	113.521
Pat Dowell	Democratic	621	-0.303	-0.484	-0.118	1.001	162.094
Michael Chandler	Democratic	137	-0.292	-0.577	0.003	1.000	314.739
Stephanie D. Coleman	Democratic	317	-0.286	-0.474	-0.079	1.000	116.599
Ed Smith	Democratic	114	-0.278	-0.563	0.014	1.001	326.359
Harry Osterman	Democratic	361	-0.261	-0.524	0.013	1.002	245.589
Shirley Coleman	Democratic	49	-0.257	-0.758	0.383	1.001	878.617
Michelle A. Harris	Democratic	629	-0.224	-0.407	-0.035	1.000	112.852
Manuel Flores	Democratic	84	-0.215	-0.543	0.118	1.000	598.686
Sharon Denise Dixon	Democratic	58	-0.212	-0.603	0.189	1.000	467.970
Ameya Pawar	Democratic	232	-0.194	-0.502	0.130	1.001	269.469
Gregory I. Mitchell	Democratic	467	-0.189	-0.384	-0.001	1.000	129.904
Rey Colon	Democratic	192	-0.189	-0.459	0.087	1.002	240.884
Milly Santiago	Democratic	144	-0.178	-0.606	0.269	1.000	956.653
Bennett R. Lawson	Democratic	204	-0.162	-0.358	0.033	1.002	156.923
Emma M. Mitts	Democratic	626	-0.145	-0.341	0.052	1.000	153.943
Nicole Lee	Democratic	255	-0.125	-0.306	0.067	1.000	154.198
James Cappleman	Democratic	364	-0.120	-0.378	0.153	1.000	258.937
Deborah L. Mell	Democratic	195	-0.101	-0.456	0.321	1.003	593.052
Madeline Haithcock	Democratic	58	-0.094	-0.431	0.218	1.001	370.151
Roberto Maldonado	Democratic	354	-0.083	-0.360	0.205	1.000	293.882
Freddrenna Lyle	Democratic	108	-0.073	-0.353	0.186	1.000	340.221
Helen Shiller	Democratic	102	-0.061	-0.365	0.245	1.000	366.182
Roderick T. Sawyer	Democratic	349	-0.052	-0.342	0.230	1.001	279.049
Bill Conway	Democratic	202	-0.043	-0.226	0.150	1.000	164.519
David H. Moore	Democratic	470	-0.009	-0.200	0.188	1.000	131.335
Michele Smith	Democratic	326	-0.004	-0.318	0.309	1.001	430.225
Howard B. Brookins Jr.	Democratic	453	0.040	-0.202	0.282	1.005	222.307
Michael Scott Jr.	Democratic	233	0.057	-0.282	0.394	1.000	450.218
Monique L. Scott	Democratic	245	0.078	-0.097	0.267	1.000	126.711
Gilbert Villegas	Democratic	450	0.082	-0.108	0.264	1.000	141.141
Samantha Nugent	Democratic	332	0.156	-0.025	0.339	1.000	160.112
Derrick G. Curtis	Democratic	451	0.158	-0.026	0.345	1.000	163.381
Gene Schulter	Democratic	104	0.215	-0.171	0.640	1.003	636.283
George A. Cardenas	Democratic	441	0.242	-0.004	0.477	1.000	274.368
Brian Hopkins	Democratic	472	0.266	0.091	0.451	1.000	141.850
Matthew J. O'Shea	Democratic	564	0.271	0.090	0.458	1.002	147.286
Debra Silverstein	Democratic	529	0.272	0.093	0.455	1.000	144.654
Thomas M. Tunney	Democratic	471	0.308	0.088	0.533	1.000	193.550
Ariel E. Reboyras	Democratic	464	0.316	0.073	0.558	1.000	201.596
Proco Joe Moreno	Democratic	249	0.325	-0.064	0.725	1.006	955.471
Natashia Holmes	Democratic	50	0.329	-0.273	1.122	1.000	1263.189
Thomas Murphy	Democratic	51	0.365	0.033	0.779	1.001	507.294
Peter Chico	Democratic	205	0.366	0.189	0.565	1.000	162.169
William D. Burns	Democratic	110	0.395	-0.120	0.999	1.004	1365.723
Nicholas Sposato	Multiple / changing designations	559	0.417	0.233	0.611	1.002	187.369
Deborah L. Graham	Democratic	100	0.424	-0.057	0.966	1.003	1370.203
Felix Cardona Jr.	Democratic	327	0.437	0.256	0.629	1.001	147.166
Willie B. Cochran	Democratic	257	0.454	0.086	0.824	1.001	689.550
Brendan Reilly	Democratic	603	0.488	0.306	0.684	1.001	150.370
Brian Doherty	Republican	109	0.502	0.113	1.079	1.000	1103.006
JoAnn Thompson	Democratic	130	0.508	0.072	1.003	1.002	824.519
Patrick Daley Thompson	Democratic	225	0.514	0.194	0.817	1.001	327.315
Timothy M. Cullerton	Democratic	87	0.557	-0.043	1.285	1.001	1313.962
