Skip to 0 minutes and 3 secondsWhen you enter your data into a processing package, like Microsoft Word or Excel, the graph you get back may well reflect the data, but will not always be the most clear or effective way to portray your results. It looks like a graph, but it could be a lot better. There are a number of different ways that you can customise a graph to better convey the information, trends, and uncertainty in your data. In this video, we're going to show you how to go from this to this. Let's start with this title. The software has picked this up from a header in the spreadsheet. In your report, the chart title should appear as a caption below the graph with a figure number.
Skip to 0 minutes and 41 secondsSo this can be deleted. The figures on these axes have trailing zeros, which do not affect the value of the number. These should be set to show the minimum number of significant figures that still accurately convey the number. You may also want to set a single zero as the origin for both axes. Often your processing package will automatically smooth the lines in between data points. This can be problematic, as it gives a false view of what the data is doing. Each bit of line could be misinterpreted as a valid data point. With smoothing, the line can extend above or below the points, implying false results. For experimental data, the only place the readings are valid is at each data point.
Skip to 1 minute and 22 secondsIt is sometimes helpful to joint each point with a straight line. But generally, it's best not to join them at all. They are discrete values, and any sort of line could falsely indicate that the results will follow it. If you're not sure what happens in between each point, you should take more data. Most processing packages will automatically include gridlines as the default format, but they are rarely necessary. If you think that gridlines will help your reader to determine the values, you can add them, but make them faint. Any content that isn't data, such as gridlines and axes, should be visually muted, so as not to distract from the data. However, it is sometimes effective to use major and minor gridlines.
Skip to 2 minutes and 4 secondsMajor gridlines enhance the display of values, while minor gridlines supplement major gridlines. This can be particularly effective for logarithmic axes. Generally though, it's best to remove gridlines entirely and use tick marks on the axes instead. You should set the values on the axes so that the data best fits the available space and ensure that there are enough values on the axes to make the graph easy to read. Make sure that these values and the increments between them are sensible. You can enlarge the data points to make them clearer. The fewer points you have, the bigger they should be. You could also make the axes thicker and darker, and increase the font size to make the values easier to read.
Skip to 2 minutes and 44 secondsYou should always add titles and units to your axes. Make sure these units are consistent with the conventions you are using in the rest of your report. The better the information you put in these titles, the shorter and clearer the overall title for the graph can be. Adding a trendline can help the reader to see the relationships in the data. If you do add a trendline, make sure it is clear that it is a trendline and not part of the original data. And be sure to show its equation and variance. You may also want to include error bars. These will help the reader appreciate the variance and uncertainty in the data.
Skip to 3 minutes and 19 secondsThere are circumstances where you may want to show more than one set of data on your graph. For example, if you have model and experimental data for the same parameters. It's important to differentiate the lines and include a clear legend so the reader can easily see which line is which. Make sure this legend is large enough to be read easily. If you're working in black and white, try making the lines thicker, lighter, or not solid. If using colour, select clear colours which are well differentiated. This is particularly important for plots where the lines are very close together. You could also use shapes. If you do, make sure that it is easy to differentiate between them.
Skip to 3 minutes and 56 secondsIn colour, it's good to have both different colours and shapes to make it easy for the reader to work out which lines and points represent which data. Each graph is different, and there is no single best way to represent all data. But, as we have seen, there are a number of different ways that you can customise a graph to better convey the data. Next time you make a graph, think back over these steps and consider how to make sure the message in your data is clear.
Getting the most out of your graphs
When you enter your data into a processing package like Microsoft Word or Excel, the graph that you first get back may reflect your data, but may not always be the most clear or effective way to portray your results.
In this video, we’ll show you how to customise a standard graph to communicate information more effectively.
This is not a tutorial on how to create a graph - these are our tips for getting the most out of your graph to make sure your message is clear.
We cover a lot of information in this video so don’t feel you have to take it all in at once. Remember you can pause and rewind whenever you need to. We have also provided a download of the key points that we cover in the video for you to take away.
Don’t forget to check out the ‘Downloads’ and ‘See Also’ section before you move on.
© The University of Sheffield