
Same numbers, different stories
Four data sets share every summary statistic. Graph them and they tell four different stories.
Four lab groups measured how a quantity \(y\) depends on a setting \(x\). All four handed in the same summary: mean \(y = 7.50\), best line \(y = 0.500\,x + 3.00\), \(R^2 = 0.67\). Did they find the same thing?
Same means, same line, same \(R^2\). What will the four graphs look like?
Graph them
Match the story
Which group's data follow a smooth curve, not a line?
Which group has one reading that is probably a copying or measuring mistake?
Which group barely varied the setting it was testing?
Which group's line is a fair summary of its data?
Fix the analysis
Above group B's graph, switch the fit to Parabola. What is \(R^2\) now?
In the table, delete group C's suspicious value (\(y_C = 12.74\) in row 3). What slope \(k\) does group C's line have now?
Group D claims: “\(y\) grows with \(x\), with slope 0.5.” Do you agree? Give a claim, evidence from the graph, and your reasoning.
The statistician Frank Anscombe built these four data sets in 1973 to make one point: graph your data before you trust any number calculated from it.
Habit to keep: graph first. Means, fits and \(R^2\) come after you've seen the shape.