如何识别资料来源中的偏见

• 点名制作者。 “据 Heritage Foundation，该机构称自己推动保守派公共政策……”
• 把事实与框架分开。 使用能在表格、方法注释或已评分选项中打开的数字。不要把随数字而来的形容词一并带入。
• 把工作配对。 使命驱动的简报是该组织论点的证据。Congressional Budget Office 的成本估计是已评分的预算或就业效应的证据。Cochrane review 是综合的证据。不要让一份顶替另外两份。

Cochrane Handbook, Chapter 7 (version 6.5, 2024; chapter last updated August 2022)

We define bias as a systematic error, or deviation from the truth, in results.

ACRL Framework for Information Literacy for Higher Education (2016)

Experts understand the need to determine the validity of the information created by different authorities and to acknowledge biases that privilege some sources of authority over others, especially in terms of others’ worldviews, gender, sexual orientation, and cultural orientations.

Chapter 7, §7.1.1

Bias should not be confused with imprecision. Bias refers to systematic error, meaning that multiple replications of the same study would reach the wrong answer on average.

Chapter 7, §7.1.1

it is more appropriate to consider whether a result is at risk of bias rather than claiming with certainty that it is biased.

官方释义

Purpose: The reason the information exists.

• What is the purpose of the information? Is it to inform, teach, sell, entertain or persuade?
• Do the authors/sponsors make their intentions or purpose clear?
• Is the information fact, opinion or propaganda?
• Does the point of view appear objective and impartial?
• Are there political, ideological, cultural, religious, institutional or personal biases?

Note that reputation and process might be apart from issues of bias: the New York Times is thought by many to have a center-left bias, the Wall Street Journal a center-right bias, and USA Today a centrist bias. Yet fact-checkers of all political stripes are happy to be able to track a fact down to one of these publications since they have reputations for a high degree of accuracy, and issue corrections when they get facts wrong.

CBO About / Introduction to CBO

CBO conducts objective, impartial analysis and hires employees without regard to political affiliation. The agency does not make policy recommendations.

An Introduction to the Congressional Budget Office

CBO makes no policy recommendations because choices about public policy inevitably involve value judgments that the agency does not and should not make.

• Heritage 简报（倡导研究）。 James Sherk, “Raising Minimum Starting Wages to $15 per Hour Would Eliminate Seven Million Jobs”（2016年7月26日）。简报主张，到 2021 年的联邦 15 美元下限将消灭“约七百万个全时当量（FTE）岗位”。把它当作 Heritage 的论点来引用。不要称它超党派，也不要把 2016 年里的 2021 年视野当成 CBO 的 2025 年选项。
• CBPP 入门（研究加倡导）。 Center on Budget and Policy Priorities, “The Minimum Wage”（Policy Basics，2026年4月16日）。该页写明联邦最低仍为 7.25 美元，截至 2026年3月有三十个州和哥伦比亚特区更高（十九个达到 15.00 美元或以上），且 BLS 统计 2024 年有 843,000 名工人按联邦最低或以下领取报酬。引用你能打开的 BLS 数字；点名 CBPP 是同时做倡导的制作者。
• CBO 报告（已评分选项，无建议）。 Congressional Budget Office, The Effects on Employment and Family Income of Increasing the Federal Minimum Wage（出版物 55410，2019年7月8日）。CBO 考察到 2025 年每小时 10、12 或 15 美元，并给出两项主要效应：多数低薪工人收入会提高，部分人会失业。仍要核对该报告中的假设。

Heritage (Sherk 2016)

Sherk, J. (2016, July 26). Raising minimum starting wages to $15 per hour would eliminate seven million jobs. The Heritage Foundation. https://www.heritage.org/jobs-and-labor/report/raising-minimum-starting-wages-15-hour-would-eliminate-seven-million-jobs

CBPP（Policy Basics，2026年4月16日）

Center on Budget and Policy Priorities. (2026, April 16). The minimum wage. https://www.cbpp.org/research/policy-basics-the-minimum-wage

