Commitments-in-Conflict
A research note — Step 5 of the commitment-typology study
The trade-off structure between identity-based, bonds-based, and need-based commitment
My task at Step 5 is to ask a question the Ren et al. material in E1 never poses directly: what happens when the forms of commitment co-occur, and what happens when they point in opposite directions? The typology is useless to a designer if it can only describe the pure cases. A community in the wild has members who joined for the topic, members who came for the people, and members who stay because leaving costs more than remaining — and these members are in the same room. So the trade-off structure is a real design question, not an academic addendum, and it has to be answered from what the source actually says, not from what a sensible theory of community "would" say.
I write this section knowing in advance how it must end: with a floor under it, and with a gap. What follows is three subsections with different weights of support — one strong, one half-strong, one empty at the source. That unevenness is not an accident of my reading. It is what E1 and E4 actually carry, and I would rather the note show the unevenness than level it with confident prose.
---
5.1 Where Identity Wins and Where Bonds Win
The two types have different failure modes, and the source names both. Identity-based commitment survives the departure of the people through whom it was first felt. E1 says so in exactly those words: in the passage distinguishing the two commitments, "when members feel strong identity-based commitment, they may stick with the group even if their friends leave." The commitment was never to the friends; it was to the group as a whole — the category, the purpose, the shared social location — and so it is not weakened when a particular friend leaves.
Bonds-based commitment has the opposite vulnerability profile. E1: "Friendship groups are vulnerable to member turnover because friends can leave as a clique." The mechanism and the dependency are stated together — "firmly dependent on close personal interaction" — and the same passage gives the underlying cause: idiosyncratic preferences. What holds a member is not a general attribute of the group that any member shares, but the specific liking she feels for specific people. That is why the loss is not one loss but a structural one: when the people she is close to leave together, the thing she was committed to is gone, and there is nothing underneath it to stand on.
So the trade-off's first axis is a robustness axis, and the two types sit on opposite ends of it: identity-based commitment is robust to dispersed turnover of individuals; bonds-based commitment is robust to nothing in particular except the continued presence of the specific people it was formed with. This is where the two types are least substitutable and most distinct — in what they survive.
The same source that gives the identity robustness result also records the empirical demonstration that the two are separable processes, not two names for one thing: Prentice, Miller, and Lightdale (1994) found that members of common identity groups "reported feeling more attached to their group as a whole than to their fellow group members," whereas members of common bond groups "reported feeling attached both to the group as a whole and to group members." That asymmetry is worth noticing in Step 5 because it is already a kind of trade-off finding: the two commitments do not simply share a mechanism at different strengths. Bond-based groups are attached to both the whole and the persons; identity-based groups to the whole alone. The identity type is narrower in one respect and broader in another — it loses the persons, but it does not depend on keeping any particular one.
---
5.2 Where They Substitute and Where They Conflict
The substitution clause has a clean source footing. E1, quoting the same Prentice, Miller, and Lightdale study, records the authors' own framing: "the two perspectives might . . . be viewed as describing two separable processes in the development and maintenance of groups, either of which might dominate under a given set of circumstances." That is the substitution claim in the source's own words: the two processes are separable — neither reduces to the other — and the conditions select which one dominates.
The strength of this clause and the weakness of the next one are both in the same sentence. "Either of which might dominate" is a claim about dominance, not about interaction. It says one process can carry the community where the other would not; it does not say what happens when both are strongly present at once, or when the design moves that strengthen one weaken the other. And here the source is silent in a way I want to mark precisely.
GAP — direct conflict not measured in E1. E1 does not report a case in which identity salience is raised inside a community and bonds are correspondingly undermined, or vice versa. The direction I am looking for — a measured trade-off where strengthening the identity type reduces the bonds type — is not in the material I hold from E1. The Prentice, Miller, and Lightdale finding is a between-type comparison, not a conflict measurement. The robustness finding in 5.1 is a difference in failure modes, not an inverse causal relation. So the substitution reading stands on a verbatim quote, and the conflict reading stands on nothing from E1 at all. If I were to go further — if I were to say that raising identity salience to the group as a whole tends to reduce attachment to individual members, or that a community rich in close personal bonds will tend to underdevelop its identity commitment — I would be supplying a claim the source does not make. I will not. The step's own instruction is to mark the conflict case GAP where E1 does not measure it, and that is exactly what it is: a named hole in the evidence, not a hedge.
There is a piece of adjacent material in E1 that sits close to the conflict case, and I want to name it so the note does not accidentally inherit it as if it were the conflict measurement. E1 records the finding, in the same Ren et al. body, that off-topic communication reduces identity-based commitment and increases bonds-based commitment. On first read that resembles a conflict result: one behavior pushing one commitment down while pushing the other up. But the relation it measures is between a design choice — permitting or encouraging off-topic communication — and the two commitment types. It is not a measurement of one commitment type acting directly against the other. It tells a designer that a single choice can move the two types in opposite directions, which is genuinely useful for Step 5; it does not establish the identity-salience-versus-bonds conflict I am after, and treating it as if it did would be exactly the kind of fill-from-memory substitution this section exists to avoid. So: GAP stands, and the off-topic finding sits beside it as adjacent, not as satisfying it.
---
5.3 The Need-Based Line
GAP — need-based commitment is not named in E1's Ren et al. material. The Ren et al. passage in E1 that carries this whole section works with two types and only two: identity-based commitment and bonds-based commitment. It "distinguish[es] between identity-based commitment and bonds-based commitment"; it traces the distinction to Festinger, Schachter, and Back's two ways that commitment forms — "through the attractiveness of the group or through the attractiveness of individual group members." Those are the two causes the source gives, and there is no third. Nowhere in the material I hold from E1 does the need-based form appear — not as a named type, not as a description that could be read into the text without the name, and not as a third attractor alongside group and members.
What the note carries on the need-based line is therefore not a claim from E1. It is a design-space inference, and I want to say exactly what that means and where it comes from, because the distinction between a source-backed type and an inferred type is the whole point of this step. The inference runs: the two types E1 names are specified by their attractor — what pulls the member toward the group — and E1 names two such attractors, the group as a whole and specific others. But a community can also retain a member by making exit expensive relative to staying — by an incentive structure inside the group and alternatives outside it — and that is a different kind of pull, not a weaker or stronger instance of either of the two named ones. That structure is the one my earlier attachment note used the name need-based commitment for. The whole reason for this GAP line is that it does not come from E1. It comes from my own design reasoning across communities, and I am carrying it forward from the earlier note because the design space seems to require a third attractor — not because E1 supplies one. I hold it as a working construct for the design space, not as evidence.
I want to be equally plain about the weaker version of this claim. One could try to read the need-based type into E1 at the margins: a member who stays because leaving would cost her a network she has built, or because the community is the only place with the content she needs, is arguably held by something E1's text gestures at when it says the forces that keep members in conventional organizations are weaker in online communities. But that is the designer's reading, not a source claim, and the Step 5 instruction is explicit that E1's Ren et al. material names only two types. So the note records the need-based line as GAP, carries the inference from the earlier attachment note marked as inference, and does not dress the inference in source clothes.
---
5.4 Closing Statement
The note's trade-off structure stands on exactly two source-backed types — identity-based commitment, robust to the departure of friends («building-successful-online-communities-evidence-based-social-design.pdf»), and bonds-based commitment, vulnerable to clique departure («building-successful-online-communities-evidence-based-social-design.pdf») — plus one honestly-marked GAP, the need-based commitment line, which E1's Ren et al. material does not name and which the note carries forward as a design-space inference from the earlier attachment note rather than a claim from the source.
---
Comments
No comments yet — be the first.