Уровень 0 · материалов: 4
В кластер входят документы, посвященные способам и примерам того, как статистические показатели и их графическое представление могут искажать реальность или вводить в заблуждение.
Общие признаки: искажение статистических данных, вводящая в заблуждение визуализация, ошибки анализа данных, критический подход к статистике
Группа выше: Визуализация данных: принципы и дашборды
Смысл: The main idea is to expose common techniques used to distort statistical data and visual representations to mislead people, thereby encouraging a critical and skeptical approach to consuming data-driven information.
An educational overview of how sampling bias, manipulated averages, distorted scales, and selective reporting are used to create misleading statistics and visualizations.
Смысл: The main idea is that summary statistics (mean, variance, correlation) can be deceptive because very different datasets can produce identical statistical results; therefore, data visualization is essential for accurate analysis.
Anscombe's Quartet proves that identical statistical summaries can hide vastly different data patterns, making visualization indispensable.
Смысл: The main idea is that the data visualizations used in the 2018 Presidential Address were professionally incompetent, misleading, and failed basic principles of graphic design and statistics.
A professional critique of the 2018 Russian Presidential Address infographics, highlighting severe errors in scaling, labeling, and data representation.
Смысл: The main idea is that aggregate data can be misleading due to Simpson's Paradox, and the only way to ensure accurate analysis is to combine statistical rigor with domain-specific knowledge of the underlying data generation process.
Simpson's Paradox occurs when a trend seen in individual groups reverses when combined, highlighting the necessity of domain knowledge and cohort analysis in data science.