神经科学与心理学2022pnas.org
对长期态度变化的广泛误解
Widespread misperceptions of long-term attitude change
入门 · 约 45 分钟读完前置知识：社会心理学基础、问卷调查方法
导读
这篇 PNAS 论文研究了人们对长期态度变迁的普遍误判：在 51 项态度议题上，参与者高估了 57% 议题的变化幅度，20% 低估，20% 连变化方向都判断错误，仅 4% 估计正确。作者认为人们普遍误以为社会态度变化比实际更快、更剧烈。
HN 关注点：其方法论问题：基于 Mechanical Turk/Prolific 的付费问卷样本是否可靠，以及标题与实际数据之间的偏差。
HN 评论精选
这几条评论里：质疑 2 补充 2
质疑质疑研究样本：MTurk/Prolific 付费问卷的参与者既不具代表性又敷衍，损害研究可信度
这项研究存在严重的人口样本问题（还有其他问题）。社会科学界目前流行用 Amazon Mechanical Turk 和 Prolific 做问卷，这些平台靠几美分报酬吸引人做杂活，聚集了大量'职业'任务执行者。本研究给的'大价钱'是 1.5-2.25 美元。两个主要问题：1) 样本本身就是社会中的一个小众子集，再怎么标准化也没用。这历来是社会科学的老毛病，以前靠社科课本科生，但至少比 MTurk 更接近代表性样本。2) 这些人只想尽快刷完任务拿钱，很多人连最低要求都做不到。本研究确实设置了注意力检查题，但平均每 6 个参与者就有 1 个因答不对'请选 C'这类问题而被剔除。期刊纵容这种方式是因为能制造无尽的'影响力'（越来越像点击诱饵的委婉语），但这正在比以往更严重地损害整个领域。
somenameforme查看英文原文
There is a severe demographic problem with this study (among others). A current trend in the social sciences is using Amazon Mechanical Turk and Prolific to run surveys. These sites are based around paying people pennies to compete menial tasks (like complete surveys), and populated with large numbers of "professional" menial task executers. This study offered the big bucks of $1.50 - $2.25. There are two major problems with this: 1) It doesn't matter how you normalize a sample when that sample is a niche subsample of society to begin with. This has always been a problem with the social sciences which previously relied heavily on undergraduates in social science classes, but that is going to at least be much closer to something like a representative sample than Amazon Mechanical Turk or Prolific can offer. 2) The people carrying out the tasks are motivated primarily by blazing through as many studies as possible while doing the bare minimum to get paid, and many fail at even that. In this study ctrl+f for "attention check." To this study's credit, they at least included numerous attention checks, but it emphasizes the issue. On average about 1 out of 6 participants were excluded for failing to correctly answer questions along the lines of "Please select C as the answer to this question." Journals are likely indulging this because it's going to generate endless "impact" (which is increasingly becoming a euphemism for clickbait), but it's undermining the entire field more than ever before.
补充类比史学史：职业历史学家也曾长期高估'变化'，普通人有此偏差不足为奇
要记得同样的问题困扰了史学至少两个世纪。从 1700 年代到 1900 年代，历史学家自视为'事件'的记录者而非'非事件'的记录者，因此自动高估了变化。直到 19 世纪末现代人类学兴起才有人考虑其他路径，而重大转变通常归功于年鉴学派和布罗代尔的'长时段'（longue durée）概念——优先研究长期历史结构而非编年史家式的短期事件史。我们不该指望普通人搞对，因为直到最近职业历史学家都强烈偏向事件（现在许多人依然如此）。
lkrubner查看英文原文
It's important to remember that this same problem haunted historiography for at least 2 centuries, if not much more. At least from the 1700s, and lasting well into the 1900s, historians considered themselves the historians of events, rather than non-events. As such, they automatically gave too much weight to change. That a different approach was possible was not even considered until modern anthropology got going in the late 1800s. But credit for the big change in historiography is typically assigned to the Annales school and to Fernand Braudel in particular, and his concept of the "Longue durée": https://en.wikipedia.org/wiki/Longue_durée "The longue durée is the French Annales School approach to the study of history. It gives priority to long-term historical structures over what François Simiand called histoire événementielle ("evental history", the short-term time-scale that is the domain of the chronicler and the journalist), concentrating instead on all-but-permanent or slowly evolving structures, and substitutes for elite biographies the broader syntheses of prosopography. The crux of the idea is to examine extended periods of time and draw conclusions from historical trends and patterns." We should not expect ordinary layman to get things right when professional historians, until recently, had a strong bias toward events (and even now, many still demonstrate this bias).
