I have a rule of thumb that if a graphic doesn't have 1 piece of information per square centimetre, then it isn't worth the ink. In your case, the graphic fails because it presents only three pieces of information. This is better put in a simple sentence.

Hello Shakil,

The waffle plot may be used for sample sizes of most any size; "huge" sample sizes simply mean that you are scaling the meaning of a single icon to represent more than one case.

However, Ronán Michael Conroy 's comment is germane here. Following his and Edward Tufte's advice (see the marvelous text, The Visual Display of Quantitative Information), you have to wonder if a graphic is indeed needed for just 3 data points (Extraction = 397/411 = 96.6% of cases, Operculectomy = 11 cases or 2.7%, and Medication = 3 cases or 0.7%). In print journals, space is at a premium, so use figures only when they help tell a story that the simple numbers (as given above) can not convey as meaningfully. If you're presenting in a "limitless" format, such as an internal report or an online journal, then the space issue isn't so pressing.

Good luck with your work.

Good point, David Morse – and the other problem with the plot is that the very large discrepancy in size between Medication, Operculectomy and the vast number of Extractions is that it's difficult for the reader to appreciate the percentages involved.

I always suspect that when a graphical display adds numbers and percentages, it is because the author realises that the reader isn't going to be able to extract them by eye. All the information that the author wants to communicate is in the graph legend, not the graph.

A waffle plot presents one-dimensional information in two dimensions, so the reader has to integrate the vertical and horizontal dimensions to arrive at an area to compare. On first principles, a one-dimensional variable requires only one dimension.

Use the GraphPad Prism Software. It will draw interesting graphs.

Thanks Dr. David Morse and Ronán Michael Conroy for your comments. I appreciate your support. I liked the rule of thumb of Dr. Ronán Michael Conroy

Abolfazl Ghoodjani I hope you are not suggesting that a pie chart is an interesting graph!

In fact, with such small groups, a pie chart would be unreadable.

Ronán Michael Conroy that's why I was trying to use a waffle plot. Because of the small values of the other 2 factors.

Can you suggest, what type of graph should I use in this case?

Shakil Ahmed You don't need a plot of any sort. I cannot see why a plot is better than telling the reader that they were almost all extractions, and giving the percentages for the other two groups.

Graphs have to be used when there is a pattern to communicate that can best be done visually. This isn't such a case. How many people would have gone to see this movie?

Dear Professor @Ronán Michael Conroy

My goal was to familiarize users with great GraphPad software.

And draw hundreds of graph examples available in this software.

Perhaps we should think, others have the understanding to teach others.

Not to know ourselves better than others.