Contribution
A field test, in people’s own goals over three weeks, of the control-process prediction that running ahead of an expected rate of progress is followed by less effort on that goal (“coasting”) and running behind by more (“pushing”). Both held. Before it, the authors say, “not much evidence bears on” coasting (p. 1022), and the clearest prior work was Louro, Pieters and Zeelenberg (2007). It adds that coasting is weaker in bipolar I disorder, a candidate route into mania.
Key insight
Effort tracks the sign of the gap between progress and expectation, in both directions: ahead, ease off; behind, push. On Carver’s account the signal that carries this is affect, positive for ahead and negative for behind, but this study reaches the effort change from the progress ratings directly. Affect is the theory’s middle term, and it was not measured.
Assumptions
- The two-loop model (Carver & Scheier 1998), stated on p. 1021: a loop regulates the rate of goal approach; “If goal progress falls below an implicit criterion (e.g., expected progress or intended progress), negative affect arises and effort toward that goal increases; if goal progress exceeds the implicit criterion, positive affect arises and effort toward that goal decreases.”
- Carver 2003’s extension to many goals (p. 1021): negative affect “is a call for higher priority for that goal”; “Carver reasoned that positive affect—a signal of surging ahead—may mean that the goal to which it relates can assume a lower priority; less effort would be expended toward this goal, which would free attention and resources for other needs. If another goal needs attention at that point, effort might shift to it; this again represents a reordering of priorities.” The result is satisficing across goals (Simon 1953, 1967), with no claim that the shifting “is intentional, or cannot be overcome, or even that it is advisable” (pp. 1021–1022).
- Expectation is the criterion. The deviation is perceived progress at t + 1 minus the progress expected at t (Fig. 1, p. 1024).
- Approach goals only (n. 1); goals were chosen to be tangible, effortful and approach-related (p. 1023).
Key results
- Design (pp. 1022–1023). 12 people with bipolar I disorder, no current episode (MHRSD < 10, BRMS < 7), and 12 with no mood-disorder history. Each named three goals and, three times a day for 21 days, rated for each the effort put in since the last assessment (0–4), how much closer they had come (0–4) and how much closer they expected to get by the next (0–4). 4% of 13,608 observations were missing (p. 1024).
- Analysis (pp. 1024–1025). Multilevel models of the change in effort from t + 1 to t + 2 on separate overshoot and undershoot vectors, with group as a level-2 predictor. Observations were not nested within goals.
- Pushing (p. 1025). Undershoot predicted more subsequent effort, b = −0.25, t(23) = −4.06, p < .01 (negative because undershoot is coded negative). Group did not moderate it, b = −0.19, p = .13; separately it was significant for BD (b = −0.34, p < .01) and marginal for controls (b = −0.16, p = .07).
- Coasting (p. 1025). Overshoot predicted less subsequent effort, b = −0.34, t(23) = −6.38, p < .001: “making unexpectedly high progress toward a goal was followed by reduced effort toward that goal.”
- Coasting attenuated in BD (p. 1025). Group moderated the overshoot slope, b = 0.32, t(22) = 4.02, p < .01. Both groups coasted, BD b = −0.19, p < .01, controls b = −0.50, p < .001.
- Robustness (p. 1025). Time of day did not change the results. People with BD reported more effort overall (M = 1.81 against 1.18) and higher expectations (1.65 against 1.14), but did not overshoot less, so neither low effort nor fewer overshoots explains the attenuation.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Progress beyond expectation on a goal is followed by less effort on it | moderate | b = −0.34, p < .001, n = 24, daily self-report |
| C2 | Progress short of expectation is followed by more effort on it | moderate | b = −0.25, p < .01 |
| C3 | Coasting is weaker in bipolar I disorder in remission | moderate | group × overshoot b = 0.32, p < .01; small groups |
| C4 | Positive affect is the signal that lets the goal take lower priority and frees resources for other goals | not tested here | stated from Carver 2003 (p. 1021); no affect or other-goal measure |
| C5 | Attenuated coasting may let an initial success spiral into mania | speculation | p. 1026, “requires further testing” |
Concepts
- Coasting — reducing effort on a goal after progress exceeds the criterion (expected progress); “a tendency to coast, or reduce effort, toward the associated goal” (p. 1023).
- Pushing — increasing effort after progress falls short of it.
- Overshoot / undershoot — positive and negative deviation of perceived progress at t + 1 from expected progress at t (p. 1023).
Connections
It is the field half of the Carver programme’s evidence for coasting; the experimental half is Thürmer, Scheier and Carver (2019/2020, filed beside it as LIT-1878), which calls this study the “only one published study” it knows to test coasting and shifting directly, and faults it for leaving “common method variance” open. The control-process model is the record’s through Carver 2018 (LIT-1449, NOTE-1102). Louro et al. (2007) is cited as the convergent prior result.
Bearing on the record
- CLAIM-468: partly. The objection’s positive half waits on Carver 2003 (LIT-1445) for the thesis that positive affect lowers a goal’s priority. This paper, by Carver’s group, states that thesis in so many words (p. 1021, quoted under Assumptions), so the attribution is now confirmed from a read source, not reported. But it is a restatement, and the study tests only a consequence: more-than-expected progress is followed by less effort on that goal. It measured no affect and no reallocation to other goals, so it does not show that positive affect is what lowers the priority.
- CLAIM-440: partly, and indirectly. Clause 3 is held for negative valence only “until that paper is read”. This reading supplies the attribution and field evidence that the hold of a goal on effort weakens after an overshoot, which the control-process account links to positive affect. Whether a lowered priority for one goal is a weaker hold of the prevailing ordering, in clause 3’s sense, or a decision of that ordering to move to another goal, is a question about the claim this paper does not answer. The negative-valence half (pushing) fits the clause.
- No instruction for machine-learning practice; nothing for the anthology.
Limitations
- Self-report throughout (p. 1025), with no time stamps to check when ratings were made; effort and progress were rated by the same person in the same booklet, which Thürmer et al. call common method variance.
- Small samples, 12 per group, though with many observations each.
- Goals not modelled. Observations were treated as independent of which goal they came from (p. 1023).
- No affect, no other-goal measure. The mechanism the introduction gives (affect) and the payoff it names (effort freed for other goals) are both unobserved.
- The BD group was in remission, so the mania inference is a conjecture (p. 1026).
Open questions
- Is affect the mediator, and does effort released from one goal go to another? A within-person study that rated affect and effort on all three goals would answer both.