← Steinbach City Council Reporting

1. What was measured

The unit of analysis is one councillor on one substantive agenda item: budget, land use, infrastructure, policy, service levels or major governance. Routine and ceremonial items are excluded. For each item every councillor present receives a record, including when they said nothing. Each record is scored 0–4 on five dimensions, or N/A when there was no fair opportunity to observe the dimension. The full rubric is in the methodology.

Headline measures. The reasoned-position rate is the share of eligible items on which a councillor took an identifiable public position (position score of at least 2) and gave at least one substantive reason (reason score of at least 2). Deliberation quality is 100 × Σ(score/4 × weight) / Σ(applicable weights), computed for records where the councillor actually spoke. Moving a motion counts as an explicit position; seconding alone or a recorded vote counts as an inferable position; a silent unanimous voice vote counts as no discernible position.

Attribution. Transcripts label voices (Speaker A, B, …), not names. Each meeting's coder first built a speaker map from the chair's hand-offs, self-identification and the minutes' mover and seconder. Votes or stances were never used to decide who spoke. Speech that could not be tied to one councillor is reported as “unidentified councillor” and is not credited to anyone.

2. The meetings

All 103 regular council meetings from January 2022 to September 2026 that have both a recording and minutes are included: a census, not a sample. They were coded in four rounds (2, 4 and 6 random meetings per year, then all remaining meetings), which allows the stability of the results to be checked across rounds. Meetings ran from 11 minutes to almost two and a half hours, with 1 to 11 substantive items each. Land use dominated (219 of 527 items), followed by budget and taxation (93) and infrastructure (89).

Attendance: Zwaagstra attended 102 of the 103 meetings, Jake and Bill Hiebert 96, Siemens and Damian Penner 93, Susan Penner 89. Rates are per eligible item, so absences do not lower anyone's score.

3. Council-level findings

• The procedural norm shapes the record. Steinbach council follows a consistent pattern: the chair names a mover and seconder, the mover explains the motion, and others are invited to add comment, usually answering “nothing further”. Low individual speaking rates therefore often reflect agreement with a rationale already given rather than an absence of judgment, but the rubric can only credit judgment stated aloud.
• Budget and governance items are the least deliberated in public. Budget and tax by-law readings averaged 1.7 speakers and governance items 1.5; nearly a quarter of budget items and over a quarter of governance items had no substantive councillor reason on the record. Fiscal deliberation likely happens at budget study sessions, which are not part of this record.
• Land-use hearings are where deliberation happens. 72 of the 127 items with a challenge on the floor and 17 of the 21 contested votes were land-use matters; five of the seven items on which every councillor spoke were land-use hearings.
• Debate widened slightly over the term. Items averaged 1.9 councillors speaking in 2022–2023 and 2.0–2.2 in 2024–2026; the share of items with two or more speakers rose from 54–58% to 60–66%. The change is modest.

4. Individual councillors

The Mayor chairs meetings and is reported separately (4.4): his procedural speech is not scored, and he receives a record only when he stated his own view on the merits, so his rates are not comparable.

4.1 Public reasoned-position rate

Intervals resample whole meetings (2,000 bootstrap draws), because a councillor's records within one meeting are not independent; they show how much a rate varies from meeting to meeting. Overlapping intervals should not be read as differences.

• Michael Zwaagstra has the highest reasoned-position rate (52%, interval 48–56%) and accounts for 29% of all councillor speech on substantive items. He spoke on a substantive item at 99 of the 102 meetings he attended, on 97% of the items he moved and on 32% of items he did not move, more than any colleague. Excluding the four meetings he chaired all or part of as Deputy Mayor, his rate is 51%.
• Jac Siemens (33%) is second. He has the highest independent-judgment average on council (2.9) and is the most frequent lone dissenter.
• Damian Penner (29%) and Susan Penner (27%) are close behind; both speak on most items they move (89% and 84%) but less often otherwise (22% and 16%).
• Jake Hiebert (23%) spoke on 68% of the items he moved (lower than his colleagues) and on 23% of others; he was the councillor most often in a losing minority (12 votes).
• Bill Hiebert (10%, interval 7–13%) has the lowest rate. He moved 21 of his 493 eligible items (colleagues moved 65–189), spoke on 29% of those and 13% of others, and spoke on a substantive item at 40 of the 96 meetings he attended. He is also the member most affected by speaker-identification merges with Jake Hiebert.