补充纠正标题偏差并补充数据：人们既高估也低估，历史上反而常低估社会支持度
论文实际标题是'对长期态度变化的普遍误解'，而不是'研究：美国人一贯高估文化变迁程度'。摘要显示：51 项态度中，参与者高估了 29 项（57%）的变化，低估 10 项（20%），方向判断错误 10 项（20%），仅 2 项（4%）估计正确。读摘要的感觉是美国人根本不了解现状——考虑到近期变化很快、且互联网让我们社交圈更封闭导致估计有偏，这更说得通。有意思的是，70 年代的人严重低估了许多事：比如 1972 年问及有严重风险的堕胎，人们估计只有 49% 支持，实际是 87%；现在感知为 71%，实际 90%。气候变化方面，1990 年人们估计 28% 关心，实际 61%；现在是 65.5 对 65。光看标题会错过很多有趣的数据。
godelski查看英文原文
The actual title is > Widespread misperceptions of long-term attitude change Not "Study: Americans consistently overestimate how much culture has changed." And from the abstract > Participants overestimated the amount of change on 29 attitudes (57%), underestimated change on 10 attitudes (20%), estimated change in the wrong direction on 10 attitudes (20%), and estimated change correctly on only two attitudes (4%). (out of 51 different attitudes) Maybe we can change the title. When I read this abstract I got the feeling that Americans just don't know what's going on. Which honestly makes far more sense. Things are changing rapidly recently. It especially makes sense when you consider we're isolating our social groups more (internet makes this easier but isn't a sole contributor) and so there would be a biased estimate for over estimation. What's also really interesting is that going through some of these is that it seems that people in the 70's greatly underestimated many things. For example when asked about abortion with there was serious risk people estimated (in 72) that only 49% of people were in favor but actually 87% were in favor. Now perception has changed to 71 and actual to 90. Another example is concerns about climate change. In 1990 people estimated 28% were concerned but actually 61% were concerned. Now it is 65.5 vs 65. If you just came here for the title it really does look like there is a lot more interesting data than what the title says. It does seem like some things we've gotten better at getting estimates on while others are more difficult. Worth at least looking at the figures. But definitely could be some interesting followup analysis.
质疑质疑自我报告数据的真实性，建议改用行为指标
这似乎假设人们在民调中说实话，但我们知道并非如此（如《纽约时报》报道美国人声称去教堂的频率远高于实际）。最好尽可能用行为指标，比如真实选举结果。
nitwit005查看英文原文
This seems to assume people are being honest on these polls, but we know they are not: https://www.nytimes.com/2014/05/18/upshot/americans-claim-to-attend-church-much-more-than-they-do.html It's better to find a behavioral measure, like actual election results, to look at if at all possible.
历次讨论
- 2022Hacker News 讨论帖208 分 · 188 评
同领域还值得读
神经科学家能理解微处理器吗？（2017）
Could a Neuroscientist Understand a Microprocessor? (2017)
提出一个思想实验：把神经科学常用的分析方法（损伤实验、电生理、功能定位等）应用到一个我们完全了解其原理的工程系统——6502 微处理器（运行游戏）上，检验这些方法能否真正“理解”…
神经科学与心理学2017 · 411 分 · 135 评进阶 · 约 60 分钟编辑精选大脑衰老呈现非线性转变，提示中年“关键窗口”
Brain aging shows nonlinear transitions, suggesting a midlife "critical window"
这篇 PNAS 论文研究大脑衰老的非线性变化，发现大脑老化并非匀速进行，而是在中年（约 40-60 岁）存在一个关键的"临界窗口期"，此时干预可能有效。
神经科学与心理学2025 · 295 分 · 159 评进阶 · 约 75 分钟编辑精选社交网络中大规模情绪传染的证据（2014）
Evidence of massive-scale emotional contagion through social networks (2014)
这篇 2014 年发表于 PNAS 的论文研究了 Facebook 上约 70 万用户的信息流中，减少正面或负面内容是否会传染性地影响用户自己发布内容的情绪。
神经科学与心理学2020 · 285 分 · 72 评入门 · 约 45 分钟编辑精选裸盖菇素使人类大脑去同步化
Psilocybin desynchronizes the human brain
这篇 Nature 研究用个体化的 fMRI 精密成像追踪受试者服用裸盖菇素（psilocybin）前后的脑功能连接变化，发现药物会造成大范围的脑网络「去同步化」，尤其是默认模式…
神经科学与心理学2024 · 273 分 · 259 评进阶 · 约 60 分钟编辑精选对智力的赞扬会削弱儿童的动机与表现（1998）
Praise for Intelligence Can Undermine Children's Motivation (1998)
这篇 1998 年的经典论文（Dweck 等人）通过实验研究表扬方式对儿童动机的影响：夸孩子“聪明”（表扬智力）会让他们回避挑战、害怕失败，而表扬努力则更能培养坚持性。
神经科学与心理学2016 · 253 分 · 155 评入门 · 约 45 分钟编辑精选