Research digest · 27 Oct 2026
Nine recent papers and one industry survey, chosen for what they say about advice and oversight, and about the people who hold office. Two concern the chatbot in the room, and they point in different directions.
1. The flattering assistant moves people least
DECISION-MAKING · AI
Conlon and Schwardmann ran a preregistered experiment with about 1,500 US adults and 10,000 conversations across 30 decision tasks. Participants stated a leaning, then decided alone, with a standard chatbot, or with one prompted to agree more. The standard model sided with their leaning in about 62% of the points it raised, yet still moved choices towards the middle by 0.22 standard deviations. The more sycophantic version barely moved anyone, while confidence rose with either.
For Sentry red-teaming: agreeable advice leaves a principal surer and no better informed. A working paper, not yet peer reviewed.
Conlon JJ, Schwardmann P (2026) AI sycophancy and decisions. Working paper, arXiv 2607.28133; CEPR DP21505.
https://arxiv.org/html/2607.28133v1
2. Boards wait too long after an interim chief, and are partly right to
SUCCESSION · BOARDS
Langan studied chief executives who succeed an interim chief. They are much less likely than successors to permanent CEOs to be removed early for poor performance. Directors appear to be sparing the firm further disruption rather than protecting their own reputations, and the caution has some basis: when these successors do leave early, market performance declines afterwards.
An interim appointment is often presented to a board as the prudent option. The paper gives succession work a sharper question: what will the board do if the permanent choice disappoints in year one, and has it decided that in advance?
Langan R (2026) Are boards reluctant to remove poorly performing successors to interim CEOs? Strategic Management Journal 47(4).
https://doi.org/10.1002/smj.70041
3. A well-connected board takes the edge off an overconfident chief
BOARDS · SENTRY
Xu and colleagues followed S&P 1500 firms from 2002 to 2018 and measured each board's position in the network of interlocking directorships. Overconfident chief executives made more acquisitions. The link weakened when the board was central in that network, and weakened further when directors were closely tied to one another and when more of them were women. Outside information helps only if the board has the habits to share and test it.
For work on board fractures, the paper separates two things often merged: a board that hears a lot, and a board that argues well with what it hears.
Xu J, Li L, Liu H, Hou W, Zhu Q (2026) Mitigating the effects of CEO overconfidence: a group information processing perspective. Corporate Governance: An International Review 34(4): 897-921.
https://doi.org/10.1111/corg.70044
4. Psychological screening tells you little about the future insider
SENTRY · INSIDER RISK
Lilja-Lolax, Alm and Annell linked 650 Swedish personnel security investigations (2010 to 2023) to selection assessments taken on entry to military training, against two control groups of 650. Higher cognitive ability and suitability for command were associated with somewhat lower odds of later investigation, with odds ratios around 0.63 to 0.75 in some models. Psychological level of function predicted nothing. The models explained only about 6% to 16% of the variance.
Clients sometimes want a test that flags the risky hire. This registry study says a one-off assessment cannot carry that weight on its own.
Lilja-Lolax K, Alm C, Annell S (2026) Psychological assessments predict insider threat risk-mitigation measures, but only marginally. Frontiers in Psychology 17.
https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2026.1926403/full
5. Fear, more than anger, is what doxing produces
FAMILIES · SECURITY
Thatcher, Stäcker and Hett asked 300 Americans to imagine being nominated to a school board, then showed each a personalised doxing scenario built from their own views on abortion and details they had shared. Fear was more common than anger, especially when the exposure felt severe. Frightened participants were less willing to stand or to discuss politics online, while anger sometimes kept people engaged. Those with weaker views were the most likely to withdraw.
Principals and family members who take public positions face the same mechanism. Exposure planning should address the fear response, since that is what drives retreat.
Thatcher JB, Stäcker A, Hett J (2026) Fury or fright: politically motivated doxing and its effects on doxees' civic engagement. MIS Quarterly.
https://doi.org/10.25300/misq/2026/18971
6. AI improves the brief, and leaves accuracy where it was
LEADING COUNSEL · AI
Schwarcz and colleagues randomised 137 upper-level law students to six realistic legal tasks, two each with no AI, a reasoning model (o1-preview) and a retrieval-based legal tool. On a seven-point scale, quality rose about 0.53 points with the reasoning model and 0.27 with the legal tool, and time used fell by roughly 20% to 28% on five tasks. Clarity and organisation improved. Accuracy did not, and the reasoning model produced 11 hallucinations against 4 without AI.