4.2 Quality of contributions when councillors spoke

Dimension columns are mean scores (0–4) over records where the councillor spoke and the dimension applied.

• When councillors spoke, contributions typically scored 2–3 out of 4 on reason-giving and independent judgment: relevant reasons with some causal link and some personal assessment, rather than either bare agreement or a full weighing of trade-offs.
• Zwaagstra (74) and Siemens (72) had the highest mean quality, followed by Susan Penner (69), Damian Penner (68), Jake Hiebert (64) and Bill Hiebert (57). The gap is mostly in reason-giving: 2.8 for Zwaagstra and 2.6 for Siemens versus 1.9 for Bill Hiebert.
• Constructive engagement was scored 370 times. Councillors usually acknowledged and answered residents' concerns (typical score 2–3); 36 contributions reached the top mark for fairly addressing the strongest counter-arguments.

4.3 Consistency across meetings

• Zwaagstra spoke on substantive items at 99 of 102 meetings, Siemens at 79 of 93 and Damian Penner at 75 of 93. Jake Hiebert (67 of 96) and Susan Penner (64 of 89) were silent at about a third of their meetings, and Bill Hiebert (40 of 96) at more than half.
• Results agree across the four coding batches (10, 20, 30 and 43 meetings) to within about 10 points per councillor, so no single batch drives the findings.

4.4 The Mayor (chair)

Records only where the Mayor gave his own view on the merits: 235 items across 90 meetings. Reasons given on 91% of them; mean quality 68.3. Dimension means: public position 3.7, reason-giving 2.5, independent judgment 2.6, constructive engagement 2.3, intellectual flexibility 1.9.

Mayor Funk stated his own view on the merits on 45% of substantive items, almost always with a reason (91%), usually as a closing summary after councillors had spoken; he was in a losing minority twice. His contributions score like a typical councillor's (quality 68), with intellectual flexibility (1.9) in the same low range as everyone else. Because he chairs and speaks selectively, his figures are not comparable with councillors'.

5. Representative quotations

For each person, the longest verified quotation from their median-quality spoken record, as the methodology's reporting rules require. Quotations are verbatim from the machine transcript and may contain transcription errors; the city's name is shown correctly as Steinbach, and residents named in them are shown by initials.

“The letter states that there's a— it's a conditional objection and basically raising concerns about sidewalks and walkways and asking either for confirmation that sidewalks sidewalks are already included or planned or that they will be added, that the plan will be changed to add them.”

Michael Zwaagstra · Public hearing and By-Law 2268 2nd & 3rd reading - rezoning Rosebrook Trail & Pacific Bay lots to RLD / RMD · 2025-12-16 · 0:13:25 · record quality 76

“It is also on a corner, so the curbing takes up a bit of that. There is a small portion of that front property that is at 25 feet. So— and it's an odd-shaped lot. It's a corner lot. I think it would work here with a single driveway as proposed.”

Jac Siemens · Variance V-2025-22 (116/118 Sterling Crescent, front yard setback 17 ft vs 25 ft RLD minimum) - public hearing and decision · 2025-09-16 · 0:15:18 · record quality 76

“And although I'm not going to say that Steinbach doesn't have serious crime— every community does have instances of it— a lot of the crime that we are seeing is, is small-scale and petty.”

