“效应量”与“统计显著”有何区别——大样本为何危险？
How do effect size and statistical significance differ—why are big samples risky?
统计显著只看是否不像巧合，大样本下极小的效应也能“显著”却无实际意义；效应量（如 Cohen's d）衡量大小。读研究要看两者，而非只盯 p 值。
Significance only asks 'unlikely by chance'; with huge samples, trivial effects turn 'significant' yet meaningless. Effect size (e.g., Cohen's d) measures magnitude. Read both, not just p.
难度 专业 · 适合 小学高段(9-12)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#效应量 #显著 #大样本
什么是“元分析”？它怎么综合许多小研究？
What is a meta-analysis—how does it combine many small studies?
元分析用统计方法把多个独立研究合并，估算整体效应并看异质性（为何结果不同）。它比单一研究稳健，但前提是纳入研究质量与发表偏倚可控。
Meta-analysis statistically combines independent studies to estimate an overall effect and examine heterogeneity (why results differ). It is sturdier than one study, yet depends on study quality and controlled publication bias.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#元分析 #效应量 #异质性
“p值”的常见误解——p<0.05 到底意味着什么？
Common p-value myths—what does p<0.05 actually mean?
p值是在“零假设为真”前提下，看到当前数据或更极端的概率，并非“假设为真的概率”，也不代表效应大小。p<0.05只说不太像巧合，不等于重要或可靠。
The p-value is the probability of the data (or more extreme) if the null is true—not the probability the hypothesis is true, nor the effect size. p<0.05 merely says not much like chance, not important or reliable.
难度 专业 · 适合 小学高段(9-12)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#p值 #零假设 #误解
心理学的重复危机是怎样被量化的？学界采取了哪些制度性改革？
How was psychology's replication crisis quantified, and what institutional reforms followed?
开放科学合作组织在 2015 年组织多个实验室重复了一百项心理学研究，结果只有约三分之一到四成达到统计显著，且平均效应量约为原始报告的一半。诊断出的原因包括对显著性阈值的选择性报告、事后调整假设、样本量不足导致的低统计效力，以及缺少直接重复的激励。随后的改革包括预注册与注册报告、公开数据与分析代码、多实验室协作的大规模重复项目，以及期刊对效应量与置信区间的报告要求，美国心理学会等机构也更新了报告标准。
The Open Science Collaboration had many laboratories repeat one hundred psychology studies in 2015, and only about a third to two-fifths reached statistical significance while average effect sizes were roughly half those originally reported. Diagnosed causes include selective reporting around significance thresholds, hypothesising after results are known, low statistical power from small samples, and the absence of incentives for direct replication. Reforms followed: preregistration and registered reports, open data and analysis code, large multi-lab replication consortia, and journal requirements to report effect sizes and confidence intervals, with bodies such as the American Psychological Association updating reporting standards.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-11-02
来源 · Source
Open Science Collaboration（2015）《Science》重复百项心理学研究；Simmons、Nelson 与 Simonsohn（2011）《Psychological Science》假阳性心理学
Open Science Collaboration (2015), Science, Estimating the reproducibility of psychological science; Simmons, Nelson and Simonsohn (2011), Psychological Science, False-positive psychology
独立核验依据 · Independent verification: 核验依据 1 CROSSREF 标识 10.1126/science.aac4716：Estimating the repro
#重复危机 #预注册 #统计效力
自我损耗理论为什么在近十年受到严重质疑？
Why has ego depletion theory come under serious challenge over the past decade?
