vieses

Vieses, the Portuguese plural of viés, refer to biases—systematic deviations that distort estimates, judgments, or measurements, causing conclusions to differ from objective truth. They can arise at any stage of empirical work, including study design, data collection, analysis, and reporting, and are distinguished from random errors by their predictable direction and magnitude.

Common types are: viés de seleção (non-representative samples), viés de medição (imprecise measurements), viés de confusão (uncontrolled confounding factors), and viés de publicação (tendency to publish significant results). Cognitive biases known in psychology, such as viés de confirmação (confirmation bias) and viés de âncora (anchoring), also influence interpretation.

Causes and contexts: Vieses originate from methodological choices, cognitive processes, and environmental pressures. They are pervasive in statistics, epidemiology, social sciences, journalism, and public policy and can compound when data are incomplete, opaque, or selectively reported.

Mitigation involves design and analysis practices that reduce systematic deviations: random sampling and randomization, blinding, validated measurement tools, calibration, preregistration of hypotheses and analysis plans, full reporting, replication, and sensitivity analyses. Transparent data and code sharing further helps identify and limit biases.

In research and decision-making, acknowledging vieses improves credibility and reliability. While biases cannot always be eliminated, recognizing their presence and documenting assumptions and limitations are essential steps toward robust conclusions.