Contribution
The paper joins two of the authors’ earlier findings into one causal sequence and tests it whole:
- Sheldon and Elliot (1998): self-concordant goals are better attained because they get sustained effort;
- Sheldon and Kasser (1998): attaining self-concordant goals raises well-being more than attaining other goals.
The new link is need satisfaction. Attaining self-concordant goals is proposed to produce daily experiences of autonomy, competence and relatedness, whose accumulation over a semester is what raises well-being. Study 2 tests that accumulation; Study 3 tests the whole model (Figure 1).
Key insight
Whether a goal is “personal” is not settled by the person having chosen and listed it. All the goals here were self-generated, “yet this does not mean that they all belong to the self in a deeper sense” (p. 494). What matters is the reason the goal is pursued: goals pursued from identification or interest keep their energy when obstacles come, and their attainment feeds the needs that well-being rests on. Goals pursued under external pressure or introjected guilt can be attained and leave the person no better off.
Assumptions
- The self is the agentic “I”: “the more-or-less integrated center of agentic activity” (p. 483), the subject-self of James and Mead, not the self-concept studied by most social psychologists.
- SDT’s continuum of perceived locus of causality (Ryan & Connell 1989) is one dimension, so a single difference score (identified + intrinsic − introjected − external) can stand for it (pp. 483, 486).
- Three universal needs: competence, autonomy and relatedness are “qualities of experience universally required by human beings in order to thrive” (p. 484). Well-being judgments are assumed to be formed partly from recalled experiences of these kinds (pp. 484–485, citing Kahneman 1997).
- Self-report throughout. Effort and attainment are the participants’ own monthly ratings (“How hard are you trying…?”, “How well are you doing?”), not observed outcomes.
- Correlational inference. The authors say causality “cannot be established with certainty in correlational data”, and present the prospective design as “the next best thing” to experiments (p. 493).
Key results
- Study 1 (N = 169 students, ten personal strivings, one semester).
- Self-concordance correlated with semester effort (r = .24) and attainment (r = .20); effort with attainment r = .76 (Table 1).
- With both in the regression, self-concordance dropped out (β = .02, ns) and effort stayed (β = .76); Sobel z = 2.93 (p. 486).
- Change in well-being: attainment β = .34; self-concordance no main effect; self-concordance × attainment β = .11, p < .05 (p. 487).
- SEM (need satisfaction unmeasured): the specified model fit poorly, χ²(7, N = 169) = 43.8, p < .01, GFI .93, NFI .87, CFI .88. After adding paths from initial well-being to effort and attainment, which the modification indices suggested, χ²(5) = 4.95, CFI 1.00 (p. 487).
- Study 2 (N = 152, three 24-hour activity samples over a semester).
- At Time 1 each of competence, autonomy and relatedness predicted well-being uniquely.
- Semester-long need satisfaction predicted change in well-being (β = .15, .21, .16 for competence, autonomy, relatedness), and the coefficients were “almost the same” with Time 1 need satisfaction controlled (pp. 488–489).
- SEM again fit only after adding a path from initial well-being to semester need satisfaction: χ²(2, N = 152) = 11.02, p < .01, then 2.83, p > .05 (p. 489, checked against the page image).
- Study 3 (N = 73, six personal projects, eight activity samples).
- Effort mediated again: effort β = .72, self-concordance β = .08, ns; Sobel z = 3.12 (p. 491).
- Change in well-being: attainment β = .28, self-concordance β = .19 (unpredicted main effect), interaction β = .23, p = .01.
- Need satisfaction was predicted by attainment (β = .46), self-concordance (β = .24) and their interaction (β = .24).
- Mediation of attainment’s effect on well-being by need satisfaction was partial: need satisfaction β = .22, attainment β = .19, p = .056; Sobel z = 1.98 (pp. 491–492).
- Whole-model SEM: paths .29 (self-concordance → effort), .71 (effort → attainment), .41 (attainment → need satisfaction), .21 (interaction → need satisfaction), .30 (need satisfaction → change in well-being); χ²(9, N = 73) = 20.3, GFI .93, NFI .89, CFI .93 (p. 492, checked against the page image). Adding a direct attainment → well-being path (β = .19, p < .07) improved fit somewhat (Δχ² = 3.7, p = .06).
- The self-concordance → effort path held with self-efficacy, implementation intentions, avoidance goals or life skills added, one at a time; each of these except avoidance had its own effect on effort.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Self-concordant goals receive more sustained effort, and effort fully mediates their better attainment | moderate: replicated in two samples (N = 169, 73) and in the authors’ 1998 work; all self-report, correlational | Studies 1, 3; pp. 486, 491 |
| C2 | Attainment raises well-being more when goals are self-concordant | moderate: interaction in two samples, small (β = .11, .23) | Studies 1, 3; pp. 487, 491 |
| C3 | Accumulated daily experiences of autonomy, competence and relatedness predict change in well-being, beyond their baseline | moderate: one sample, prospective | Study 2; pp. 488–489 |
| C4 | Attaining self-concordant goals yields more need-satisfying experience, which partly carries attainment’s effect on well-being | weak to moderate: one sample of 73, partial mediation, attainment keeps a near-significant direct effect | Study 3; pp. 491–492 |
| C5 | The self-concordance effects are not reducible to self-efficacy, implementation intentions, avoidance framing or life skills | moderate for the four named controls, entered one at a time | Study 3; p. 492 |
| C6 | Goals people freely choose and list may still not “belong to the self in a deeper sense” | interpretive, resting on C1–C2 | p. 494 |
| C7 | The distinction between one’s own goals and “alien injunctions” is “subtle and shifting”, because internalizing culture works by the same infiltration | interpretive; stated in the conclusion, not tested | p. 495 |
Concepts
- self-concordance: “the degree to which stated goals express enduring interests and values” (p. 482); operationally, identified + intrinsic − introjected − external reasons. In earlier work called goal self-determination (n. 7).
