How to read potentially confusing charts

This guide is one in a series on different aspects of statistical literacy. The others in this series can be found in the House of Commons Library's Good Information Toolkit page.

The way readers interpret charts tends to be a combination of an initial impression or ‘gut feeling’ and conscious analysis of the chart’s individual elements. A well-designed chart should clearly illustrate the patterns in its underlying data and should be straightforward to understand – even if it has multiple variables, and uses different visual elements and colours.

Charts can illustrate a variety of patterns, such as the direction of change in a value over time, the contribution of different parts to the whole, the scale of differences between groups and so on.

This guide looks at some common examples of charts where basic patterns in the underlying data are not always clear to readers. This could be because of the chart’s type or design, or because of the nature of the underlying data.

Charts with non-standard axes

Charts with non-standard axes can be more difficult to interpret at first glance. This can include axes that don’t start at zero or don’t increase in regular increments.

Shortened value axis

in many charts is a shortened value axis. This is where instead of starting at zero the axis starts at some greater value. As a result the variations shown in the chart can appear greater than they are.

The two charts below, showing accepted applicants to UK universities through UCAS, illustrate this point. The chart on the left has a full-value vertical axis, which starts at zero. The chart on the right has a shortened value vertical axis, which starts at 0.4 million.

At first glance, the chart on the right appears to show an increase of around 150% in accepted applicants. However, the actual increase is just 22% – as shown by the chart on the left. When looking at the right-hand chart, it is easy to miss that the vertical axis does not start at zero and therefore the increase it shows appears greater than it actually is.

Source: UCAS Undergraduate end of cycle data resources 2024

To better understand the data, we first need to spot a shortened value axis. In some cases, there will be a gap at the foot of the vertical axis (as in the right-hand chart above), or a symbol like a zig-zag line. In other cases, there will be no visible indication and the only way to be sure is to carefully read the axis values.

Compressed date axis

Charts that show changes in data over a period of time (also called ‘time series charts’) sometimes use a compressed date axis. This means they might show data from a series of irregular dates (such as 2006, 2009, 2011 and 2016) as if they were regular, with equal gaps. However, by doing so they effectively compress or skew the horizontal axis. The two charts below, showing the estimated proportion of households in fuel poverty in Northern Ireland, illustrate how compressed dates can change the patterns we see in charts.

Source: NI Housing Executive, Estimates of fuel poverty in Northern Ireland in 2019 (and earlier)

The chart on the left condenses the dates and appears to show a sharper increase at the start of the period, and an even sharper decrease after 2011. The chart on the right shows how it would look if this data was presented with appropriate gaps between years. It illustrates the ‘true’ visual pattern in the underlying data: it shows that there were gaps in the data and if we were to interpret the trends over the gap years, the annual changes over these periods were much smaller than those implied by the chart on the left.

The most important steps here are to identify whether a chart contains any coby carefully reading the axis labels, and to recognise that such charts might not accurately reflect patterns in the underlying data. It will not always be possible to picture the ‘true’ pattern, especially if a number of gaps have been compressed (like dates in the left-hand chart below). In such cases, we might need to get the underlying data or estimate its approximate values if the data isn’t available.

Logarithmic charts

Logarithmic charts use values that appear to ‘jump’ – they might start with 1 and then increase to 10, 100, 1000 and so on. Instead of a standard linear scale with even gaps (such as 10, 20, 30 and so on) they have a logarithmic scale where each step is a multiple of the last (normally 10).

Logarithmic charts compress data ranges and allow the reader to view trends which can be difficult to understand on a linear scale.

As with a conventional scale, a rise in the data will be shown as an upward slope and a fall in the data will be shown as a downwards slope. However, with a logarithmic chart an upward sloping straight line indicates that the data is increasing at a constant rate (meaning the value grows by the same percentage each period). It does not mean a constant increase in absolute terms (where the value grows by the same amount each period). If the rate of change is increasing, the line will bend upwards in a logarithmic chart. If the rate of change is falling, the line will flatten towards the horizontal axis. This helps to focus on relative rather than absolute changes.

Logarithmic charts are often used to show exponential growth where a conventional linear scale would make it difficult to identify changes in the rate of growth or decline. For example, the two charts below look at the same covid-19 test data. The first chart has a conventional linear scale, while the second chart has a logarithmic scale. In the first chart it is difficult to see how the number of covid-19 cases was progressing towards its peak. The vertical differences are too large to see when the rate of growth started to slow. This is clearer on the logarithmic chart.

Source: UKHSA, COVID-19 archive data download (Overview, new cases by specimen date)

It is harder to get an accurate impression of absolute values in charts with logarithmic scales as often you cannot simply read off values from the vertical axis. Therefore, the key here is to identify whether charts have a logarithmic scale by carefully reading the axis labels, and then focus on how the line depicts proportionate changes.

Potentially confusing chart types 3D charts

3D charts are difficult to read because of their ambiguity. This is illustrated by the chart below, which shows education spending per capita in the UK for financial year 2022-23. It is not clear whether the value should be taken from the front of the bar, the back of the bar, or from somewhere else. In this case, data labels (in the form of numbers) have been added to the bars to highlight this ambiguity.

The point of a chart is to illustrate patterns rather than give precise values. In this example, it is difficult to judge accurately what the actual values are from the size of the bars alone. The 3D effects also distract from the underlying patterns of the data.

Source: HM Treasury, Public Expenditure Statistical Analyses 2024

There is no easy way around this issue. The use of data labels and gridlines in 3D charts might help (like in the chart above). However, if those are missing, the best approach is to focus on the relative height of the bars rather than attempt to read across to the vertical axis.

