Graphing Descriptive Ecologic Data

 

Return to Home

How to Use

References

Acknowledgements

   
  Previous 1 2 3 4 5 6 7 8 9 Next  

Graphing Descriptive Studies on Two Y-Axes Graphs

It is easy to misrepresent information simply by changing the axes on the graph.  Graphs with 2-Y axes and 1-X axis are particularly challenging. 

Here are some basic guidelines to be sure that your graphs are showing undistorted correlations.

Start the axes at zero
In your first attempt at developing a graph, you should always anchor the Y axis at zero.  Then you can modify the minimum value if doing so does not change the relationships shown by the graph. If you are not starting an axis at zero, you should explicitly state why you have chosen an alternative. Here are two common reasons for not starting a graph at zero.

  1. When the lowest values in the data are far from zero, you may choose to start at a higher value to save space. Note that this may exaggerate the size of any differences.

  2. When zero is not a biologically plausible value for the data being graphed, set the minimum values just below the lowest values in the data. For example, serum sodium levels are always far above zero. 

Set the axes at logical levels
When you are analyzing data, set the minimum and maximum values on the axes at logical levels. To determine the range of values that need to be displayed on the graph, identify the largest value to be graphed and round up to a slightly larger number. For example, if the largest integer is 778, it might be logical to round to 800 as the largest value on the axis.

As long as the axes emphasize a valid relationship, it is appropriate to modify the minimum values.  However, it is important to indicate you have done this by putting a short gap between 0 and the minimum number you chose so the reader is visually aware of how you made the graph.  Another alternative is to provide a full-set graph anchored at zero and a sub-set graph that contains a break in the axis to provide the reader with all the information used to develop the modified graph.

Establish an appropriate interval size
Select an interval size that does not crowd the data together and prevent the relationship from being visible.  Often, increments of 10, 100, 1000, etc. are used as standardized intervals.

Both Y axes must be comparable
When you use a2-Y axes graph, the range on both Y axis needs to be similar. In other words, if the scale on the left Y axis is 0-100, the scale on the right Y axis should be in an equivalent range.

Use the same scale if several graphs will be compared
To provide accurate analysis and evaluation of multiple graphs, you might set the same scale for the sake of comparison. It is important to be consistently clear what scales and legends are being used and what they mean.

A Two Y-axes Graph in Practice 
This graph was also cited by Kaplan and Wein based on reinterpretation of data presented by Foege, Millar, and Henderson. (3,4)

This series is data.series1, that is smallpox casesthis refers to the other dataset, data1.series1, which is unvaccinated

Question #4
DIRECTIONS:
 
Complete the following based upon the instructions provided within each question and provide answers based upon the relationships that appear.

Do the lines look closely correlated?


If you change the range values of the right Y axes from 0 - 100%, is the correlation still evident?


 

Because you can change the slopes of each line by changing its Y axis, you can choose axes that make the two lines look like they have the same slope. This might lead your reader to think they are correlated.

It is your responsibility to display valid correlations clearly.
Be careful when using 2-Y axes graphs.

 

   
  Previous 1 2 3 4 5 6 7 8 9 Next