These come from syntheses and trials on organizational change generally, not only middle managers, graded with the same rubric as the Atlas studies.
Molan et al. 2018 · cluster randomized trial, 49 small units of one disability-services nonprofit (N = 430) · rated High strength
Employees rated their service quality lower than families did, and staff who got their survey results back rated the relational side of their own service higher. Rated from the abstract, so the grade is provisional.
The Atlas cases show gaps between how different levels see a change; this trial measured a different gap, between staff and the families they serve, and found that feeding survey results back shifted staff perceptions. The outcome is perception, not change progress.
An et al. 2022 · randomized field experiment, 672 leaders randomized (504 remained) · rated Medium strength
Across all leaders, leadership training did not significantly narrow the gap between how leaders and their employees saw the leaders' behavior; the gap narrowed only in public organizations and tended to widen in private ones.
It supports the Atlas advice to ask how managers and staff describe things, since leaders' own view can stay out of step even after training.
Kluger & DeNisi 1996 · meta-analysis of 607 effect sizes (N = 23,663) · rated Medium strength
Feedback interventions improved performance on average (d = .41), but more than a third of them decreased performance, and feedback was less effective when it drew attention away from the task and toward the self.
It is about feedback on individual task performance, not about spotting a stuck change; it serves as a caution that returning feedback helps on average but not reliably, which sits alongside the one Atlas study of when survey feedback changed practice.