Why isn't the mode reported?

Home > Statistics Every Writer Should Know > The Stats Board > Discusssion

Why isn't the mode reported?
Message posted by Reg Aubry on October 5, 2000 at 12:00 AM (ET)

Neat site!

One thing I really really really wish reporters understood - and used - is the concept of mode. When people want to know the cost of the average house in a neighborhood, or the average salary at a company, they truly are looking for the mode: "The number or range of numbers in a set that occurs the most frequently."

Now, it turns out that using the median gives you that number most of the time. But technically that's not what people really are asking for.

I remember reading that mathematicians don't like the mode because it's messy...in that you could have more than one mode. For example, if four of the employees in your example had a salary of $15K and four had a salary of $23K, the modes are 15 and 23. Messy, yes...but exactly the kind of information a reader of news would find informative!!!! And if, as the standard reporter would do, one averaged the two, the resulting figure of $19K would be inaccurate and misleading to all concerned, inside and out of that company!

Why oh why is the mode ignored? I don't understand it. Behind practically every median or mean reported, the real question the reader thinks is being answered is the mode!

Enlighten me!

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: Why isn't the mode reported?
Message posted by SOF on October 5, 2000 at 12:00 AM (ET)

The mode is very unreliable and often meaningless. The median is much better.
To verify this fact, roll a die 20 times repeatedly and look the the mode for each set of 20 rolls as well as looking the median for each set of 20 rolls.

Re: Why isn't the mode reported?
Message posted by phil on October 7, 2000 at 12:00 AM (ET)

Very good observation. However, let me go one step further and say that what people want sometimes is to see what the shape of the distribution looks like. That shows you where the mode(s) is, the center, and the spread. I find that to be more useful in some settings.

A histogram is sometimes the closest thing you have to the distribution, so it is good, especially if you have 50-100 observations (or more) that make up the histogram.

Mathematical statisticians don't like the mode because it doesn't have nice properties like the mean does. The median is an "order statistic" which is non-parametric. It is not as easy to work with as the mean.

Your $5 contribution helps cover part the $500 annual cost of keeping this site online.