# Comparing histograms

The data I have are the ages male or female won an award (79 data points, ages, for each gender). When constructing a freq table I used to find the number of classes, $$n = 79$$ $$\frac{\log(n)}{\log(2)} +1 \approx 8$$

to find the class width for each gender

$$\frac{max - min}{8}$$

The freq tables look like this

Female
Values      Frequency   Rel Freq.
1. 21 – 28   20          25%
2. 29 – 36   31          39%
3. 37 – 44   17          22%
4. 45 – 52   4           5%
5. 53 – 60   2           3%
6. 61 – 68   3           4%
7. 69 – 76   1           1%
8. 77 – 84   1           1%

Male
Values      Frequency   Rel Freq.
1. 29 – 34   10          13%
2. 35 – 40   20          25%
3. 41 – 46   24          30%
4. 47 – 52   13          17%
5. 53 – 58   6           8%
6. 59 – 64   5           6%
7. 65 – 70   0           0%
8. 71 – 76   1           1%


The histograms end up looking like this (STATDISK), Histograms http://www.xdcclan.com/images/histograms.jpg

Using a class start with the lowest data point in each data set, the histograms appear uneven with one starting at 0 and ending at 100 and the other starting at 20 and ending at 80

The histograms need the same amount of classes, but wouldn't it be better if I did at values as,

Values      Freq
20 - 24     #
25 - 30     #
35 - 39     #
etc


to get a better histogram to compare with? or does this not matter?

Do the # of classes only have to be the same? Do the histograms both have to start at the same point (0 or 20)? Are these histograms sufficient/correct? what if I said it is mandatory to use the formulas above to find the # of classes and class width? Should I use the same class start for each in the histograms?

• You have illustrated one of the problems with histograms. But why use a histogram at all? What are you trying to do? Sep 26, 2013 at 22:50
• Unfortunately the image in your post has gone missing. Could you provide an imgur-hosted link to the correct image, so it won't rot? Oct 20, 2014 at 16:18

Welcome to CV!

Do the # of classes only have to be the same?

Not necessarily.

Do the histograms both have to start at the same point (0 or 20)?

Highly recommended, and the axes should also end at the same number, and better if they are of the same length as well.

More babbling: It depends what do you mean by "comparing." From just the two histograms I can definitely compare and contrast the skewness of the distribution. But beyond that, visually it's not easy to say anything about the frequencies at different age because the class widths are different, and what's more crucial, the x-axes are different.

A panelled histogram that can successfully enhance visual comparison should have common y and x axes like this example:

If you allow the x- and y-axes to extend to wherever they want, it's difficult to compare across graphs.

Are these histograms sufficient/correct?

By themselves they are correct, but if it's for cross-gender comparison it's insufficient.

What if I said it is mandatory to use the formulas above to find the # of classes and class width?

I'd have a hard time not laughing. For many reasons:

1. The formula was probably aimed for making one histogram, and not for making two or more then put together for comparison. It sounds to me like a shoe-smith always wants to use up the whole piece of leather in making a shoe, and he can never sell any pair cause no any two of them actually make a pair.

2. Histogram has more parameters than just sample size. The min and the max, the distribution of the people, the nuances one would like to highlight, etc. It's unfair to overlook all other important information.

3. When it comes to visualization of data, it's important to realize that we are to communicate the data, not package it and force them down the throat of the viewers'. I'd probably stay away from all these pragmatic "you have to do this and do that..." bossy instructions. (Though I give those instructions as well, guilty as charged.)

But(!), if your professor/supervisor--who can decide if you will pass/fail the course--said you have to use this formula, then I'll suggest pick your battle wisely.

Should I use the same class start for each in the histograms?

Not necessarily, but recommended because this form is easiest to perceive. There are histogram that uses unequal bin sizes. If for some reason a certain age categorization is more important in females than males, you may make them unequal.

For instance, if some country's legal drinking age is 18, and another is 21, you may group all <18 into one column for the first country, and <21 into another column for the second country. The question is that, would this be more meaningful? Or would you see the same pattern if you bin them at single year age?

In a nut shell, you'll need to know what you want the viewers to know, and work backward. Avoid starting with recipes.

1. Your histograms need to not just have the same number of bins, they need to have the same bins. Set your histograms relaive to the max and min of your whole dataset, not just relative to each gender.

2. Designing a good histogram is more art than science. The heuristic you used to determine the number of classes is fine as a rule of thumb, but really you should play with the number of classes a bit to determine which bin width makes the most sense. Start with wide bins and then increasingly narrow your bins until you can tell that you're bins are too high resolution. Backtrack to an appropriate bin width.

3. To facilitate visual comparison, you should plot the two histograms in the same chart. You have lots of options here. I think your best bet is a violin plot, i.e. a "back to back" histogram, e.g.

If you're not a fan of this strategy, here are some alternatives:

• Treat each histogram as a class in a 2-class vertical bar chart
• Plot the height of each histogram as a point in a scatterplot
• Change the color and drop the opacity of one histogram and overlay it on top of the other.
• Subtract the values in one histogram from the other and plot the "delta" bars as positive or negative deviations from 0, such that positive could be "more male" and negative could be "more female

It would be useful to line up the bar widths for each gender if you wish to compare the two. What I would do is make a side by side histogram (or boxplot or stem and leaf plot). This way, you're pretty much guaranteed that the bars will line up, regardless of the statistical program you are using.