该理论主张自控像肌肉一样消耗有限资源，先完成一项抑制任务会削弱随后的自控表现，早期实验大量支持这一预测。质疑来自三方面：多实验室大规模预注册重复研究未能稳定复现效应，效应量接近于零；元分析在校正发表偏差与小样本偏倚后，原有证据大幅缩水；血糖作为损耗资源的解释也与生理测量不符。当前较受支持的替代解释强调动机、注意分配与对努力的主观意愿变化，而非固定能量池的耗尽。
The theory held that self-control draws on a limited resource like a muscle, so performing an inhibition task first weakens subsequent self-control, and early experiments supplied abundant support. Three lines of doubt followed: large multi-lab preregistered replications failed to reproduce the effect reliably, with estimates near zero; meta-analyses corrected for publication bias and small-study effects shrank the original evidence drastically; and the glucose-as-fuel account fitted poorly with physiological measurements. Favoured alternatives now emphasise shifting motivation, attention allocation, and willingness to exert effort rather than depletion of a fixed energy pool.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-11-04
来源 · Source
Hagger 等（2016）《Perspectives on Psychological Science》多实验室重复研究；Inzlicht 与 Schmeichel（2012）《Perspectives on Psychological Science》
Hagger et al. (2016), Perspectives on Psychological Science, multi-lab replication; Inzlicht and Schmeichel (2012), Perspectives on Psychological Science
独立核验依据 · Independent verification: 核验依据 1 Crossref DOI 10.1177/1745691612454134（复合著录按段核验：本标识只担保第2段
#自我损耗 #重复危机 #发表偏差
什么是“抽样偏差”——为什么“网上调查”可能不准？
What is sampling bias—why might an online survey mislead?
样本若不能代表总体（如只访问网民、志愿者），结论就被偏向。代表性随机抽样本难但关键；方便样本适合探索，不宜直接外推。
If the sample can't represent the population (only netizens, volunteers), conclusions skew. Representative random sampling is hard but key; convenience samples suit exploration, not direct generalization.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#抽样偏差 #代表性 #样本
邓宁-克鲁格效应是真实的心理现象，还是统计假象？
Is the Dunning-Kruger effect a real psychological phenomenon or a statistical artefact?
原始研究发现，能力最低的被试显著高估自己的相对排名，而能力最高者略微低估，被通俗解读为无知者最自信。批评者指出，这种图形也可能由回归到均值与量表天花板效应自然产生：由于自评与客观成绩相关不完全，任何按真实成绩分组的比较都会出现低分组高估、高分组低估的模式。经统计模拟等方法重新分析后，学界目前认为元认知能力差异确实存在，但效应的强度被原始呈现方式放大了，且不支持无知者比专家更自信的强版本。
The original studies found that the least skilled participants substantially overestimated their percentile standing while the most skilled slightly underestimated theirs, popularly read as the ignorant being the most confident. Critics noted the same graph can arise from regression to the mean plus scale ceiling effects: because self-assessment correlates imperfectly with performance, any grouping by true score will show low scorers overestimating and high scorers underestimating. After reanalyses using simulation and other methods, the consensus is that genuine differences in metacognitive accuracy exist but the effect's magnitude was inflated by the original presentation, and the strong claim that novices are more confident than experts is not supported.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-11-04
来源 · Source
Kruger 与 Dunning（1999）《Journal of Personality and Social Psychology》；Nuhfer 等（2017）《Numeracy》统计再分析
Kruger and Dunning (1999), Journal of Personality and Social Psychology; Nuhfer et al. (2017), Numeracy, statistical reanalysis
独立核验依据 · Independent verification: 核验依据 1 Crossref DOI 10.1037/0022-3514.77.6.1121
#邓宁-克鲁格效应 #回归到均值 #元认知
为什么“纵向研究”比一次性调查更有说服力？
Why is a longitudinal study more convincing than a one-time survey?
纵向研究在多个时间点追踪同一批人，能看变化顺序与长期结果，比横断面（同一时点）更能逼近因果与发育轨迹，但更贵、易流失。
Longitudinal studies track the same people over time, revealing sequence and long-term outcomes; more than cross-sectional snaps, they approach causation and developmental paths, but cost more and lose subjects.
