Data presentation
In a minute
A long list of numbers says little until you organise it. You sort data into frequency tables, draw , read the , and off a graph, and use to compare two data sets and judge whether data may be normal.
You'll be able to
- Present and in frequency tables, with written as inequalities that leave no gaps.
- Draw and read a frequency histogram with equal class widths, and describe its shape in context.
- Use a cumulative frequency graph to estimate the median, quartiles, percentiles, range and IQR, and interpret them.
- Draw and read a box and whisker diagram, with each marked by a cross.
- Compare two distributions using their box plots, and judge from symmetry whether data may be normally distributed.
Before you start3 quick questions on what this builds on
Revise first: 4.1 Sampling, data quality and outliers
Frequency tables
Thirty numbers in a row are hard to take in. Count how often each value occurs, and the pattern jumps out.
For with only a few different values, give each value its own row or column. Tally the list once, then count the tally marks.
Try it first, then reveal one step at a time.
- (a) The values run from to , so the table needs a column for each of them.
- Work through the list once, putting a tally mark under each value, then count.
- Check the total: cars.
- (b) “At least ” means , , or people: cars.
- As a percentage: . So almost two thirds of the cars could use the lane.
is different. Measured values are rarely equal, so you group them into classes.
The classes must leave no gaps and must not overlap. With – and –, a mass of g fits nowhere. With and , every mass fits exactly one class.
A value on a boundary goes in the class whose inequality it satisfies. Some tables use the other form, such as , so read the signs every time.
Try it first, then reveal one step at a time.
- (a) Place the awkward values first. satisfies , and satisfies .
- is less than , so it goes in . goes in .
- Tally the rest, then count.
- Check: avocados.
- (b) “ g or more” is the last two classes: large avocados, a little under half the batch.
Histograms
A table gives you the numbers. A picture shows at a glance where the data bunches up, and whether it trails off to one side.
In this course every class in a histogram has the same width. Then each bar's height is simply its frequency, and a taller bar always means more values.
Try it first, then reveal one step at a time.
- Horizontal axis: mass, grams, from to , marked at each class boundary.
- Vertical axis: frequency, from to with an even scale, so the tallest bar, , fits.
- Draw each bar from its lower to its upper boundary, as tall as its frequency: , , and . The bars touch.
- Four touching bars from to g, with heights , , and .
Now look at the shape. If the two sides roughly mirror each other, the data is symmetric. If one side trails off in a long tail, the data is skewed: positively if the tail is on the right, negatively if it is on the left.
Try it first, then reveal one step at a time.
- (a) “At least an hour” is : the three bars from to . Their heights are , and .
- So students, and .
- (b) The bars rise to a peak at , then fall away slowly to . The long tail is on the right, so the times are positively skewed.
- In context: most students spent under an hour, but a few spent far longer than the rest.
Cumulative frequency graphs
A helpline logs the length of calls. How many ended within minutes? Add the frequencies of every class up to : a running total.
1. Add a cumulative frequency column to the table, adding each class's frequency to the total so far.
2. Plot each cumulative frequency above the upper boundary of its class, because only there has the whole class been counted. Start with a point at the first lower boundary, with cumulative frequency .
3. Join the points with straight lines or a smooth curve. This lesson uses straight lines; a curve gives slightly different readings, and both are accepted as estimates.
Try it first, then reveal one step at a time.
- Keep a running total, adding the next frequency each time.
- The last total equals the number of calls, . That is your check.
- Plot , , , , , and , then join them in order.
Now read the graph the other way. The has half the data below it. The lower quartile has a quarter below it, and the upper quartile three quarters. With values, read across from these heights:
The measures the spread of the middle half of the data. It is in the formula booklet.
The is the largest value minus the smallest. A grouped graph hides the actual values, so estimate it from where the graph starts and ends: here minutes. The true range can't be larger than this.
Try it first, then reveal one step at a time.
- (a) The median is at . Read across and down: minutes.
- is at , which gives minutes.
- is at . That lies between and : is of the way up, so minutes.
- So minutes: the middle half of the calls spans about minutes.
- (b) is the th percentile, at . Read across and down: minutes.
- Interpret: about of calls lasted minutes or less, and about lasted longer.
Box and whisker diagrams
A box and whisker diagram, or box plot, draws five key values to scale. You see at once where the data starts and ends, and where its middle half lies.
1. Draw a scale. Draw the box from to , with a line across it at the median.
2. Find the fences and from 4.1. Any value strictly beyond a fence is an .
3. Draw each whisker to the most extreme value that is not an outlier. Mark each outlier with a cross.
Try it first, then reveal one step at a time.
- (a) Enter the nights in one list and the guests in a second, and run one-variable statistics with the second list as the frequency list. Check that .
- Write down what you read: smallest , , , , largest .
- (b) . The fences are and .
- Nothing is below . equals the upper fence, so it is not an outlier. Only , so nights is an outlier.
- (c) Box from to with a line at . Left whisker to , right whisker to , and a cross at .
- A stay of nights is unusual but quite possible, for example a work trip. It is a genuine value, so keep it.
Reading a box plot: about a quarter of the data lies in each of the four parts, from the smallest value to , to the median, the median to , and to the largest.
To compare two data sets, draw their box plots on one scale. Make two comparisons, each with numbers and in context: the medians for a typical value, and the IQRs or ranges for spread. A box plot doesn't show the mean, so don't compare means.
Shape matters too. If the median sits in the middle of the box and the two whiskers are about the same length, the data is roughly symmetric, so it may be normally distributed (4.9). A lopsided box or one long whisker means skewed data, which is unlikely to be normal.
Try it first, then reveal one step at a time.
- First check for outliers. Model P: , fences and . Model Q: , fences and . Every value is inside its fences.
- (a) Typical value: model P has the higher median, h against h. So a model P battery typically lasts longer.
- Spread: model P has the smaller IQR, h against h. So model P's battery life is more consistent.
- (b) Model P: the median is h from each quartile, and both whiskers are h long. Model P is symmetric, so it may be normally distributed.
- Model Q: the median is h above but h below . Model Q is positively skewed, so it is unlikely to be normal.
- Model P has the higher median ( h against h) and the smaller IQR ( h against h), so it typically lasts longer and more consistently. Model P is symmetric, so it may be normal; model Q is positively skewed.
Common mistakes
- Writing classes with gaps, such as – and –, for measured data. Use inequalities such as and .
- Plotting cumulative frequency at class mid-points. Plot it at each upper class boundary.
- Reading “more than” straight off a cumulative frequency graph. The graph counts values at or below, so subtract the reading from .
- Drawing a whisker out to an outlier, or leaving the outlier off. The whisker stops at the most extreme value that is not an outlier, and the outlier gets a cross.
- Leaving out the frequency list when you enter a frequency table on the GDC. Each value then counts once, and the quartiles are wrong.
- Comparing means when asked to use box plots. A box plot shows medians, quartiles and the range, so compare those.
Exam tips
- Estimate from a cumulative frequency graph: write down the height you read across from, such as for , and draw the reading lines on the graph. That earns the method mark.
- Draw a box plot: plot all five values accurately on the given scale, with whiskers to the most extreme non-outliers and a cross for each outlier. The mark needs both whiskers and every outlier right.
- Compare two distributions: write one sentence on the medians and one on the IQRs or ranges, each with the values and in the context of the question.
Key takeaways
Practise this
16 past-paper questions on Statistical measures and frequency tables in the AI HL bank, each with a mark scheme and a worked explanation.