Pierce and Aguinis (2013) proposed the too-much-of-a-good-thing (TMGT) effect as a meta-theoretical principle for management research: antecedents widely regarded as monotonically beneficial are claimed to reach context-specific inflection points beyond which their relations turn asymptotic and often negative. The article itself left the framework's scientific core unresolved: it offers no general method for predicting or locating inflection points across domains, no rule for deciding when high levels of a construct merely saturate benefits versus reverse them into harm, an incomplete account of the moderators that shift inflection points and slopes, acknowledged obstacles of measurement, range restriction, and statistical power to valid testing, and no criterion for when a construct at extreme levels becomes a different construct. The open core — a predictive, falsifiable theory of TMGT relations with a valid cross-domain testing apparatus — remains open.

Background

A long-standing default in management and organizational research is that more of a beneficial antecedent is better: higher leadership quality, more conscientiousness, richer job design, more planning, more organizational slack (absorbed, uncommitted resources a firm can deploy) are each assumed to improve outcomes monotonically. Against this default, a body of empirical findings showed the same antecedents producing neutral or negative outcomes at high levels. Grant and Schwartz (2011) synthesized the psychology analog, arguing that positive traits, states, and experiences follow inverted U-shaped relations (benefits rise, reach a peak, then decline), echoing Aristotle's idea of virtue as a mean between deficiency and excess. Pierce and Aguinis (2013) gave the management version a name and a meta-theoretical status: the too-much-of-a-good-thing (TMGT) effect, the claim that all seemingly monotonic positive relations reach context-specific inflection points (the level of the antecedent at which the relation's slope changes sign — the turning point, the maximizer of the inverted U and the vertex of a quadratic — as distinct from a mathematical inflection, at which concavity changes and the slope need not change sign, and from the saturation level, the level beyond which a merely asymptotic relation levels off without reversing) after which the relation turns asymptotic (benefits level off, or saturate) and often negative, producing overall curvilinearity. Where the items below speak of locating, estimating, power-analyzing, or moderator-shifting an inflection point, the object is the turning point when the relation reverses and the saturation level when it merely saturates. They illustrated the principle across organizational behavior (leadership, personality), human resource management (job design, personnel selection), entrepreneurship (new venture planning, firm growth rate), and strategic management (diversification, organizational slack), and argued that TMGT undermines the validity of linearly specified models.

The framework's statistical workhorse is the quadratic (squared-term) regression, including its moderated form, which Pierce and Aguinis (2013, p. 326, Equation 4) recommend for estimating how a moderator (a variable that changes the strength or form of a relation) shifts the inflection point of an inverted U. But the article explicitly defers the hard scientific work: theory development, not the article, should predict where inflection points lie; future studies should estimate them with existing empirical tools; it remains undecided when a relation saturates rather than reverses — the article offers only a tentative classificatory expectation, following Aristotle's taxonomy, that predictors good in themselves (e.g., general mental ability, wisdom) show asymptotic relations while predictors classified as actions or passions show inverted U-shaped ones, leaving open when each pattern is likely to emerge; the moderators that move inflection points and slopes must be worked out domain by domain; measurement problems, range restriction (the attenuation of observed relations when a sample covers only a narrow range of the antecedent, so that extremes near the inflection point are unobserved), and low statistical power (the ability of a design to detect a true nonlinear effect) are flagged as obstacles to valid tests; and it is left open whether a construct taken to an extreme is still the same construct or borders on a different one, even psychopathology.

