Plotting With Confidence

• Published in: Statistics in Corpus Linguistics Research Vol. 1; pp. 116 - 133
• Main Author: Wallis, Sean
• Format: Book Chapter
• Language: English
• Published: Routledge 2021
• Edition: 1
• Subjects: Confidence Intervals; Wilson Score Interval; Wald Interval; Vice Versa; Time Magazine; Corpus Linguistics Research; Interval Widths; Percentage Difference; Binomial Proportions; Cochran's Rule; Quotative Change; Difference Interval; GB; Cell Scores; Raw Frequencies; OLS Regression; Typical Error Level; Diagrammatic Reasoning; Wilson Centre; 1920s Data; Steepest Gradient; Independent Proportions; Continuity Correction; Independent Binomial Proportions; Contingency Table
• ISBN: 9781138589377; 9781138589384; 1138589381; 1138589373
• DOI: 10.4324/9780429491696-10

Magnus Levin examined the changing meaning of the verb think in progressive contexts (i.e., bethinking) in the TIME Magazine corpus over three decades: 1920s, 1960s and 2000s. Confidence intervals are easier to interpret and more decisive: many can be 'eyeballed' and evaluated, and if only intervals overlap, they can be compared with a robust test. This chapter might remarks on the higher values in the quotative column ('quot', second from the right). Sure enough, this column alone contributes nearly 60% of the total score. Plotting confidence intervals on graphs is a simple way to picture this uncertainty and (hopefully) make the intangible tangible. The chapter shows how the same approach may be applied to Multinomial tables. Plotting data with Wilson score intervals turned out to be considerably more revealing than a traditional approach based on exploring cell values. The process of plotting and visualising data is often overlooked as a statistical method in its own right.