难度 进阶 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#纵向 #横断 #追踪
什么是“双盲”实验？为什么研究者和被试都不能知情？
What is a double-blind experiment—why must both sides be unaware?
双盲中研究者和参与者都不知道谁在接受处理，避免期望与行为泄露影响结果（如安慰剂效应、观察者偏差）。单盲只藏一方，三盲还含数据分析者。
In double-blind, neither researcher nor participant knows who gets the treatment, blocking expectation and behavior leakage (placebo, observer bias). Single-blind hides one side; triple-blind also hides analysts.
难度 进阶 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#双盲 #期望 #安慰剂
什么是“生态效度”？实验室结论能用到真实生活吗？
What is ecological validity—do lab findings apply to real life?
生态效度指研究发现推广到真实情境的程度。高度控制常牺牲真实（如记忆用无意义音节），所以结论外推要小心，需用现场研究交叉验证。
Ecological validity is how well findings generalize to real life. Tight control often costs realism (memorizing nonsense syllables), so generalization needs field studies to cross-check.
难度 进阶 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#生态效度 #外推 #现场
为什么“没结果”的研究常常看不见？——发表偏倚与文件抽屉
Why do null-result studies often vanish—publication bias and the file drawer?
显著、正面结果更易被发表，空结果常沉在抽屉（文件抽屉问题），使文献整体偏向夸大效应。预注册与要求发表空结果是对策。
Significant, positive results get published more; null results sit in file drawers, biasing the literature to overstate effects (file-drawer problem). Preregistration and mandating null-result publication are remedies.
难度 进阶 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#发表偏倚 #文件抽屉 #空结果
什么是“实验组 vs 对照组”和“随机分配”？
What are treatment vs control groups and random assignment?
随机分配把参与者分入处理组与对照组，平衡已知未知差异，使两组除处理外可比。这是实验能推论因果的核心，比“自然分组”更可靠。
Random assignment splits participants into treatment and control, balancing known and unknown differences so groups differ only by the manipulation. It is the core that lets experiments infer causation, more than natural grouping.
难度 进阶 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#随机分配 #对照组 #因果
什么是“复制危机”？为什么有些心理学发现重复不出来？
What is the replication crisis—why do some findings not replicate?
复制危机指许多已发表结果在重新实验时消失，原因含小样本、p-hacking、发表偏倚（只发显著）。它推动了预注册、大样本与开放数据的改革。
The replication crisis: many published results vanish on re-test, from small samples, p-hacking and publication bias (only significant gets printed). It spurred preregistration, big samples and open data.
难度 进阶 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#复制危机 #p-hacking #预注册
“相关”等于“因果”吗？为什么两件事一起变不代表谁导致谁？
Does correlation equal causation—why co-variation isn't cause?
相关只说明两变量共变，可能 A→B、B→A，或被第三变量 C 同时驱动（如冰淇淋销量与溺水同升，因夏天）。要证因果需控制变量的实验或纵向设计。
Correlation only shows co-variation—A→B, B→A, or both driven by C (ice cream and drowning rise together via summer). Causation needs controlled experiments or longitudinal designs.
难度 入门 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#相关 #因果 #第三变量
问卷真的能量出“真实想法”吗？它有哪些局限？
Can questionnaires really measure true thoughts—what are their limits?
问卷依赖自我报告，易受社会赞许（想答得好看）、记忆偏差与措辞影响，且只能测意识层面。行为、生理与大数据常作补充，单一方法都该谨慎。
Surveys rely on self-report, vulnerable to social desirability, memory bias and wording, capturing only conscious levels. Behavior, physiology and big data help; any single method deserves caution.
难度 入门 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Roger Hock,《Forty Studies That Changed Psychology》Pearson 多版;中译《改变心理学的40项研究》
R. R. Hock, Forty Studies That Changed Psychology, Pearson, various editions; Chinese ed.
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL1845342W：Forty studies that chan
#问卷 #自我报告 #局限