Compound bar chart - to round or not to round? (percentages)

  • MHB
  • Thread starter 3vo
  • Start date
In summary, the conversation discusses the creation of a compound bar chart that includes sub data within each grouping. The sub data is represented as a percentage out of 100% of the bar. The question arises on whether to round the sub data to a whole number or a specific number of decimal places or significant figures. The suggested approach is to leave them as they are or round them to a significant number of digits. It is also noted that rounding errors may lead to the total not adding up to 100%, but this is a normal and accepted occurrence.
  • #1
3vo
7
0
Hi guys,

I'm currently drawing a compound bar chart.
As I understand, this is type of bar chart which involves sub data within each grouping. The sub data is incorporated into each of their grouping bar as a percentage out of 100% of that bar. So basically the bar is 100% in total and the sub data is drawn into the bar, as their respective percentage of the bar. (I could of prob explained that a lot better).

Anyway... I've drawn up my chart and worked the percentages of each sub data within each grouping out of a 100%. However for some of the groups, the sub data are NOT integers.

Do I round these figures up to a whole number? or a specific number of dp or sf? What do I do if after rounding the total doesn't add up to 100%?

Or is it better to leave them as they are and try to best incorporate them into the graph?

An example of my data: Imagine a table below (dont know how to do one here)
First number is freq, 2nd is %
Monday | Tuesday |
Red 7 30.'43% 3 18.75%
Blue 10 43.'47% 7 43.75%
Black 6 26.08% 6 36.5%
Totals 23 100% 16 100%
 
Mathematics news on Phys.org
  • #2
3vo said:
Do I round these figures up to a whole number? or a specific number of dp or sf? What do I do if after rounding the total doesn't add up to 100%?

Or is it better to leave them as they are and try to best incorporate them into the graph?

Hi 3vo! :)

Yep. Leave them as they are, or round them to some significant number of digits.
Up to you basically. What you think looks best and is most informative.

But if you do round them, don't round up, but round normally.
0.5 goes up, 0.4999 goes down.

And if they don't add up to 100%, don't worry about it.
That's normal - the result of rounding errors.
This is known and accepted.
 

FAQ: Compound bar chart - to round or not to round? (percentages)

What is a compound bar chart?

A compound bar chart is a type of graph that displays multiple sets of data side by side, using bars of different colors to represent each set of data. It is commonly used to compare the relationship between two or more variables.

When is a compound bar chart used?

A compound bar chart is used when there are two or more variables that are being compared and when the data can be divided into subgroups or categories. It is especially useful when the data includes percentages, as it can easily show the relationship between the different categories.

Should percentages be rounded in a compound bar chart?

The decision to round percentages in a compound bar chart depends on the purpose of the chart and the level of precision required. If the purpose is to give a general overview of the data, rounding percentages may make the chart more visually appealing and easier to read. However, if the data analysis requires a high level of precision, it may be better to display the exact percentages without rounding.

What are the benefits of rounding percentages in a compound bar chart?

Rounding percentages in a compound bar chart can make the data easier to interpret and compare. It can also make the chart more visually appealing and less cluttered. Additionally, rounding can help to reduce any potential bias that may arise from displaying exact percentages.

What are the potential drawbacks of rounding percentages in a compound bar chart?

Rounding percentages in a compound bar chart can potentially lead to a loss of accuracy and precision in the data. It may also hide important differences between data points and make it difficult to identify small changes in the data. Additionally, rounding can introduce bias if the rounded values do not accurately reflect the true values of the data.

Back
Top