Damian Penner · City Manager report on RCMP shortages - authorize agreement with private security firm, funded from policing budget surplus · 2025-06-17 · 0:12:42 · record quality 71

“as indicated by our administration, it'll just give our bylaw employees the power to issue penalties and It's obviously updating our penalties as well. So it's long overdue.”

Jake Hiebert · By-Law 2209 3rd reading - amending Administrative Penalty by-law · 2022-05-17 · 0:31:26 · record quality 65

“Yeah, I would just like to say I've had the opportunity to work with Jennifer a few times over the years, and she is a very competent woman, and she will be a great addition to that board.”

Susan Penner · Ratify appointment of J.F. to Jake Epp Library Board (two-year term) · 2024-03-05 · 0:59:46 · record quality 73

“I think SPC has played a vital part of this community for many, many years, as Mr. R. mentioned, and it's— I'm glad to see that we're moving forward.”

Bill Hiebert · By-Law 2250 public hearing and 2nd reading - rezone 333 Loewen Blvd from C4 to E/I for Steinbach Bible College · 2024-10-15 · 0:21:25 · record quality 58

“Do you feel it would be best to take this back, reword it, and then vote on it as a resolution later, once the wording is changed, or vote on it now and then you change the wording?”

Mayor Earl Funk · In Memoriam Policy - increase donation to $150 and direct to family's charity of choice · 2023-09-05 · 0:16:13 · record quality 71

6. Data quality and limitations

• Not yet human-reviewed. Every score was suggested by an AI coder following codebook v0.2. The methodology requires human review of disputed, low-confidence and publication-bound ratings; that has not happened. Seven records where a coder counted a deference-only remark (“X said it all”, “I concur with Councillor Penner”) as speech were corrected by the analyst to match the codebook.
• Reliability not measured on this coding. Two instances of the same AI model are not independent coders; human double-coding of 20–30% of records is still needed.
• Speaker attribution is good but imperfect. Diarization often merged the two Hieberts, or a councillor with staff or a delegate. 77 speech turns could not be tied to one councillor and are credited to no one, mostly Hiebert or Penner turns, so those members may be somewhat undercounted.
• Transcription incidents were caught and fixed. Two recordings were initially truncated by a download fault; both were re-transcribed in full and re-coded, and every transcript's length was checked against its YouTube recording.
• Minutes are sometimes inconsistent. From February to May 2023 the attendance lists omit Bill Hiebert, yet the recordings show him opening meetings and moving motions; coders treated him as present.
• Only open-meeting speech counts. Committee work, budget study sessions, closed sessions, constituent work and written questions to staff are invisible to this method. A councillor who does their deliberating elsewhere will score low here.
• Judgment calls in item selection. Delegations without a council direction and permit lists were excluded even where councillors commented; routine tax adjustments and minor statutory appointments were treated inconsistently across coders.

7. What still needs to happen

• A human reviewer should listen to the 77 unidentified turns and the flagged records, and double-code a random 20–30% of records, computing weighted kappa against the AI coding before relying on any individual figure.
• Councillors should be offered the chance to correct factual errors, as the methodology's reporting rules require. Corrections can be submitted here.
• Budget study sessions and committee-of-the-whole meetings, where fiscal deliberation appears to happen, should be added.
• Codebook v0.3 should settle identification by elimination and speaking order, the acting-chair case, deference statements and routine tax items.

Appendix: every substantive item

All 527 items with the number of councillors who spoke, whether any councillor gave a substantive reason, whether a challenge was on the floor, and whether any councillor opposed. Dates link to the start of the item in the recording where available. Residents are identified by initials.

Sources and reproducibility

• PJDI v1.0 rubric methodology (the instrument), operationalized in codebook v0.2.
• Transcripts: AssemblyAI speaker-diarized transcripts of the City of Steinbach's YouTube meeting recordings; official council minutes.
• Data workbook: every rating with its quotation, timestamp, link to the recording and the coder's rationale, so any figure on this page can be audited and corrected.