Multiple pie charts

Pie charts are commonly used to visualise contributions of different elements to a whole. They can also compare these for different groups, areas or periods. The pie charts below compare the UK’s public expenditure on transport for two financial years: 2019-20 and 2023-24.

Source: HM Treasury, Public Expenditure Statistical Analyses 2024

These charts show that the relative size of roads spending was smaller in 2023-24, even though its absolute value increased to £12.2 billion. Its share of the total fell in 2023-24 because overall transport spending increased proportionately more than roads spending. Pie charts are better at showing proportionate contributions from a small number of elements and might give the wrong impression in cases like this.

Occasionally the slices in pie charts are scaled in proportion to the difference in totals. In the example above, the total area of the right-hand chart (the circle) would be 34% bigger than the left-hand circle, to represent the increase in total transport expenditure from £18.2 billion to £26.8 billion. However, the pattern of the underlying data could still get lost as the human eye can find it more difficult to compare curved areas than the length of bars or positions on a line.

The key to better understanding data presented in multiple pie charts is to carefully read the values to identify if they are looking at different totals. If you know the individual values or the totals, you can make some assessment of relative changes – although this is not always straightforward. If there are no absolute values available, either for individual elements or the total, charts could potentially be misleading.

Charts with two trendlines and value axes

A time series is a set of values of a particular variable (such as the price of something), which can be represented in a chart as a trendline, with time on the bottom (x) axis and the value on the side (y) axis.

Charts with multiple time series all using the same value axis, where the value is in the same units, let the reader compare how the different series have changed over time. For example, this could be two lines showing the price of two items, in pounds per item, over time. Here the reader can identify relative values of each series, gaps between them, crossing points and so on.

However, it is not possible to show multiple data series which use different units on a single axis. The example below is such a case.

Sources: DESNZ, Monthly and annual prices of road fuels and petroleum products; Ofgem, Energy price cap levels 1 July to 30 September 2025: Final levelised cap rates model (Annex 9)

The oil prices (per litre) are plotted on the left-hand axis, while the gas prices (per kilowatt hour) on the right-hand axis.

The first step in interpreting such charts is to identify which series is plotted on which axis (in the chart above, this is clear from annotations on the chart and different coloured lines). The second step is to focus separately on how each series has changed over time: is it going up or down? Has that changed over the time period shown? If so, when?

The relative positions of the different series are not important, such as the crossing points or the gaps between them. The version below illustrates this. Changes to the chart scales change the relative positions of the two lines, but not how each series has changed over time.

Sources: DESNZ, Monthly and annual prices of road fuels and petroleum products; Ofgem, Energy price cap levels 1 July to 30 September 2025: Final levelised cap rates model (Annex 9)

Scatter plots

Scatter plots are charts normally used to look at the relationship between two variables, to help identify whether and how they are associated (a positive association means when one goes up the other goes up too; a negative association means when one goes down the other goes up). They can also help identify ‘outliers’ from a general pattern.

However, a visual display like a scatter plot can only give an impression of the relationship between two variables. It will be open to interpretation by the reader unless the association is very strong (meaning all the dots form a sharp line) or very weak (meaning the dots are distributed randomly). In practice, virtually all scatter plots will be somewhere between these two extremes.

Sources: Defra, Local authority collected waste management - annual results

The example above appears to show a negative association: authorities with higher amounts of residual waste per household had lower recycling rates. However, there are many outliers and the degree of association shown is open to interpretation.

The only way to tell definitively is to read any accompanying text to see whether the relevant regression statistics have been calculated (the percentage of the variation in one variable that was associated with variation in the other). These tell us how changes in one variable affect the other, and the strength of any association.

In the example above the regression statistics show the following:

• The slope of the line of best fit was significantly less than zero (negative).
• A 10 kg increase in residual waste per capita was associated with recycling rates that were 0.8 percentage points lower.
• 48% of the variation in one indicator was explained by variation in the other.

The guides in this series on regression and confidence intervals and statistical significance give more background.

Charts with potentially confusing data Index data

Index charts show change over time in relative (percentage) terms. They can help to simplify data, but can also be confusing for readers, especially when they contain more than one time series in index form.

In most cases the underlying data is converted to an index where the base value is set to 100. The title of the chart and/or the axis should indicate that it is showing an index rather than actual values. It should also give the base year, for example ‘2017=100’ would indicate that the value in 2017 is the base value, set to 100, and the values for all other years are relative to this year.

The index number for a specific point in time shows how much it differs from the base value in percentage terms. For example, a value of 110 means it is 10% higher than the base value, not 10 units of the underlying values (such as £10 or 10 people). A chart showing index values will have the same shape as a chart showing underlying data, but the values on the vertical axis will be different.

An index chart may include more than one time series. This can be useful when the series are measured in different units which cannot be charted on the same axis, such as number of an item sold and price per item. We might look at the gaps between the series or their crossing points and interpret these as patterns in the underlying data.

The two charts below show passenger transport by mode in Great Britain. The left-hand chart shows that the value for rail was higher than for air travel. This does not mean that rail travel became more popular than air travel in 2010, but that the increase in rail travel between 1990 and 2010 was greater than the increase for air travel.

Source: DfT, Transport Statistics Great Britain (Table TSGB0101)

The choice of the base year can also be a source of confusion. The two different base years (1990 and 2005) in the charts above give very different relative values for passenger transport. This is because the change in air travel between 1990 and 2005 was much greater than for road and air travel.

It would be very difficult for a reader to make mental calculations about the impact of a particular base year on the index series. However, it is important to realise that the pattern of relative changes shown in an index chart is only for the one base value it uses. Using a different base year could potentially give different patterns, especially where the series included have very different trends over time.

For more on index values see the statistical literacy guide Time series index numbers.