Welcome to the Faculty Bearometer

On Nov. 19, 2025, the Faculty Bearometer distributed the following question to 2,579 senate members.

To what extent, if at all, does ideological homogeneity within your department limit open and vigorous discussion of important issues in your discipline?

N=143

The Faculty Bearometer sent the following question on November 11, 2025 to 2,589 faculty senate members:

How, if at all, have you changed formative or summative assessment of your students because of generative AI?

N=113. Of the participants, 108 were regular faculty, 5 have emeritus status. 61 of the participants had a primary appointment in a STEM department. Members from 30 of our 35 unit/departments categories participated.

The Bearometer is an independent faculty project that poses a single poll question to Berkeley Senate faculty on issues of teaching and governance. It is unaffiliated with the Academic Senate or the administration and is modeled on MIT’s Faculty Pulse. Faculty propose and vote on questions, which are then selected by the Question Keepers (currently Chris Hoofnagle, Law; and William Fithian, Statistics) for distribution via university email. The Bearometer emphasizes anonymity to protect candid participation and typically receives responses from about 200 faculty per poll, providing rapid, faculty-driven feedback between formal Senate surveys. The Bearometer has run a dozen iterations on topics such as academic freedom, student preparedness, and faculty governance.

The Bearometer emphasizes anonymity to protect faculty speech, though this limits demographic representativeness.

Faculty speech is strongly protected by academic freedom, yet academics have many incentives to keep their views closely held. Thus, the Bearometer has both technological and procedural methods to ensure that any participant will not be identified. Anonymity is a fundamental feature of the Bearometer.

This anonymity leads to a problem: is the Bearometer valid? Or are its participants basically the same people every time?

On internal validity: we are confident that only Berkeley faculty are completing the Bearometer because we have an authoritative list of senate members, and because we distribute to email with unique links. We receive Qualtrics reports on possible fraud (typically someone has submitted twice) and these reports signal that double dipping is quite rare—typically just 1 or 2 submissions.

The unknown problem is representativeness: is the Bearometer a valid measurement of the faculty itself?

To answer this question, we examined response patterns from Bearometers 5-12 (the last 8 surveys, as we deleted earlier data consistent with privacy protections), focusing on our regular faculty (non-emerita).

In sum, this is what we found: Across eight surveys, 52% of regular faculty never participated. Of those that did, 60% have just done 1 or 2. Only 13% have done 5 or more.

Bearometer Participants

As one can see, most regular faculty haven’t done Bearometers 5-12 (we no longer have data for Bearometers 1-4). Of those that do, the Bearometer is not dominated by any small faction. Only 11 people have done all 8. 15 have done 7. 27 did 6.

Some of these Bearometers were quite popular. We have an overall 14.6% response rate from regular faculty based on Bearometers 5-12.

We also studied STEM versus non-STEM participation.

STEM and Non-STEM Participation

Berkeley has more STEM faculty (based on primary appointment). The different in participation between the groups is not significant.

There are no statistically significant differences across departments.

Few emerita participate in the Bearometer. Of the 1,016 in our distribution list, 824 have never completed a Bearometer. 100 have completed 1. Another 92 have completed more than 1.

What we do not know: we do not have data on sex, nor on pre/post tenure. Both of these variables, especially combined with departments, could undermine the anonymity of the Bearometer. But we also know that less powerful faculty may have the strongest speech concerns. Perhaps one might think that more senior faculty attitudes predominate, however, the Bearometer’s anonymity operates to protect these classes of faculty where other alternatives, such as the climate surveys, cannot because of climate surveys’ intense focus on demographics and combinatorial options.

The Bearometer has tradeoffs. It can rapidly and reliably get feedback from about 200 regular faculty in just days, using a single email invitation. Campus climate surveys can get higher participation, but only after weeks of recruitment, and these surveys come years apart (2009, 2011, 2019). In addition, the Bearometer is more democratic: the questions come from faculty members themselves. This helps give signals to decision makers free of the kinds of restraints on the senate and faculty administrators.