- conation: “the human capacity to set and pursue personal agendas” (p. 482), separated from cognitive and affective processes.
- need satisfaction: daily, activity-based experiences of competence, autonomy and relatedness; the autonomy score uses the same four-reason method as self-concordance, applied to the day’s activities (n. 5).
- self-integrated vs. nonintegrated action (Figure 2): intrinsic and identified reasons on one side, introjected and external on the other.
Connections
- Ryan & Deci 2000 (LIT-559) and Deci & Ryan 2000 (LIT-558). The paper is the SDT goal literature these reviews later draw on, and it states the three needs in their terms. Its own SDT sources are Deci and Ryan 1985 and 1991, which the record does not hold.
- THEORY-054 (autonomy as graded integration). Evidence for the autonomous end of the continuum, not for integration as distinct from identification: there is no integrated-regulation item, and identified and intrinsic are pooled. THEORY-054 notes SDT studies “usually” merge these; this is one. The measure is the person’s own report of reasons, which is the endorsement-based measurement THEORY-054’s refutation clause worries about.
- THEORY-047 (history of acquisition fixes autonomy). Not tested: no history is measured. C7 states the difficulty THEORY-047 faces from inside SDT: the process that makes an injunction alien is also the one by which culture is internalized.
- THEORY-052 (need frustration as transdiagnostic). C3 is the well-being side of the same need theory, in a nonclinical sample; it says nothing about frustration or disorder.
- THEORY-029 (no attitude owns a motive by its order). The paper’s criterion is felt ownership, “whether the person feels ownership as he or she pursues the goal” (p. 484), not order of endorsement. It takes the person’s report as the measure and does not ask whether that report could itself be alienated, which is THEORY-029’s question.
- Authenticity (LIT-574, LIT-594). The paper speaks of goals that “represent … that person’s authentic interests and values” (p. 483) and of the “true self” (pp. 485, 493), and draws on Rogers’s congruence (pp. 485, 495). Its Kaczynski passage (p. 495) concedes that self-concordance is silent on the content of goals; Taylor’s book is the philosophical statement of why that matters.
- Schlegel et al. (LIT-606). Sheldon is thanked in that paper’s acknowledgements; it measures the true self-concept directly rather than through goal reasons.
Bearing on the record
- THEORY filed as THEORY-077. “Goals pursued for identified or intrinsic reasons receive more sustained effort, which carries their better attainment” is well supported by this paper and by the 1998 work it recaps, and is filed Proposed with this paper as source. It should promote on a replication with an effort or attainment measure that is not the participant’s own rating.
- For THEORY-054, the paper is supporting but not decisive evidence; it should be cited as showing that the autonomous end predicts, not that integration does.
- No ML instruction. Nothing here belongs in the anthology.
Limitations
- All self-report, including effort and attainment; shared method variance could carry the effort → attainment path (r = .76).
- Small, homogeneous samples: Rochester students, one semester each; Study 3 has 73 participants for a model with many paths.
- Post hoc model revision. Studies 1 and 2 reached acceptable fit only after adding paths from initial well-being that the modification indices suggested, and Study 3’s “adequate” fit has a χ² of 20.3 on 9 df, which is significant at .05. The authors say the added paths are “not inconsistent with our conceptual model”.
- The model needs amending by its own data: attainment’s direct effect on well-being is not carried by need satisfaction (Study 3).
- Integration is not measured apart from identification (see Connections).
- Content is out of scope: goals with “manifestly evil” content are acknowledged and set aside (pp. 494–495).
Open questions
- Would an observed measure of attainment (grades, gym visits) keep the effort mediation?
- Does integrated regulation, measured separately from identified, add anything to the prediction?
- Does the attainment × self-concordance interaction survive in a sample large enough to estimate it precisely?
Corrections
- none to a seeded skim (there was no seed)
- Citation. One search summary gave the pages as 546–557; the printed article is pp. 482–497. The brief’s citation (JPSP 76(3), 1999) is right.
- Study 3 fit. The paper calls χ²(9, N = 73) = 20.3 “p > .01” and the fit “adequate”; 20.3 on 9 df has p ≈ .016, so the model is rejected at .05 and passes only at .01. The printed statement is true but reads as more reassuring than the test is.
- Study 2 degrees of freedom. Both the original and the revised Study 2 models are printed as χ²(2, N = 152) (p. 489), but the revision adds a free path, which should cost a degree of freedom. One of the two df values is a misprint, or the revision changed something else that the text does not report.