Tuesday, June 9, 2009

The graduation pie bake-off (updated)

Public and nonpublic graduates and nongraduates, when combined, should add up to a credible estimate of that year's eighteen-year-olds. Starting today, I plan to use pie charts to push that point. On finding fresh counts or estimates of public graduates and nongraduates, I'll combine them in a single graph with total eighteen-year-olds and private-school and home-school figures. All numbers will be rounded just enough for easier consumption.

My first pie comes from Education Week's 2009 Dropout Counts report, being released this morning, which estimates Kentucky will have 41,556 public school graduates and 16,193 public school nongraduates in 2008-09. Here's the resulting pie:

That tiny nonpublic piece isn't one I'd want to serve to company. The last estimate I saw had close to 3,800 private school graduates, and I saw no sign that home-schoolers or nongraduates were included. I think EdWeek included far too many public nongraduates in its recipe.

EdWeek's national estimates offers 2,889,430 graduates and 1,286,915 nongraduates, which I'm trying to match to a Census count of roughly 4,344,000 reaching age 18 at the right time. Again, the nonpublic slice seems highly unlikely: the most current private-only, graduates-only number I can find is over 300,000.

Finally, here's my own 2007-08 pie, using ingredients from the Department's most recent nonacademic briefing. This one leaves a credible helping for the nonpublic sector.

Why am I spending all this time in a hot kitchen? I'm doing it because I think most approaches to public school graduation rates make the same big mistake. Specifically, they assume that many students who spend two years in ninth-grade can across the stage with their first freshman class and with their second freshman class, collecting diplomas both times.

Those approaches either result in implausibly low nonpublic school numbers or implausibly high numbers old enough to graduate. The pies are my way of pointing out which numbers come out of the oven making sense, and which ones seem, even on brief inspection, to be half-baked.

Update: below is the pie I originally included above. It uses the EdWeek 2009 graduation estimate and the Census Bureau's estimate of 2008 17-year-olds. I regret the error. The pie now shown above is the one I should have displayed originally. I apologize for the error.


  1. This post is fascinating, Susan, and underscores how critical is to use commonsense before swallowing the data whole. I particularly appreciated your pie puns. They really take the cake.

  2. Susan,

    Your last pie chart is problematic.

    The KDE's Excel spreadsheet for graduations through 2008 only lists 41,258 total diplomas plus certificates awarded, so it looks like your "pie" has almost 1,000 too many students in this sector. That error ripples into the other sectors.

    Also, you say in the main post that the private sector only graduates around 3,800 students. However, your "pie" includes a total of 5,000 private sector students plus dropouts. That implies the graduation rate for the private sector is only 76 percent, about the same as the public school rate. That certainly looks dubious, though there isn't any data to support much of anything about private schools and homeschools in Kentucky, including your 5,000 figure, so who knows what is really happening? Census can't help with that.

    There are other problems with using Census data for the sorts of analysis you attempt. For example, do you know what the standard error is for the 18-year old population in Kentucky? I have not looked, but the error band might make your analysis pretty fuzzy.

  3. Anon,

    Please feel free to share your own recipe for Kentucky graduation pie. That is, please share:

    • how many students you think would count as 100% graduation.

    • how many you think graduated from public school.

    • how many you think failed to graduate from public school.

    • how many you think were in nonpublic settings, including home-schools as well as private schools, either with or without a breakdown of how many you think graduated.

    It's your choice what year to address and what sources to use, so long you're clear about the size of the total pie, and the relevant slices.


Updates and data on Kentucky education!