« Worst statistical graphic nominated | Main | A good question deserves good data »



The problem with teacher pay is that in most countries it is virtually all public, and the public systems act virtually as a cartel, and a cartel that is quite prepared to produce a poor product if it keeps the costs low. Probably the best comparison would be to find private sector employees with comparable level of skill and compare their salaries.


These are the countries that are members of the OECD.


There are good reasons to look at starting pay, as that removes demographic effects (older vs. younger populations of teachers) and focuses on the question - who is being attracted into the field?

But if that's the question, not accounting for cost of living is a fail. Rather than going into GDP/PPP comparisons, Ken's suggestion makes more sense: look at ratio of teacher salary to alternatives for college graduates.

Finally there's no way average starting salary for a teacher in the US is 42K either. Average overall salary, maybe. Here in very well paid Montgomery County MD teachers start at 46K. If you include all benefits 42K is within reason, but not salary. Is there some adjustment already in the OECD data?

Rick Wicklin

From a statistical point of view, this plot is lacking an important feature: confidence limits (or "error bars" as they are known in some fields). Now, to be clear, I don't recommend putting confidence limits on a poster that is intended for general consumption! However, the fact that the nations are RANKED by starting salary implicitly suggests that Korea is "better" than Germany which is "better" than Ireland, and so forth.

Replacing the bar chart with a dot plot that includes "error bars" for the first and third quartiles would help convey the message that teachers in these nations have a range of starting salaries, and would indicate how wide the range is. Presumably, Luxembourg has a narrow range whereas other countries (USA?) have larger variation.

For an idea of how the dot plot might look (using different data), see http://blogs.sas.com/iml/index.php?/archives/108-Ranking-with-Confidence-Part-1.html


I think you are being unfair to the original graph source. You can always come up with many more questions than what is shown on the plot. I'm curious, could you come up with a graph that answered all the points you brought up?

I can count 6 to 8 different points you question. Let's see someone make a plot that answers all of that?!


Good god, you have done our eyes a service by not including the full original graphic from Soshable! Precisely why STEMs Need Blooms:



To be honest I feel Ortiz will stop Diaz, though it'll be late, I just think the age of Diaz (and the poor quality he's been facing recently) and the speed and power or Ortiz will see him stopped. It just is smart that you would want to pay only for good information. The first reason for people using the online sites for football betting is that it is so convenient to place bets in this manner.

The comments to this entry are closed.


Link to Principal Analytics Prep

See our curriculum, instructors. Apply.
Kaiser Fung. Business analytics and data visualization expert. Author and Speaker.
Visit my website. Follow my Twitter. See my articles at Daily Beast, 538, HBR.

See my Youtube and Flickr.

Book Blog

Link to junkcharts

Graphics design by Amanda Lee

The Read

Keep in Touch

follow me on Twitter