Posts Tagged ‘graphs’

Let’s assume you have a scale that measures a variable “hotness” on a scale of 1 to 5. According to most people’s intuition, 1 is less hot, and 5 is very hot. A shorter column in a graph means less hotness, a longer one means more hotness. It makes sense, doesn’t it?

Now, look at all your scales and all the graphs you created. Do they ALL make sense?

Sometimes, because of the way you laid out your answers in Qualtrics and because of the way Qualtrics assigns values to answers, you may end up with a reverse scale that is very confusing.

In the examples below, the scales are very confusing. In a culture that reads left to right, were things increase from left to right and from bottom to the top, the image below means that the actual difficulty was higher than the expected difficulty – but that’s not what the authors mean!

Similarly, when you look at the column graph below, you’d think the blue one indicates more difficulty, and the red, less. Alas, that’s not true…

Solution: Flip the scale!

For this measure, go into Excel, replace 5 with 1, 4 with 2, draw the graphs again, and voila! – they make sense.

In usability principles, this falls under consistency and standards – use accepted standards in your interface.

Please make sure to check your scales and graphs, make sure they make sense – in the generally accepted way in American culture.

How do you decide what is the best way to present information? When should you use a table, a bar graph, or a pie chart? It takes a bit of thinking about the nature of the information and the message you want to get across – then, Excel can do the rest.

Here is a quick resource for you that explains a bit about making these decisions (link opens pdf).

I believe tables and bar graphs will be most useful to you. So, how do you decide whether to use a table or a bar graph?

  • Tables are great for presenting individual values, but if you cram too much information into one table, it becomes overwhelming. Tables do not facilitate comparisons. In a table, the reader has to search for comparisons among data and compute them mentally, then remember them for later. This is a lot of hard work!
  • Bar graphs, on the other hand, are great for rankings and comparisons.

One thing to be careful about are the sides of bar graphs: What do the axes represent? Are they accurately titled? If you display scaled along the axes, are they correct? Check what types of units you are displaying (absolute numbers, percentages, etc.).  When you work with 5 participants, one participant represents 20%. While marking 20% on an axis is technically accurate, it is a bit misleading. When working with such small numbers, I suggest being very cautious about using percentages, if using them at all.

Let’s look below at three ways of presenting the same information: Level of agreement on a 5-point SA-SD scale, where 5 is SA, and 1 is SD. Imagine the agreement is that the task was easy to complete.

The table presents all the individual values. Look at it for 3 seconds or less and answer: Which task was easiest to accomplish?

Here is a different view of the values each participant gave. It doesn’t show the averages, but, can a quick look at the colors give you an overall idea about which task was the easiest?

(Stacked bar graphs can be very effective. We saw a great example in one presentation yesterday of a stacked bar chart showing task completion rates).

Below is a simple column graph representing only the means for each task. You lose the detailed information about each participant’s responses, but you gain even more clarity about what task was easiest to complete:

The examples in this post are all useful for the first part of the results section, where you present combined results. Things should be much easier in the second part, where you present the data for each task. See my previous post for an overview of the three parts of the results section.