The Faculty Bearometer sent the following question on October 28, 2025 to 2596 faculty senate members:

What have you observed about how disability accommodations are implemented at Berkeley (e.g., benefits, costs, process, communication, fairness, inclusion, workload, small/large class dynamics)?

N=130 (4 submissions are blank). Of the participants, 119 were regular faculty, 11 have emeritus status. 70 of the participants had a primary appointment in a STEM department. Members from 31 departments participated.

The Faculty Bearometer sent the following question on October 14, 2025 to 2,614 faculty senate members:

What percentage of undergraduate students attend your lecture on any given day?

180 faculty answered this survey. The results are available at the link provided to faculty.

The Faculty Bearometer sent the following question on October 7, 2025 to 2,620 faculty senate members:

What do you enjoy most about being a faculty member at Berkeley? (open text response). 138 responded.

The Bearometer also sought votes for the following nominated questions (in randomized order). 114 responded.

• Should the faculty have opportunities to rate senior administrators? (likert)
• Ought the undergraduate colleges adopt a grade curve? (Scaled response: No, Yes, with X, Y, or Z distribution)
• How, if at all, have you changed formative and summative assessment of your students because of generative AI? (open text)
• What percentage of undergraduate students attend your lecture on any given day? (two digit numeric)
• How often have you “not done the right thing” at Berkeley because it would require cycles of your time spent on paperwork or compliance? (likert)
• How would you describe current practices around disability accommodation at Berkeley, based on what you have observed? (open text)

The Faculty Bearometer sent the following question on September 23, 2025 to 2,630 faculty senate members:

Has the use of student evaluations of teaching in faculty merit and promotion cases affected the difficulty of your course? If so, how? 238 responded.

Respondents could choose multiple responses:

• It hasn’t affected the difficulty of my course.
• Yes, it has led me to reduce course rigor/demand.
• Yes, it has led me to increase course rigor/demand.
• Yes, it has led me to award higher grades (more As or lenient distributions)
• Yes, it has led me to award lower grades (fewer As or stricter distributions)
• Other

All, here are some methods updates on the Bearometer for Fall 2025.

Our faculty changes, and so our N fluctuates. In September 2025, we added new faculty who have joined the campus this fall, and eliminated entries for those departed. Our N also decreases over time as faculty opt out of the Bearometer.

When we report units, we do so in an aggregated fashion. We aggregate units according to the table below so that all reported units have at least 50 members. Columns A and C are the most granular unit identifiers; these come directly out of CalNet and represent the respondent’s home department. To be clear, respondent identity is suppressed in Qualtrics; survey responses only carry the “Reported category” below.

We have also added an isStem flag based upon column A. We did this based on our own judgment. If you believe we should reclassify an entity, please email us. There are a few disciplines that could go either way, and departments that have subunits that are arguably stem, with others in social sciences and humanities.

Column B is what the Bearometer uses to report results in order to preserve respondent privacy. In some cases, we suppress results (e.g. Faculty administrators) because the number of participants is so small that reporting may identify the respondent.

Here are some interesting aspects of our categorization:

• We count 2,713 senate faculty members. However, our N will always be smaller because of email delivery errors, and opt outs.
• Our methods tag 1,414 members as STEM based upon their primary department categorization
• 1,017 have emeritus status
• This is our population breakdown by reported unit

The Faculty Bearometer sent the following question on April 29, 2025 to 2,564 faculty senate members:

Thinking about the students you teach, what portion of them are exceptionally prepared, adequately prepared, minimally prepared, or not at all prepared for the level of rigor you believe UC Berkeley coursework requires? 238 responded. The results are available at the link available to faculty.

On April 22, 2025, we distributed the following question to senate faculty (2567 members) via Qualtrics email contacts. 237 completed the survey. Results are available in the link circulated to senate faculty.

Over the past 12 months, approximately how much have you personally spent on work-related expenses without getting fully reimbursed, because of policy rules or because it wasn’t worth your time to seek reimbursement?

The choices were: $0-$50; $51-$250; $251-$1,000; $1,001-$5,000; >$5,000