Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› jcp-theory-development
jcp-theory-development
GitHub专为JCP论文构建心理机制与假设，明确中介变量、推导理论预测的调节变量并排除竞争过程。旨在将效应解释转化为核心贡献，避免薄弱的理论描述，确保研究通过因果链逻辑验证心理过程。
触发场景
消费者效应缺乏解释性心理过程
审稿人指出理论薄弱或仅为寻找过程的效应
调节变量仅为稳健性检验而非机制测试
无法用一句话命名产生效应的中介变量
安装
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill jcp-theory-development -g -y
SKILL.md
Frontmatter
{
"name": "jcp-theory-development",
"description": "Use when building the psychological theory and hypotheses for a Journal of Consumer Psychology (JCP) manuscript — naming the mechanism (mediator), deriving theory-predicted moderators, and ruling out rival processes. Builds the mechanism that IS the contribution; it does not run the studies (jcp-methods)."
}
Theory Development (jcp-theory-development)
When to trigger
- You have a consumer effect but no explanatory psychological process
- Hypotheses read as "A increases B" with no account of why it happens in the mind
- A reviewer wrote "the theory is thin" or "this is an effect in search of a process"
- Your moderators feel like a robustness dump rather than tests of the mechanism
- You cannot name, in one clause, the mediator that produces the effect
At JCP the mechanism IS the contribution
JCP does not reward effects; it rewards explanations of effects. The theory section must specify a psychological process — a shift in inference, construal, attention, affect, motivation, goal activation, identity, or processing fluency — that generates the consumer outcome, and then derive hypotheses that test that process a priori. The standard frame is a causal chain: antecedent → psychological process (mediator) → consumer outcome, with moderators that the process predicts. A paper whose contribution is "X affects Y" will be desk-rejected or sent back; a paper whose contribution is "X affects Y because it changes mental state M, and only when condition C holds" is a JCP paper.
Build the process, not the prediction
- Name the mediator concretely. "Increased motivation" is too vague; "an inflated sense of personal control that lowers perceived risk" is a process. The mediator should be a known or definable psychological construct with established measurement.
- Derive moderation from the mechanism. The most persuasive JCP moderators are ones that switch the process on or off: if the effect runs through M, then blocking M (cognitive load, an alternative goal, a manipulation that satisfies the need M serves) should attenuate or reverse it. That is a theoretical claim, not a sensitivity check.
- State hypotheses a priori; avoid HARKing. Each study should test a distinct link in the chain (effect → mediation → moderation → boundary). Post-hoc explanations dressed as predictions are the rigor-era's cardinal sin and reviewers look for it.
- Pre-empt rival processes. A JCP reviewer will propose a competing mechanism (demand, mood, a confounded construal). Name the leading rivals and design the theory so a study can adjudicate — ideally a moderator that your process predicts but the rival does not.
Theory archetypes JCP rewards
| Archetype | What the contribution looks like |
|---|---|
| Novel mediator | A familiar effect is shown to run through an unexpected mental process |
| Process adjudication | Two rival mechanisms are pitted against each other; data favor one |
| Moderation-as-theory-test | A theory-predicted moderator turns the effect off, proving the process |
| Boundary / when-not | Specifying conditions under which a "known" effect reverses, via the mechanism |
| New construct | A new consumer-psychological construct, defined, measured, and shown to do work |
Checklist
- The mediator is named as a concrete psychological construct, not a vague "motivation/感觉"
- The causal chain antecedent → process → outcome is explicit
- At least one moderator is derived from the mechanism (switches the process on/off)
- Hypotheses are a priori and map onto a planned study chain
- The leading rival process(es) are named and there is a plan to rule them out
- The theory engages and changes an existing account, not just labels an effect
Anti-patterns
- Effect in search of a process: a robust result with no mechanism — the classic JCP reject
- Mediator-as-restatement: a "mediator" that is just the dependent variable measured earlier
- Moderator dump: boundary conditions with no theoretical rationale tied to the process
- HARKing: presenting post-hoc accounts as a priori hypotheses
- Borrowed-label theory: invoking a construct's name (e.g., "construal level") without using its logic or measuring it
- Single rival blindness: ruling out one alternative while ignoring the one the reviewer will actually raise
Output format
【Effect】antecedent → consumer outcome
【Process (mediator)】the named psychological mechanism = the contribution
【Moderators】theory-predicted, with which switch the process on/off
【Rival processes】named + how a study adjudicates
【Study chain】effect → mediation → moderation → boundary (study-by-study)
【What changes in the literature】advance / deepen / overturn which account
【Next skill】jcp-literature-positioning
版本历史
- 1839142 当前 2026-07-05 13:29