Neighboring methodological work sharpens these difficulties. Lind and Mehlum (2010) supplied a formal test (building on Sasabuchi's procedure) for whether a regression's extremum lies inside the data range. Simonsohn (2018) showed that testing U-shaped hypotheses with quadratic regressions can produce near-certain false positives (for example, declaring logarithmic growth U-shaped) and proposed a two-lines procedure that estimates a breakpoint and two regression segments instead. Haans, Pieters, and He (2016) reviewed how strategy research theorizes and tests U-shaped relations, distinguishing moderation that shifts or tilts the curve from moderation that flips it entirely. Busse, Mahlendorf, and Bode (2016) reframed TMGT through competitive mediation — an antecedent simultaneously activating a benefit-generating and a cost-generating mechanism whose net effect is curvilinear — and argued TMGT is frequent but not omnipresent. Arin, Minniti, Murtinu, and Spagnolo (2022) contributed a statistical framework for detecting inflection points, kinks, and jumps endogenously, as a precursor step before choosing an estimation model. None of these, individually or jointly, supplies what the TMGT agenda still lacks: a way to predict, before data are collected, whether and where a given beneficial antecedent stops helping.

Problem Statement

The too-much-of-a-good-thing (TMGT) effect is asserted as a meta-theoretical principle: every seemingly monotonic positive relation between an antecedent and a desirable outcome reaches a context-specific inflection point beyond which the relation turns asymptotic and often negative. Pierce and Aguinis (2013) left the principle's scientific core unresolved, and the open problem is to convert it from a descriptive meta-theory into a predictive, falsifiable theory with a valid testing apparatus. Concretely, for antecedent-outcome relations in management and organizational research (spanning organizational behavior, human resource management, entrepreneurship, and strategic management, at micro and macro levels of analysis), the article's future-research agenda leaves the following unresolved:

(i) a general theoretical method for predicting, before data are collected, whether a given beneficial antecedent's relation with a given outcome contains an inflection point at all, and where that context-specific inflection point lies, together with validated procedures for estimating its location and uncertainty from data; (ii) a rule for deciding whether the relation beyond the inflection point merely saturates (asymptotic benefits) or reverses into harm (inverted U), that is, for understanding when each pattern is likely to emerge; (iii) an account of the moderators of these relations — which variables shift the inflection point and which change the slopes on either side of it — and the circumstances under which each operates, worked out across rather than within isolated domains; (iv) the measurement, sampling, and analysis conditions under which TMGT relations can be validly detected in real data, addressing the obstacles the source article flagged: measurement quality at extreme levels, range restriction that hides the inflection region, the trade-off between the false positives of quadratic-regression tests and the power of flexible alternatives, and statistical power as a function of inflection-point placement;

and (v) a criterion for deciding when a construct measured at extreme levels should be reclassified as a different construct (or a clinically distinct phenomenon) rather than the same mechanism taken too far; the article raises this question but considers the TMGT useful as a meta-theoretical principle for describing and explaining relations involving such a construct regardless of how the question is resolved.

The five matters (i)–(v) are jointly posed; the source's remark that TMGT remains useful for describing and explaining relations involving such a construct, however the question is answered, does not put (v) outside the agenda.

The four domains named above are illustrative of the field in which the method and account must hold rather than an exhaustive target set: a general result holding beyond them is required. For (i), "validated procedures" means estimation procedures for the turning point and its uncertainty whose validity is established from data.

A resolution of the agenda gives, for each item, a domain-independent result with its conditions of application and a supporting argument rather than a mere enumeration of possibilities. These five matters are interdependent — the moderator account (iii) presupposes the method (i) and the rule (ii) — so they are resolved as one agenda rather than as separable results.

