Contribution
The largest synthesis of SDT’s contextual claim: that others’ support for autonomy, competence and relatedness goes with need satisfaction, autonomous motivation, well-being and performance. It is the first to treat competence and relatedness support alongside autonomy support across domains, to estimate their incremental contribution from meta-analytic matrices, and to compare lateral (peer) with vertical (authority) sources (pp. 2–7).
Key insight
The three supports are not three separate levers in the data. They correlate .67–.71 with each other, and each correlates about as strongly with every need as with its own. The specificity SDT’s model draws (Fig. 2: each support “primarily” feeds its own need) shows only in the joint models, and most clearly for relatedness: with all three in, relatedness support’s weight on relatedness satisfaction is .56 and autonomy support’s is .007 (Table 2).
Assumptions
- Correlations, read as direct effects. Only zero-order correlations were pooled; support is “treated as the predictors” on theoretical grounds, and where lagged correlations existed the most distant was used “consistent with the causal direction implied by SDT” (p. 10). Mediation through need satisfaction is not tested (p. 5).
- Schmidt–Hunter psychometric meta-analysis, correcting for unreliability by artifact distributions; Gignac and Szodorai benchmarks .15/.25/.35 for weak/moderate/strong (pp. 9–10).
- Culture as Hofstede’s individualism percentile of the sample’s country (p. 14), which the authors call an assumption of homogeneity (p. 20).
- Structure and involvement counted as competence and relatedness support (p. 8).
- Incremental validity from pooled matrices, with harmonic-mean N (32,933–113,363) as the sample size (p. 11).
Key results
The database (pp. 8–9)
- Seven databases to 27 November 2021, plus citation searches; 768 records, 881 samples; 50 matrices added by emailing 173 authors. Screening by the first author, 20% double-screened (98% agreement, κ = .93). Grey literature not actively sought. Not preregistered; data and scripts at OSF (osf.io/2ht3s).
Main associations (pp. 11–12; ρ, corrected)
- Need satisfaction: autonomy support with autonomy .57 (k = 235), competence .45 (k = 274), relatedness .44 (k = 251); competence support .50, .48, .48; relatedness support .53, .45, .62.
- Need frustration (k = 8 for competence and relatedness support): autonomy support −.30, −.20, −.35; competence support −.33, −.38, −.33; relatedness support −.30, −.30, −.55.
- Motivation: autonomous motivation .40–.44; controlled motivation .01–.10; amotivation −.17 to −.31.
- Subjective well-being: positive affect .39–.50; negative affect −.24 to −.44; life satisfaction .42–.50 (competence support k = 3).
- Performance: overall .19 (autonomy), .14 (competence), .25 (relatedness); self-reported performance stronger than objective.
- Heterogeneity: “very high values for I² (typically exceeding 90%)” (p. 12).
Moderators (pp. 13–14)
- Measures: domain-specific scales (TASC, POPS) gave weaker effects, e.g. autonomy support with intrinsic motivation in classrooms .17 (TASC) against .57 (climate questionnaires).
- Domain: physical education strongest; autonomy support with autonomy satisfaction in health care weakest (.23, k = 11).
- Source: vertical autonomy support stronger than lateral for competence satisfaction and autonomous motivation; lateral relatedness support stronger than vertical for relatedness satisfaction (difference −.20).
- Culture: 2 of 18 relations moderated by individualism, autonomy support → autonomous motivation weaker in more individualist samples (β = −.0010 per percentile) and relatedness support → intrinsic motivation stronger (β = .0030). Age and time lag weakened some relations.
Incremental validity (pp. 14–17, Tables 1–4)
- Table 1: supports intercorrelate .67, .68, .71; satisfactions .61–.72.
- Table 2, all three supports (Model 7): R² = .369 (autonomy satisfaction), .269 (competence), .388 (relatedness); ΔR² over autonomy support alone .044, .067, .194. Joint betas for relatedness satisfaction: autonomy support .007, competence support .078, relatedness support .560. Relative weights: each support at least 18% of R².
- Reproduced here from the Table 1 matrix: Models 1–7 for all three needs, to three decimals.
