Contribution
It turns Deci’s (1980) three types of person into three measured dispositions that everyone has in some degree, and gives the first instrument for them, the General Causality Orientations Scale (GCOS). It renames the internal and external orientations “autonomy” and “control” so that introjects (“I have to do that to feel worthwhile”) count as control rather than internal causation (p. 111). It places causality orientations in a network of other measures and claims a parallel between the three orientations and the three kinds of event of cognitive evaluation theory.
Key insight
A person’s orientation is to the events themselves: the autonomy-oriented “interpret their existing situations as more autonomy promoting (i.e., as informational)”, the control-oriented “seek out, select, or interpret events as controlling”, the impersonal experience behaviour “as being beyond their intentional control” (p. 112). So the same reward or feedback the experimental literature manipulated is, on this account, read differently by different people, and “a high level of autonomy orientation would lead to the same types of behaviors, cognitions, and affects that tend to occur in informational environments” (p. 113).
Assumptions
- Three orientations, scored separately. “The scale was constructed to measure the level of each orientation within a person rather than to classify a person as being one of three types” (p. 113).
- A rank order of self-determination: autonomy > control > impersonal, so a self-determination measure “would be positively correlated with the autonomy orientation, slightly negatively correlated with the control orientation, and highly negatively correlated with the impersonal” (p. 115).
- Locus of causality is not locus of control. Rotter’s internal control can be autonomous or controlled; his external control should relate to the impersonal orientation (pp. 113–114).
- Generality. The scale crosses domains (achievement, relationships) on purpose; domain-specific scales are expected to predict better (pp. 113, 131).
Key results
Construction (pp. 117–118)
- 96 items written “by persons familiar with the theory”; over 200 students; 12 vignettes kept, “5 of the 36 items … were rewritten in an attempt to improve their loadings”. The revised scale went to 923 undergraduates and 193 non-students, in varied samples and procedures.
- Example vignette (p. 118): offered a new position — (A) “I wonder if the new work will be interesting?” (C) “Will I make more at this position?” (I) “What if I can’t live up to the new responsibility?”
Reliability (pp. 118–119, Table 1)
- α (n = 636): autonomy .744, control .694, impersonal .741.
- Mean item–own-subscale (minus itself) r: A .522, C .483, I .508; the highest mean with another subscale .147 (I items with C). Lowest single own-subscale r .355; highest single cross r .301: “no overlap of the two distributions”.
- Retest over about 2 months (n = 51): A .749, C .711, I .778.
Subscale relations (pp. 119–120)
- A–C r = .034; A–I −.248 (p < .001); C–I .273 (p < .001), n = 636.
Other constructs (Table 2, p. 120; pp. 120–123)
| measure (n) | A | C | I |
|---|---|---|---|
| Social desirability (51) | −.02 | −.03 | −.18 |
| Supporting autonomy in children (51) | .55*** | −.04 | .09 |
| Type-A (73) | .16 | .26* | .00 |
| Rotter external control (73) | −.16 | .29** | .52*** |
| Self-derogation (73) | −.20* | −.06 | .38*** |
| Depression, BDI (87) | −.12 | .09 | .28** |
| Private self-consciousness (70) | .21* | .23* | .24* |
| Public self-consciousness (70) | .11 | .22* | .41*** |
| Social anxiety (70) | −.16 | .06 | .58*** |
| Ego development (42) | .43** | −.22 | −.32* |
| Self-esteem (70) | .35*** | .01 | −.61*** |
Emotions (pp. 123–126, Table 3)
- Ryan et al. 1984 (n = 86): A with interest .24, with shame −.34; C with distress .22, guilt .24.
- Ryan 1982’s experiment (n = 128; only significant correlations shown): I with shame .29, fear .38, guilt .21, relaxed −.36, pressured .30; C with hostility A .21, “important to do well” .27, “how hard did you try” −.19; A with hostility B −.16, guilt −.26. Correlating across eight experimental cells is “a tenuous procedure” (footnote 1, p. 124).
- Kernis (1983, dissertation): control-oriented people were happier and less fearful in controlling than in informational conditions (p. 126).
Behaviour (pp. 126–129, Tables 4, 5)
- Exam grades (n = 88): C r = −.25 (p < .02); A −.04; I .01. The predicted negative I relation did not appear.
- Fairness ratings after grades were partialled: A with essay fairness β = .51, and I with short-answer fairness −.29, both at p < .07; “the findings are quite weak”, “since all six B values were in the predicted direction, is suggestive” (p. 127).
- Mother–infant attachment (Bridges et al. 1983): n = 4 avoidant, 31 secure, 6 resistant; the predicted pattern (I highest for avoidant, A for secure, C for resistant), “overall significance … marginal (p = .06)” (p. 128).