References

  1. Pierce, Jason R., and Herman Aguinis (2013). The Too-Much-of-a-Good-Thing Effect in Management. Journal of Management 39(2):313-338. DOI: 10.1177/0149206311410060. URL: https://doi.org/10.1177/0149206311410060
  2. Grant, Adam M., and Barry Schwartz (2011). Too Much of a Good Thing: The Challenge and Opportunity of the Inverted U. Perspectives on Psychological Science 6(1):61-76. DOI: 10.1177/1745691610393523. URL: https://doi.org/10.1177/1745691610393523
  3. Lind, Jo Thori, and Halvor Mehlum (2010). With or Without U? The Appropriate Test for a U-Shaped Relationship. Oxford Bulletin of Economics and Statistics 72(1):109-118. DOI: 10.1111/j.1468-0084.2009.00569.x. URL: https://doi.org/10.1111/j.1468-0084.2009.00569.x
  4. Simonsohn, Uri (2018). Two-Lines: A Valid Alternative to the Invalid Testing of U-Shaped Relationships with Quadratic Regressions. Preprint, SAGE Advance. DOI: 10.31124/advance.7222010. URL: https://doi.org/10.31124/advance.7222010
  5. Haans, Richard F. J., Constant Pieters, and Zi-Lin He (2016). Thinking about U: Theorizing and testing U- and inverted U-shaped relationships in strategy research. Strategic Management Journal 37(7):1177-1195. DOI: 10.1002/smj.2399. URL: https://doi.org/10.1002/smj.2399
  6. Busse, Christian, Matthias D. Mahlendorf, and Christoph Bode (2016). The ABC for Studying the Too-Much-of-a-Good-Thing Effect. Organizational Research Methods 19(1):131-153. DOI: 10.1177/1094428115579699. URL: https://doi.org/10.1177/1094428115579699
  7. Antonakis, John, Robert J. House, and Dean Keith Simonton (2017). Can super smart leaders suffer from too much of a good thing? The curvilinear effect of intelligence on perceived leadership behavior. Journal of Applied Psychology 102(7):1003-1021. DOI: 10.1037/apl0000221. URL: https://doi.org/10.1037/apl0000221
  8. Nickel, Lauren B., Brent W. Roberts, and Oleksandr S. Chernyshenko (2019). No evidence of a curvilinear relation between conscientiousness and relationship, work, and health outcomes. Journal of Personality and Social Psychology 116(2):296-312. DOI: 10.1037/pspp0000176. URL: https://doi.org/10.1037/pspp0000176
  9. Arin, Peren, Maria Minniti, Samuele Murtinu, and Nicola Spagnolo (2022). Inflection Points, Kinks, and Jumps: A Statistical Approach to Detecting Nonlinearities. Organizational Research Methods 25(4):786-814. DOI: 10.1177/10944281211058466. URL: https://doi.org/10.1177/10944281211058466
  10. Vleugels, Wouter, and Huw Flatau-Harrison (2023). Curvilinear relationships in person-environment fit research: Is there evidence for a too-much-of-a-good-thing effect? Journal of Occupational and Organizational Psychology 96(3):669-677. DOI: 10.1111/joop.12432. URL: https://doi.org/10.1111/joop.12432
  11. Cheng, Jinkai, Yating Luo, Feng Hu, and Kunjie Cui (2025). Exploring the Curvilinear Effect of Motivation to Lead on Leadership Emergence: The Moderating Role of Shared Team Vision. Behavioral Sciences 15(10):1295. DOI: 10.3390/bs15101295. URL: https://doi.org/10.3390/bs15101295
  12. Czerwiński, Stanisław K., Paweł A. Atroszko, and Roman Konarski (2025). Mental health of intellectually gifted individuals: investigating the nonlinearity of the relationship between intelligence and general mental health. Health Psychology Report 13(1):39-54. DOI: 10.5114/hpr/187337. URL: https://doi.org/10.5114/hpr/187337
  13. Yamaura, Kazuho, Shima Okada, Yoko Nishihara, Naruhiro Shiozawa, Eri Mukai, and Yusuke Sakaue (2026). Revisiting leader-member exchange: a review through the lens of the "too much of a good thing" effect. Frontiers in Psychology 17:1722767. DOI: 10.3389/fpsyg.2026.1722767. URL: https://doi.org/10.3389/fpsyg.2026.1722767
  14. Lamrani Abou Elassad, Yassine, Giuseppe Criaco, Justin J. P. Jansen, and Tom J. M. Mom (2026). Is the Venture Growth-Survival Relationship Inverted U-Shaped or Not? A Replication and Extension. Entrepreneurship Theory and Practice. DOI: 10.1177/10422587261441597. URL: https://doi.org/10.1177/10422587261441597