First decision: who counts as the public
A survey of residents and a survey of registered voters are different studies, and they can disagree. Residents include everyone who lives in the city: renters, non-citizens, people who have never voted. Registered voters are the subset the electoral system recognizes, and they skew older and more established almost everywhere.Which frame is right depends on the decision. A referendum, a bond question or anything heading to a ballot should be weighted to the electorate that will decide it. Service delivery, parks, code enforcement and communications reach are questions about everyone who lives there. For the City of Pembroke Pines we weighted to the city’s active registered voters by age, gender, race, party and district, because that program was built to inform decisions heading toward a ballot. That choice is stated on the dashboard rather than buried in an appendix.
Ask any firm you are considering which frame they intend to use and why. If the answer is not immediate, the study has not been designed yet.
The questionnaire is where most surveys go wrong
A leading question produces a number that is real, repeatable and worthless. We draft the instrument with the people who will have to act on the results, then test it for order effects, double-barrelled items and the quiet assumption that residents already know what a department does.Length is a budget decision as much as a design one. Every additional question costs completion rate, and a survey that people abandon halfway produces a sample that is no longer the population you set out to measure.
We publish the full instrument with the results. Anyone can read what was asked, in what order, and check that the wording supports the finding.
Reaching residents who are not already listening
The easiest people to survey are the ones already subscribed to city email, and they are not the city. A study built only on the existing list measures the people already paying attention and calls it public opinion.We field in English, Spanish, Haitian Creole and French. In South Florida that is not a courtesy; it is the difference between a representative sample and a sample of one part of the city. Translation is done for meaning, not word-for-word, because a literal rendering of a survey item frequently changes what is being asked.
For Pembroke Pines the two waves drew 3,569 responses from 2,876 unique residents, nearly three times the committed minimum of 1,200 across both surveys.
Weighting, and the honest error band
Raw counts from any survey are not what the population thinks; they are what the respondents think. Weighting corrects for who answered against who lives there. It is a modeling step, and it should be described, not hidden.What comes back is a modeled estimate with an error band, not a fact. Pembroke Pines Wave 1 was n=1,933 with a modeled margin of plus or minus 2.9 points. That band sits beside the number on the dashboard, so a two-point gap is never read as a result when it is not one.
When a result is close enough that we cannot call it, we say so. When the honest reading is that a problem is about cost or trust rather than messaging, we say that too.
District analysis, and what it can and cannot support
A citywide number can hide two districts that disagree completely. Cutting results by district is usually the most useful thing a community survey does for a commission, because it turns one average into a map of where a policy is actually contested.It also costs precision. A district subgroup is a fraction of the full sample, so its error band is wider than the citywide one, and small differences between two districts often cannot be distinguished. We size the sample for the district cuts the city actually intends to use, and we mark the cuts that will not support a confident read.
What arrives: a dashboard, not a PDF
Most polling is handed over as a deck of toplines and crosstabs. We build an interactive dashboard instead, so the commission, staff and anyone the city chooses to share it with can move through the findings themselves, question by question and district by district.The Pembroke Pines dashboards are public and are the honest example of what a city receives: key findings, an interactive district map, district comparison, the full topline, written answers, and a methods section. The case study carries the dates, sample sizes and sources.
A study appears publicly only once the client clears it for release. Some cities publish everything; some publish nothing. Both are normal, and it is the city’s decision, not ours.
Common questions
What does a community survey cost?
It is driven by four things: how many residents you need to reach, how many languages you field in, questionnaire length, and how finely you need to cut the results. District-level analysis is usually the largest single driver, because supporting confident district cuts requires a larger sample than a citywide read does. We scope against the decisions the survey has to inform rather than quoting a standard package.
How long does a community survey take?
The fielding window is the short part. For Pembroke Pines, Wave 1 was in the field July 6 to 9, 2026 and Wave 2 from July 21 to 24. Questionnaire development, translation, weighting and dashboard build sit either side of that, and the schedule is set by when the city needs the findings in hand for a decision.
Should we survey residents or registered voters?
It depends on the decision. Anything heading to a ballot should be weighted to the electorate that will decide it. Questions about services, facilities and communications reach are about everyone who lives in the city. The two frames can produce different answers to the same question, which is why the choice is made before the instrument is written.
Can we survey in languages other than English?
Yes. We field in English, Spanish, Haitian Creole and French.
Do the results have to be published?
No. Publication is the client's decision. When a study is released, we publish the sample, the frame, the weighting and the error band beside the numbers, and the full instrument alongside them.