- Cardiac surgery (King 1984, n = 50): A predicted seeing the operation as a challenge.
Norms (pp. 129–130, Tables 6, 7)
- Students (n = 636): A 70.54 (6.62), C 49.05 (8.13), I 39.03 (8.99), on a 12–84 range. Women higher on A (71.89 vs 68.83), men on C (50.51 vs 47.83), both p < .001.
- 83 engineers and 19 research scientists: scientists lower on C (40.8 vs 45.6, p < .02).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | The GCOS’s three subscales are internally consistent and stable over two months | moderate | α .69–.74 (n = 636); retest .71–.78 (n = 51), pp. 118–119 |
| C2 | Autonomy and control orientations are independent dimensions, not opposite poles | strong (for this scale) | A–C r = .034, n = 636, p. 119 |
| C3 | The impersonal orientation tracks external locus of control, self-derogation, social anxiety, depression and low self-esteem | moderate | Table 2, n = 70–87 |
| C4 | The autonomy orientation tracks supporting children’s autonomy, ego development and self-esteem | moderate | Table 2, n = 42–70 |
| C5 | The orientations predict behaviour | weak | one exam (C, −.25), fairness trends at p < .07, attachment p = .06 with cells of 4 and 6, one dissertation |
| C6 | Orientations parallel the three kinds of event: autonomy behaves as informational contexts do, control as controlling ones, impersonal as amotivating ones | weak | asserted (p. 113); supported only by the pattern of correlations; Kernis’s person × context interaction is cited, not shown |
Concepts
- Causality orientation — a relatively enduring aspect of people that characterizes “the source of initiation and regulation, and thus the degree of self-determination, of their behavior” (abstract, p. 109).
- Autonomy orientation — choice based on information; seeks self-determination; experiences rewards “more as affirmations of their competence” (p. 112).
- Control orientation — behaviour organized by controls “either in the environment or inside themselves”; covers compliance and defiance alike (p. 112).
- Impersonal orientation — behaviour felt to be “beyond their intentional control”; “pervasive sense of incompetence” (p. 112).
- Locus of causality vs locus of control — why one behaves vs what controls one’s outcomes (p. 113).
Connections
It is the trait counterpart of the experimental work in Ryan 1982 and Ryan, Mims and Koestner 1983 (LIT-1970), which it cites as manipulating “the interpersonal context of rewards or feedback” (p. 110). The record held causality orientations only as one-line mentions in NOTE-439 (p. 240 of LIT-558) and NOTE-441 (p. 1562 of LIT-560). The paper anticipates Deci and Ryan’s 1985 book and a causality-orientations-at-work scale (p. 131).
Bearing on the record
- The registration’s “the source of Causality Orientations Theory” is right, with the qualification that the paper presents a scale and its validity, not a theory with propositions: COT’s propositions are stated later (Ryan et al. 2022, LIT-1967, cites this paper as COT’s source and reports one or two meta-analyses).
- THEORY-054 (graded autonomy, not order). The paper’s renaming is evidence for the account’s “Not order” half in the SDT sources: inner events such as “I have to do that to feel worthwhile” “function more like external controls” (p. 111), so a motive inside the person can be controlled. It is the same position as LIT-558’s introjects, ten years earlier, and is not a test of it.
- THEORY-049 (functional significance). The paper adds a person-side route to the same account: whether an event is informational or controlling depends on the recipient’s orientation as well as the event. That makes the account’s refutation harder to set up, since an event can be controlling for one person and informational for another; a test of THEORY-049’s clause needs the orientation measured or randomized away.
- QUESTION-071 and CLAIM-016 (history and manipulation). The paper holds that orientations develop from the person’s history of contexts, but measures only the present disposition, so it does not bear on history against present structure.
- No instruction for machine-learning practice.
Limitations
- Its validity summary overstates its table. “All of the predicted relationships appeared” (p. 123), but the predicted negative relation of control to ego development was “not quite significant” (−.22, n = 42), the predicted I–performance relation did not appear (p. 127), and the fairness and attachment results are at p < .07 and p = .06.
- Small validation samples (42 to 128) for most constructs, and many correlations, with Table 3 reporting only the significant ones.
- Items written by the theory’s own group and rewritten to improve loadings, then validated on overlapping samples; no independent sample confirmation is reported.
- The scale is not printed, so its content cannot be checked from the paper.
- Unpublished manuscripts and dissertations carry several validity results (Bridges et al.; King; Kernis; Weil and Ryan; Ryan et al. 1984).
Open questions
- Do the orientations moderate the effects of manipulated events, as the Kernis interaction suggests, in a preregistered experiment?
- Does the autonomy orientation predict behaviour beyond self-report outcomes? Ryan et al. 2022 report correlations with outcomes “small-to-medium” (Hagger and Hamilton 2021).