Apps

🧪 This platform is in early beta. Features may change and you might encounter bugs. We appreciate your patience!

Proxy Bias

Also Known As: Measurement error bias Construct validity bias
Aspect ID: proxy_bias

Definition

Proxy bias occurs when indirect measures are used in place of the true construct of interest, and the gap between the proxy and the true construct is correlated with other variables in the statistical model. Unlike random measurement error (which attenuates associations), proxy bias creates systematic distortions because the mismatch between what is measured and what is meant is not random.

Examples

Household income is used as a proxy for socioeconomic status in a model that also includes race. If the income-to-SES gap differs systematically by race (e.g., because of wealth disparities not captured by income), then the race estimate in the model partly reflects the proxy-SES mismatch, biasing both coefficients.

A tech company uses number of GitHub commits as a proxy for software engineer productivity in performance reviews. This disadvantages engineers who do deep architectural work, mentoring, or documentation — contributions that generate few commits but enormous team value. The proxy captures only one visible slice of the true construct.

A public health study uses zip code as a proxy for environmental pollution exposure. In heterogeneous urban zip codes, some residents live next to a highway while others live far from it. The proxy introduces substantial measurement error that is not random — it systematically mismeasures exposure for residents in large, mixed zip codes, biasing effect estimates.

Verification Steps
Verification Steps
Binary yes/no questions that an AI must answer to detect a reasoning pattern in a text.
Each of the 452 aspects has verification steps — simple yes/no questions designed to systematically detect whether a pattern appears in a text. For ad hominem: "Does the argument attack a person rather than their claim?" For false dichotomy: "Are only two options presented when more exist?" This ensures consistent, reproducible analysis.

Binary (yes/no) questions an LLM must answer to identify this aspect:

  1. 1

    Is the study using an indirect measure (proxy) rather than directly measuring the construct of interest?

    Type: binary
  2. 2

    Is the gap between the proxy and the true construct correlated with other variables in the model?

    Type: binary
  3. 3

    Could the proxy-target measurement gap introduce systematic bias in estimates?

    Type: binary
  4. 4

    Has the validity of the proxy been assessed against the true construct in the study population?

    Type: binary
Deep Dive
The expandable detail section on each aspect page with examples, psychology, and counter-strategies.
The Deep Dive section provides in-depth information about each aspect: a real-world example showing the pattern in action, an explanation of why it works psychologically, practical advice on how to counter it, alternative names, and links to related aspects.