CBO（出版物 55410，2019年7月8日）

Congressional Budget Office. (2019, July 8). The effects on employment and family income of increasing the federal minimum wage. https://www.cbo.gov/publication/55410

据 Heritage Foundation，该机构称自己推动保守派公共政策，到 2021 年的联邦 15 美元起始工资将消灭约七百万个 FTE 岗位（Sherk, 2016）。声明自己不做政策建议的 Congressional Budget Office 考察了到 2025 年的 10、12 和 15 美元选项，并报告更高的联邦下限会提高多数低薪工人的收入，同时使其中一些人失业（Congressional Budget Office, 2019）。CBPP 2026年4月的入门来自也把自己的工作描述为倡导的机构，报告了现行联邦下限 7.25 美元，以及 BLS 2024 年按该下限或以下领取报酬的工人人数。

参考文献表

Lundh, A., Lexchin, J., Mintzes, B., Schroll, J. B., & Bero, L. (2017). Industry sponsorship and research outcome. Cochrane Database of Systematic Reviews, 2017(2), Article MR000033. https://doi.org/10.1002/14651858.MR000033.pub3

正文中

一项涵盖 75 篇论文的 Cochrane 方法学综述发现，产业资助的药品与器械研究比其他资助者的研究更常报告有利的疗效结果（Lundh et al., 2017）。这是关于赞助的发现，不是丢掉每一项产业试验、或把每一项非产业试验都当成干净的许可证。

The potential for conflict of interest and bias exists when professional judgment concerning a primary interest (such as patients' welfare or the validity of research) may be influenced by a secondary interest (such as financial gain).

Perceptions of conflict of interest are as important as actual conflicts of interest.

Cochrane Handbook, Chapter 7, Table of reporting biases

The publication or non-publication of research findings, depending on the nature and direction of the results.

Cochrane Handbook, Chapter 13 (version 6.5, 2024)

ROB-ME is the first structured approach for assessing the risk of bias that arises when entire studies, or particular results within studies, are missing from a meta-analysis because of the P value, magnitude or direction of the study results (Page et al 2023).

John P. A. Ioannidis, PLoS Medicine（2005年8月30日）

Moreover, for many current scientific fields, claimed research findings may often be simply accurate measures of the prevailing bias.

Ioannidis, J. P. A. (2005). Why most published research findings are false. PLoS Medicine, 2(8), Article e124. https://doi.org/10.1371/journal.pmed.0020124

• 作为视角接受。 “Heritage 主张……”、“CBPP 认为……”、“在一封你实际打开过的信中……”。在句子里点名制作者，而不只在括号里。
• 作为可核验事实接受。 表格、法规、方法注释、你打开过的 DOI。引用数字，不要引用口号。
• 用不同工作来平衡。 把倡导简报与机构评分、系统综述或另一学派的论文配对。把两篇已经同意的博客配对不是平衡。把 Sherk (2016) 与 CBO (2019) 配对更接近，你仍须说明它们评的是不同日历。
• 不要洗白。 当唯一来源是一份受资助简报时，不要写“研究表明”。不要把智库主张移进你的语气，再引用一篇你没读过的期刊。
• 不要编造另一边。 若打不开真实对应物，就如实写，或收窄主张。

• 确认。 只查询已经符合论题的短语。
• 锚定。 把第一篇文章当成此后所有文章的框架。
• 可得性。 过度加权你记得住、或图书馆立刻载入的研究。
• 权威 / 内群体。 不经横向检索就信任著名实验室、熟悉报纸或“我们这类人”。

• 事实被捏造，或你打开的数字与正文中的数字不符。
• 相反证据被压制，且方法无法核对。
• 本应重要的利益冲突未被披露。
• 设计被建成保证想要的结果（未设盲的结局、不可比的剂量、看过之后切换结局）。
• 反对观点只以稻草人出现，或页面是为销售、招募或煽动而存在。
• 你无法判断谁该负责。