Choosing and judging statistical representations

Choosing and judging statistical representations

Secondary 4

Which display for which purpose

Each display keeps some information and loses some. Choose the display that shows what the question needs.

DisplayShows wellCannot show, or drawback
Stem-and-leaf diagramevery value, in order; the median and quartiles exactlyis hard to draw and read for a large data set
Dot diagrameach value as a dot; clusters, gaps and an isolated valuesuits only small data sets
Histogram (equal class widths)the frequency of each class and the shape of the distributionthe individual values; the median only to its class
Cumulative frequency curvethe number of values up to any value; estimates of the median, quartiles and percentilesthe individual values; class frequencies only by subtracting readings
Box-and-whisker plotthe five-number summary; a quick comparison of two groups on one scalethe number of values, the individual values, or how the values lie inside each quarter
Bar graphcounts of separate categoriescontinuous data such as heights or times
Pie charteach category as a part of one wholethe total; two pie charts of different totals cannot compare counts

Misleading displays

Both charts show the same sales: \(60\) and \(64\) units. On the left the vertical axis starts at \(56\), so the second bar looks twice as tall although sales rose by only \(\dfrac{4}{60}\times100\%\approx6.7\%\). On the right the axis starts at \(0\), so the bar heights keep the true ratio \(60:64\) and the two bars look almost the same height.

  • A vertical axis that does not start at zero exaggerates differences between bar heights. Describe the effect, not just that the chart “looks wrong”.
  • Percentages or pie charts without totals cannot show which group has more items.
  • In a box plot each of the four sections holds about a quarter of the values. A longer whisker shows a wider spread of that quarter, not more values.

Choosing an average

The mean uses every value, so one extreme value pulls it towards itself. The median depends only on the middle of the ordered data. The mode is the most common value and can be used for categories. The waiting times \(4,5,5,6,7,30\) minutes have mean \(\dfrac{57}6=9.5\) minutes, larger than five of the six times; the median, \(\dfrac{5+6}2=5.5\) minutes, describes a typical wait better.

Checks

  • Read the scale and where each axis starts before comparing heights.
  • Compare groups of different sizes with percentages, not counts.
  • Justify a choice of average with the data: name the extreme value or the reason the mean is pulled.

Need help? Join our E-Maths tuition for O-Level success.

Learn more