The paradox of withdrawal: How managerial turnover intention and attitudes shape performance appraisal quality in high-contact service

Purpose Performance appraisal quality depends on the discretionary effort of the rater. This study examines rater-side psychological antecedents of appraisal rigor among hospitality managers, focusing on how managers’ positive attitudes toward the job and the organization drive appraisal quality and whether that relationship is disrupted by the managers’ own turnover intention. Design/methodology/approach Data were collected from 318 U.S. hospitality managers through a multi-stage screening protocol on Amazon Mechanical Turk. Because job satisfaction and organizational commitment were empirically indistinguishable in the sample (r = .976), a higher-order Managerial Attitudes factor was specified to test the two attitudinal hypotheses jointly. Structural Equation Modeling (SEM) was used to estimate direct and moderating effects. Common method variance was addressed procedurally and through a marker-variable technique combined with an unmeasured latent method construct. Findings Managerial Attitudes had a strong positive relationship with self-reported appraisal quality (β =.71 in the method-adjusted model). Contrary to the “withdrawal–negligence” prediction of Conservation of Resources (COR) theory, turnover intention showed a small but significant positive direct association with appraisal rigor, and its interaction with attitudes was not statistically significant. The pattern is consistent with a “Lame Duck Objectivity” account in which anticipated exit releases managers from certain political constraints on evaluative candor, but reverse causality, statistical power, and measurement limitations remain plausible alternatives that the design cannot rule out. Originality/value The study reframes managerial withdrawal as a scope condition of COR rather than a uniform trigger of task disengagement, and it demonstrates the utility of a higher-order attitudinal factor for handling discriminant-validity breakdown in hospitality data. Boundary conditions on generalizability — including racial and gender homogeneity of the sample, single-source measurement, and cross-sectional design — are discussed in detail, and a program of multi-source, longitudinal, and gender- and race-balanced replications is proposed.