2026 judicial retention · Maricopa voter scope
Your ballot.
Your priorities.
Who should stay on the bench? Explore what people who interact with these judges say—and decide how much each perspective matters to you.
Before you explore
Useful information.
Check the original.
This independent site is for information only, not an official election resource or a recommendation to retain or remove a judge.
AI helped extract these numbers from difficult-to-read, inconsistently formatted PDFs. AI makes mistakes. Checks and spot checks can catch some errors, but they do not guarantee that every value is correct or complete.
Each judge’s card links to the full original report, and the detailed tables link to source pages. Read those reports to verify important details and see the full context. About the data and calculations ↓
Check your own ballot. This collection follows the state pamphlet. Mark H. Brain appears there but is absent from the supplied county ballot list; his ballot status remains unresolved. Data snapshot: October 7, 2026.
02 / Explore the feedback
The judges
N: how many answers are behind the feedback?
N counts valid answers to one question from one respondent group. For example, N = 8 for lawyers means eight lawyers answered that question. It is not the total number of people across a judge’s report.
A section or overall result shows a range, such as N 8–40: the smallest and largest question/group samples contributing to it. The same person may answer several questions, so we do not add those Ns together.
! Includes N < 10 At least one contributing sample has fewer than 10 answers.
With fewer answers, one person’s response can have more influence. These colors flag sample size, not judge quality. Ten is our display cutoff, not a guarantee of reliable or representative feedback.
Removes each question/group result with fewer than 10 valid answers; N = 10 stays included. Remaining groups receive the weight for that question. Questions with no remaining answers are omitted. Percentages and coverage may change because different evidence contributes. Missing answers never count as zero ratings.
The context behind the numbers
A tool for your judgment.
What your rollup means
For each question, each rating category’s count is divided by that group’s valid n. We combine these percentages using your respondent weights, then average the selected questions equally. Sections roll up their selected subquestions; a section is never counted again as a separate question.
Overall and section higher-category feedback percentages are the average share in the higher two source rating categories. They use the same distributions shown in the charts. These are custom average response distributions, not percentages of unique people, official section summaries, official JPR scores, or voting recommendations. Missing groups are excluded within each question; questions with no usable selected groups are omitted. Coverage is the fraction of requested question × group weight with usable data. Different coverage can affect comparisons.
Read the sample size
n is the number of valid answers to one question from one group. Rollups show its range, never a pooled respondent total. “Small sample” means n < 10, a display flag we chose—not an official cutoff or a poor-performance label. Missing answers are not zero ratings. The optional N filter removes question/group results below 10, redistributes weight among remaining groups within each question, and omits questions with none remaining. Coverage keeps the original selection as its denominator.
Groups may be missing or answer different questions. Coverage accounts for both your questions and respondent weights. Compare the group details before interpreting an ordering.
Understand the source
AI helped extract data from difficult-to-read PDF tables. Extraction and interpretation errors are possible despite automated checks and limited manual spot checks. This site is for information only. Use each judge’s original report to verify figures and read the full context.
These voluntary surveys may not represent everyone who interacted with a judge. New surveys use a 1–5 scale; legacy surveys use 0–4 with different labels. They remain separate. Presiding-judge surveys are a separate instrument.
Official commission findings are shown separately. No numerical commission vote tally was supplied. The extracted data includes source inconsistencies; row warnings are preserved in the details.
Chart design informed by Steve Wexler’s Likert-chart work: separate the midpoint, retain every rating category, and use consistent axes. Acceptable and Satisfactory are source labels, not “neutral.”
Official JPR reports ↗ · State publicity pamphlet ↗ · Download the site's source data