Lesson 3 of 8 · 12 min

Percentaged two-way tables

Convert counts to percentages within each category of the explanatory variable, so that groups of different sizes can be compared fairly.

What you will be able to do

  • Explain why counts can mislead when the groups being compared are different sizes.
  • Decide whether to calculate row or column percentages from where the explanatory variable sits.
  • Construct an appropriately percentaged two-way frequency table and check it.

Back to the study survey. 66 Year 11s study at home, compared with only 28 Year 12s. Does that mean Year 11s are more likely to study at home?

The counts can't tell you, because the groups are different sizes: there are 120 Year 11 students and only 80 Year 12 students. A bigger group will usually have bigger counts in every column. To compare the year levels fairly, put both on the same scale — out of 100.

Key idea

Compare percentages, not counts

To look for an association, calculate percentages within each category of the explanatory variable. Each category then adds to 100%, and the groups can be compared however big or small they are.

A table set up this way is an appropriately percentaged two-way frequency table.

Percentaging a two-way frequency table

  1. Identify the explanatory variable

    It defines the groups you want to compare — here, year level.

  2. Find where its categories are

    Down the side (rows) or across the top (columns).

  3. Divide each count by its category total and multiply by 100%

    Use row sums if the explanatory variable is in the rows, column sums if it is in the columns.

  4. Check each category adds to 100%

    Rounding can make a total 99.9% or 100.1%. Anything further out means a slip.

The study survey, percentaged within each year level (row percentages). Year 11: 66120×100%=55%\frac{66}{120} \times 100\% = 55\%, and so on.
Year levelHomeLibrarySchoolTotal
Year 1155%25%20%100%
Year 1235%45%20%100%

Now the comparison is fair. 55% of Year 11s study at home compared with 35% of Year 12s, and 45% of Year 12s study in a library compared with 25% of Year 11s. The same 20% of each year level study at school.

A segmented bar chart shows the same percentages as a picture: one bar for each category of the explanatory variable, each bar 100% tall.

Study location within each year level. The segments differ between the bars, which is the first sign of an association.

A survey asked 400 adults for their age group and their main source of news.

Age groupOnlineTVRadio or printTotal
18–34963321150
35–54525226130
55 and over186636120
Total16615183400

Age group is the explanatory variable. What percentage of adults aged 55 and over get their news mainly from TV?

in \%

Tables don't always put the explanatory variable down the side. When its categories are across the top, percentage down each column instead, so each column adds to 100%.

Worked example

Percentage a table with the explanatory variable in the columns

  • Simple familiar
  • Technology active

A Queensland council surveyed 300 residents about a proposed off-leash dog park, recording each resident's age group and opinion.

Opinion18–3940–5960 and overTotal
Support724824144
Oppose303844112
Unsure18141244
Total12010080300

(a) Identify the explanatory variable.

(b) Construct an appropriately percentaged table.

5 steps

Try it yourself first, then check the working one step at a time.

Method

  1. Identify the explanatory variable and whether its categories are the rows or the columns.

  2. Divide each count by its category's total (here, the column sum) and multiply by 100%.

  3. Check each category adds to 100%.

Watch out for

  • Dividing by the row sums, e.g. 72144×100%=50%\frac{72}{144} \times 100\% = 50\%. That is the percentage of supporters who are aged 18–39, which depends on how many residents of each age group were surveyed.

Common mistake

Percentaging by the response variable

Percentaging the study survey down the columns gives "45.5% of library users are in Year 11 and 54.5% are in Year 12". That's a true statement, but it answers a different question: who uses the library, not how likely each year level is to use it. Because there are more Year 11s overall, these percentages are pulled around by the group sizes. Always percentage within the explanatory variable.

A survey records each student's type of school (State, Catholic, Independent) across the top of a two-way table and whether they play a school sport (Yes, No) down the side. To see whether playing a school sport is associated with type of school, how should the table be percentaged?

Choose one answer

Tip

Rounding percentages

Give percentages to one decimal place unless the question says otherwise. Because of rounding, a category may add to 99.9% or 100.1% — that's fine, and you can say so.

Summary

  • Counts can't be compared fairly when the groups are different sizes; percentages can.

  • Percentage within each category of the explanatory variable, so each category adds to 100%.

  • Explanatory variable in the rows means row percentages; in the columns means column percentages.

  • A segmented bar chart shows a percentaged table with one 100% bar per category of the explanatory variable.

Finished the lesson?

Try the 2 quick checks above and the lesson completes itself, or mark it complete now.

Lesson 3 of 80% read

Save your progress with a free account