Numbers in the News: A Survey

Some 250 number-focused news articles have been analyzed for the prevalence of statistically-related content. Ninety-five factors were analyzed for each story. These included (a) elements of study design (controlled, control for related factors, longitudinal, cohort, experi-ment and random assignment), (b) statistical measures (ranging from simple counts and percentages to confidence intervals and p-values) and (c) types of grammar used to describe and compare ratios (percent, percent-age, rate, chance, and likely). Of these articles, 65% involved assembly in constructing categories or measures (e.g., bullying), 55% presented associations that implied causation and 41% used percents. While about half (42%) gave the sample size, only 6% mentioned that the sample was random. While 10% used the word “significant”, only 2% mentioned “statistically significant.” None mentioned p-value. Articles were classified as study-focused, study-related, surveys, diagnostic tests and others. Of the survey articles, only a third (35%) gave the margin of error. Hopefully this kind of empirical data will be useful in determining the emphasis of these topics in teaching statistical literacy as applied critical thinking.