Senior counsel reviewing AI-assisted drafts should expect cleaner prose and still check every citation.
Schwarcz D, Manning S, Prescott JJ, Barry P, Cleveland DR, Rich B et al (2026) AI-powered lawyering: AI reasoning models, retrieval augmented generation, and the future of legal practice. Journal of Law and Empirical Analysis 3(1): 220-250.
https://doi.org/10.1177/2755323X261427048
7. One in four national squad athletes sleeps badly enough to matter
PREP · ELITE SPORT
Vorster, Schmid and Erlacher surveyed 2,293 members of Swiss junior and elite national squads, drawn from up to 100 sports, and mapped their answers onto clinical sleep disorder categories. A quarter (25.8%) reported clinically relevant symptoms, women (29.9%) more often than men (22.3%). Parasomnias, insomnia and restless legs were the commonest. About one in five had poor sleep efficiency and 7.5% slept under six hours. These are screening answers, not diagnoses.
The figures support routine sleep screening in Prep for athletes, and the same short questionnaire would suit principals who travel heavily.
Vorster APA, Schmid MJ, Erlacher D (2026) Sleep health and sleep disorders in Swiss elite athletes. Discover Mental Health 6(1): 131.
https://pmc.ncbi.nlm.nih.gov/articles/PMC13346400/
8. Sleep interventions lift vigour and ease tension in athletes
PREP · SLEEP
Walsh and colleagues reviewed 21 sleep or circadian interventions in athletes and meta-analysed 13 full-text studies covering 558 participants, mean age 21. Positive affect rose relative to negative affect. Tension fell (beta about -0.71) and vigour rose (about 0.72), both with reasonably tight intervals. The anxiety effect was borderline, and results for depression and anger were inconsistent. Most studies were small, carried some risk of bias, and included few women.
Modest evidence, but the right kind for Prep: sleep work changes how a person feels on the day, and tension is where it shows.
Walsh E, Leota J, Huynh M, Liddle SK, Drummond SPA, Facer-Childs ER et al (2026) Sleep and circadian interventions to improve athletes' mental health, mood and well-being: a systematic review and meta-analysis. Sports Medicine 56(4): 997-1011.
https://link.springer.com/article/10.1007/s40279-025-02387-z
9. What the successor inherits decides how far the successor travels
SUCCESSION · FAMILIES
Del Barone and colleagues studied 16 Dutch family-owned food firms through 39 interviews with successors, incumbents and long-serving staff. They found four patterns after handover, from successors who kept the founder's approach to those who broke from it. The pattern depended on how two things met: what the incumbent had imprinted, in routines and in behaviour learned at close quarters, and the successor's own sense of autonomy, competence and belonging. Strong imprinting held back successors whose self-determination was weak.
A useful frame for the trainee guide: assess the successor's motivation as carefully as the plan.
Del Barone L, Annosi MC, Micelotta E, Buonocore F (2026) Post-succession innovation in family businesses: exploring the tension between incumbent imprinting and successor self-determination. Entrepreneurship Theory and Practice 50(2): 502-535.
https://doi.org/10.1177/10422587251382824
10. Family offices name successors and rarely prepare them
FAMILIES · INDUSTRY REPORT
Industry report. Bank of America Private Bank and Escalent surveyed 355 US family office decision makers in early 2026. About a quarter had changed leadership in the past decade, and only 37% of those said the incoming principal was very prepared. Some 83% had identified a successor, but only 27% had a trust distribution policy. Nearly all (91%) said family education keeps families aligned, yet 40% ran a formal programme. Family dynamics were the top governance challenge, cited by 45%.
The gap between naming a successor and readying one is where family-office work sits.
Bank of America Private Bank (2026) Bank of America Family Office Study. Industry survey, conducted by Escalent.
Summaries are paraphrased; the working paper and the industry report are flagged as such.