Link Search Menu Expand Document
Data visualisation toolbox

4.3 Line graphs

When to use: Line charts should be used when you have a continuous dataset that changes over-time (longitudinal). Further, they can be useful to compare different ‘series’ over the same timeline, i.e., showing the different trajectories of the values of variables overtime. In sum, they are very useful to portray trends, but not exact values.

Our case study is cross-sectional, and therefore we do not have a series of data of one variable that can be compared in a line chart over time. However, imagine we had access to comparable regional data through secondary sources that provided access to improved water source (as defined above) each year across a ten-year period.

In this circumstance, we could compare the regional data with our survey results to determine any longitudinal trends through producing a chart that shows changes in access to improved water sources over time. We could then create a line of the variable ‘access to improved water source’, with the percentage value in the y-axis, and the timeframe across the x-axis.

image info

Based on the above graph, we can visually see that there has been a gradual improvement over the past decade in the percentage of households having access to improved water sources. There was a dip between 2016 and 2017, which we can correspond with significant flooding in the region between the two different data collection exercises (through our contextual knowledge).

Best practices: Ensure that the axes are clearly labelled, so the evaluator knows exactly what is being evaluated. Given that the main purpose is to show trends, it is best to avoid distractions that can take way from the line itself, such as grids, varying colours, bulky legends, etc. Finally, avoid comparing too many variables (aim for a max of 5 lines), as again, you want to avoid clutter to ensure ease of understanding.

When to avoid: Avoid using a line chart if you have a smaller dataset, as they are best for showing patterns over significant numbers of data points.