THEORY-1307: In Slemp et al.'s meta-analysis of 881 mostly cross-sectional samples, perceived support for autonomy, competence and relatedness each correlated strongly with satisfaction of all three needs (ρ = .44–.62) and negatively with their frustration, but the three supports intercorrelated .67–.71, and their need-specificity appears only in joint models: with all three entered, relatedness support carried relatedness satisfaction (β = .56) and autonomy support added nothing to it (β = .007)

Source

Slemp, Field, Ryan, Forner, Van den Broeck and Lewis (2024), LIT-1969, pp. 8–17 of the Online First PDF; read in NOTE-1507.

What was actually shown

Pooled correlations (corrected for unreliability, pp. 11–12): autonomy support with autonomy, competence and relatedness satisfaction .57, .45, .44 (k = 235–274); competence support .50, .48, .48 (k = 38–47); relatedness support .53, .45, .62 (k = 40–47). With frustration, −.20 to −.55 (k = 8–31).

The supports overlap. Their pooled intercorrelations are .67 (autonomy–competence), .68 (autonomy–relatedness) and .71 (competence–relatedness) (Table 1).

Joint models (Table 2, harmonic-mean N 32,933–113,363). All three supports explain R² = .369, .269 and .388 of autonomy, competence and relatedness satisfaction, against .325, .202 and .194 for autonomy support alone. For relatedness satisfaction the joint betas are autonomy .007, competence .078, relatedness .560.

Checked. Solving the normal equations on Table 1’s correlations reproduces every β and R² of Table 2’s Models 1–7 to three decimals, and Table 4’s relatedness block (vertical .108, lateral .713, R² = .616) from Table 3c. The reproduction checks the arithmetic, not the pooled correlations.

The specificity could have failed to appear: had each support related to each need through a common factor, the joint betas would have been roughly equal. For relatedness they are not; for competence satisfaction they nearly are (.179, .256, .147).

What this does not say

• That support causes need satisfaction. Over 70% of the studies are cross-sectional, support and outcome are mostly reported by the same person, and the authors could not rule out reverse causation (p. 20).
• That autonomy support is unimportant for relatedness. Alone it correlates .44 with relatedness satisfaction; it adds nothing only alongside relatedness support, with which it correlates .68.
• That the needs are distinct. The satisfactions themselves correlate .61–.72 (Table 1).
• That the matrix cells are equally well estimated. Cells involving competence or relatedness support rest on k = 38–135; lateral cells on k = 2–4.