- Table 4: lateral and vertical supports add to each other; lateral relatedness support with relatedness satisfaction ρ = .78 (k = 4) leaves vertical relatedness support ΔR² = .008. Several cells rest on k = 2–4 (Table 3).
Robustness (pp. 14–15)
- Outliers rarely mattered; publication-bias analyses flagged 41% of distributions as possibly misestimated by more than 20%, but by the average-range estimate 90% were within 20%.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Perceived need support correlates strongly with need satisfaction, autonomous motivation and subjective well-being, across domains | strong as an association | k up to 357, pp. 11–12 |
| C2 | It correlates negatively, somewhat more weakly, with need frustration | moderate | k = 8–31 |
| C3 | Support relates weakly to performance | moderate | ρ = .14–.25; self-report inflates |
| C4 | The relations are largely stable across individualist and collectivist countries | moderate | 2 of 18 moderated; country-level coding |
| C5 | Competence and relatedness support add to autonomy support in explaining need satisfaction | moderate | ΔR² from pooled matrices; supports correlate ~.7 |
| C6 | Lateral support adds to vertical | weak–moderate | several cells k = 2–4 |
| C7 | Supports cause these outcomes | weak | assumed; >70% cross-sectional; the authors’ own limitation, p. 20 |
Concepts
- Interpersonal supports for basic needs — “an umbrella term” for behaviours that nurture the three needs (Fig. 2): autonomy support (choice, rationale, acknowledging feelings, avoiding controls), competence support including structure (clear expectations, feedback, expressed confidence), relatedness support including involvement (interest, care, unconditional regard).
- Vertical / lateral — support from an authority (teacher, manager, parent) or from a peer (colleague, sibling, partner) (p. 6).
Connections
- Updates Slemp et al. 2018 (leader autonomy support), Ng et al. 2012 (health care) and Vasconcellos et al. 2020 (PE) (pp. 2–3).
- Cites Ntoumanis et al. 2021 (LIT-1962) among “experimental work demonstrating that need-supportive interpersonal behaviors yield positive motivation and well-being-based outcomes” (p. 20).
- Chirkov et al. 2003 (LIT-1989) and Iyengar and Lepper 1999 (LIT-1949) frame the culture question (p. 5).
Bearing on the record
- “Effects” in the abstract and title wording are correlations. The registration calls it evidence “for SDT’s contextual claims”; it is evidence for the association. The authors say their “interpretation of need-supportive behaviors as the antecedent was based purely on theoretical grounds” and that reverse causation could not be ruled out (p. 20). Both sides are mostly the same person’s report.
- QUESTION-369. The three-way intercorrelation of the supports (~.7) and of the satisfactions (.61–.72) is a discriminant-validity question the record’s question inherits: if the needs are distinct, their measures and their antecedents are hard to tell apart in the data. The specificity in the joint models (relatedness support → relatedness satisfaction) is the best evidence here that they are not one factor.
- THEORY-052. Support correlates with frustration at −.20 to −.55, on k = 8 for two of the three supports. Nothing about psychopathology, direction, or onset; the promote_when is not touched.
- THEORY-047 / internalization. Support’s near-zero relation to controlled motivation (.01–.10) and strong one to autonomous motivation (.40–.44) is consistent with SDT’s claim that support fosters internalization rather than less regulation.
- Filed from it: THEORY-1307 (Active).
- No instruction for machine-learning practice. Its practical recommendations concern teaching, management and health care.
Limitations
- Mostly cross-sectional, recipient-reported support and outcomes; time lag weakened some relations, “indicating a possible upward bias in cross-sectional effects” (p. 20).
- Heterogeneity is very high and largely unexplained; 60 distinct autonomy support measures (p. 19).
- Culture coded at country level from Hofstede.
- Some incremental and lateral analyses rest on very few studies (k = 2–4 for some cells), which the authors flag as “more affected by second-order sampling error” (p. 20).
- Not preregistered; grey literature not sought; single first-pass screener.
Open questions
- Do the supports predict need satisfaction prospectively, or does observed satisfaction elicit support (p. 20)?
- Is the ~.7 intercorrelation of the supports in the behaviour or in the measures (p. 20)?
- Do the cross-domain correlations hold up in a meta-analytic structural model with need satisfaction as mediator (p. 20)?