Anthony Beale	Democratic	639	0.569	0.385	0.767	1.001	195.180
Vi Daley	Democratic	113	0.605	0.224	1.090	1.000	749.990
Margaret Laurino	Democratic	354	0.631	0.347	0.945	1.002	310.465
Daniel Solis	Democratic	332	0.633	0.329	0.941	1.000	265.006
Patrick O'Connor	Democratic	322	0.637	0.332	0.961	1.000	446.159
Lona Lane	Democratic	144	0.641	0.187	1.136	1.000	732.937
Edward M. Burke	Democratic	430	0.689	0.464	0.919	1.003	269.757
Michael Zalewski	Democratic	316	0.690	0.393	1.008	1.001	423.782
Thomas Allen	Democratic	107	0.710	0.304	1.219	1.000	692.724
James Gardiner	Democratic	333	0.735	0.533	0.935	1.002	162.295
Richard Mell	Democratic	150	0.745	0.349	1.188	1.002	648.214
Silvana Tabares	Democratic	357	0.760	0.569	0.964	1.003	201.763
Mary O'Connor	Democratic	80	0.763	-0.067	1.633	1.009	1165.689
Raymond A. Lopez	Democratic	462	0.806	0.610	1.015	1.000	217.104
John Rice	Democratic	25	0.816	0.054	1.648	1.001	1680.539
Anthony V. Napolitano	Multiple / changing designations	478	0.825	0.633	1.029	1.007	167.706
Latasha R. Thomas	Democratic	186	0.835	0.418	1.334	1.001	672.127
Frank Olivo	Democratic	111	0.846	0.376	1.433	1.000	772.889
Marty Quinn	Democratic	573	0.902	0.705	1.109	1.001	156.982
John Pope	Democratic	202	0.951	0.523	1.415	1.002	582.907
Patrick Levar	Democratic	107	1.001	0.444	1.635	1.000	1027.714
Todd H. Stroger	Democratic	51	1.032	0.347	1.869	1.001	1228.656
Mary Ann Smith	Democratic	100	1.062	0.530	1.717	1.001	755.768
William Banks	Democratic	85	1.087	0.498	1.795	1.000	691.870
Virginia Rugai	Democratic	116	1.098	0.573	1.647	1.001	901.455
Isaac Carothers	Democratic	96	1.138	0.538	1.830	1.000	888.828
Ray Suarez	Democratic	193	1.186	0.623	1.722	1.000	515.749
Carrie Austin	Democratic	419	1.243	1.004	1.487	1.000	197.980
Bernard Stone	Democratic	112	1.644	1.019	2.309	1.005	557.043
Ted Matlak	Democratic	58	1.757	0.888	2.605	1.000	695.762
William Beavers	Democratic	53	1.796	0.775	2.794	1.001	514.652
Burton Natarus	Democratic	60	1.844	0.793	2.823	1.000	510.569
Jesse Granato	Democratic	7	1.882	0.348	4.927	1.001	482.786
Darcel A. Beavers	Unknown	5	1.950	0.093	5.034	1.001	1490.618
Leonard DeVille	Unknown	7	1.966	0.374	5.049	1.001	396.498
Vilma Colom	Democratic	7	1.983	0.341	5.131	1.002	337.688
Michael Wojcik	Democratic	7	2.007	0.348	4.844	1.002	488.755
James Balcer	Democratic	202	2.078	1.567	2.708	1.000	438.964

Reproduction and retained versions

Run from the Local project root:

Rscript Code/66_prepare_chicago_2003.r
Rscript Code/67_fit_chicago_2003.r
Rscript Code/69_chicago_2003_report.r
quarto render chicago-report.qmd
quarto render chicago-technical-appendix.qmd

The corrected raw delivery, replay inputs and checks are under Jacob/2026-10-16-chicago/; prepared data, chains, scores, audit tables and exact session information are in Results/Chicago-2003-oct16/. Research inputs are under Data/Chicago/party_research_2003/, Data/Chicago/party_research_remaining_2026-10-15/, Data/Chicago/party_research_final15_2026-10-16/ and Data/Chicago/party_research_final10_2026-10-18/. Code/70 consolidates the last follow-up; Code/69 applies it to report outputs without changing the fitted draws. The preceding pooled fit remains in Results/Chicago-2003/ and its pages in Results/chicago-report-before-oct16/. Recent-only results and archived pages are also preserved.

Historical repeated-pattern sensitivity and alternative starts from the earlier recent-only appendix are not newly completed checks of this expanded input.