Saturday, May 23, 2009

High school attainment

The above results show Kentucky as 36th of the fifty states in high school completion.

The last time I did this analysis, Kentucky was 35th with 80%, and South Dakota was 20th with 83%. That's an important reminder that we don't just need to improve our attainment: we need to improve faster than other states to move into the top 20.

These results are also useful in thinking about public school graduation reports and methods. The figures above include GED recipients and non-public diplomas. To me, they suggest yet again that KDE reports of graduation rates of 79 to 84 percent from 2001 to 2006 are likely a bit high, and that other reports of rates well below 80 percent are almost certainly too low.


  1. Susan,

    You say the overall diploma or equivalent rate from Census data for 18 to 24 year olds is 82 percent. You also say that GEDs are included along with non-public school diplomas in this 82 percent figure.

    As I recall, the Kentucky CPE has shown a pretty consistent annual award of GEDs on the order of 9,000 to 10,000 for years. Those GEDs certainly can't be considered a high school success, can they?

    Furthermore, I think I heard somewhere that most of the GEDs in Kentucky go to recipients under age 24. If so, the 82 percent figure you show has a healthy contribution of GEDs in it.

    If we also have to take out the private school diploma holders from that 82 percent figure, why would you find it so surprising if the public school graduation rate, especially for NCLB purposes where only on-time diplomas are counted, would be in the low 70 percent range?

    Can you amplify on this?

  2. Take 73 percent as a simple version of the low seventies.

    The class of 2004 had 37,330 graduates. If 37,300 was 73 percent of the group that should have graduated, then the group must have had more than 51,100 students in it.

    Leaving out their ninth-grade year, which always bulges with students repeating that specific grade, I haven't seen any credible student count where there were 51,100 students in the class of 2004.

    Reasoning the same way, I can't find any reason to believe that there were:

    • 51,500 or more in the class of 2005.
    • 51,800 or more in the class of 2006.
    • 53,500 or more in the class of 2007.
    • 55,800 or more in the class of 2008.

  3. Alternate version:

    Is 4,000 students a reasonable estimate of the number attending private schools or being home-schooled in a given class?

    If it is, then to think we had a 73% public school graduation rate, I'd have to think Kentucky has:

    • 55,100 residents born roughly in 1986.
    • 55,500 residents born roughly in 1987.
    • 55,800 residents born roughly in 1988.
    • 57,500 residents born roughly in 1989.
    • 59,800 residents born roughly in 1990.

    Of course, they wouldn't all have to born in Kentucky. My own daughters were born in 1988 and 1989 and graduated in 2006 and 2008. Still, I'm looking at the KidsCount website reporting Kentucky births for 1990 of 54,061. That would require 5,700 more immigrants than emigrants, and I don't think Kentucky's population has been growing that way.

  4. Susan,

    RE: Your first comments (second comment in series).

    I've heard the most reliable data on public versus private school enrollment in Kentucky probably comes from the Decennial Census. I have also heard that the Kentucky Department of Education does not have high confidence in its non-public data.

    If you look at the year 2000 Decennial Census Summary File 3 Question 36 on school enrollment, it shows there were 102,714 males and 96,362 females enrolled in grades 5 to 8 in Kentucky public schools that year. That totals to 199,076 students, an average of 49,769 per grade. If you then allow for the transfer of about 1 to 2 thousand private school students who move to public schools in grade 9 (because their private schools don't offer high school grades), and if you allow that the overall student population in Kentucky showed a rise from the 1990 Census to the 2000 Census, implying that trend might have continued thereafter, then a count of 51,100 or so looks very reasonable.

    The big issue is those private to public school transfers in the ninth grade. Check with KDE on that. I think they know about it.

    RE: Your Second Comment (Third in the series)

    The 2000 Census showed a total of 55,163 males plus females enrolled in public or private Kindergarten in that year. That's pretty much in line with the most of the numbers you are talking about given the rather general nature of the discussion and the years involved (I don't recall seeing anyone say the graduation rate was constant at 73 percent across all those years).

    Anyway, we won't know for sure until the KDE finally gets its high quality student tracking system on line (assuming they do that properly) and generating graduation rates -- and that is years off according to the new Nonacademic Report.

  5. So, choose your sources, choose your cohort, and show the numbers you find plausible for:

    • Residents that age.
    • Nonpublic students that age.
    • Public graduates that age.

    I think public schools are being held accountable for student who don't exist. Show numbers that suggest otherwise. Not percents, but numbers that you think are respectable.

  6. Susan,

    You wrote, "So, choose your sources, choose your cohort, and show the numbers you find plausible."

    Actually, a lot of researchers have done that -- working within the limits of the quality of available data. Perhaps the most compelling work is found in "User’s Guide to Computing High School Graduation Rates," a two-volume report from NCES with numbers NCES 2006–604 and NCES 2006–605. The conclusion is that the best performing formula among all the estimation formulas is the Averaged Freshman Graduation Rate. It returns results only a point or two higher than the true graduation rates for two states that had high quality graduation data used to confirm the study.

    In 2004, the AFGR for Kentucky, as published in the "Digest of Education Statistics 2008" just happened to be that magical 73.0 figure you question.

    Until we can get better answers for Kentucky with a high quality system, which afte nearly two decades after KERA is only now just cranking up, I suppose the best efforts of the national expert team that created the NCES study will have to suffice for me.

    One more question. Some of your earlier posts raise concerns about thousands of students who are disappearing to the current Kentucky data system between grade 8 and graduation. How do you reconcile that concern with your current statement that "I think public schools are being held accountable for student who don't exist"? Somehow, it seems from your earlier statements that you are saying the schools are not being held accountable for students they should have responsibility to track.

    However, it has been well-reported elsewhere that once other states implemented high quality student tracking and commensurate high quality graduation and dropout rate reporting, that the official graduation rates dropped and the dropout rates soared, usually a lot.

    One of the earliest examples is what happened to Louisiana when it adopted a better student tracking system in the 1995-96 school year. That state's dropout rate more than tripled.

    The NCES dropout report containing this information explains: "Effective in the 1995–96 school year, Louisiana changed its dropout data collection from school-level aggregate counts reported to districts to an individual student-record system. The apparent increase in the dropout rate is partly due to the increased ability to track students." See Table 2 and its footnotes in "Dropout Rates in the United States: 1999," NCES NCES 2001–022.

    I am also confused by something else. Some of your earlier posts in this area comment on thousands of students who are being lost in the Kentucky data between 8th grade and graduation. Wouldn't that have to lower your numbers?

  7. The earlier post ("Graduation improves, mysteries continue")showed that starting from 2004 eighth grade, 2008 diplomas seemed to account for 81 percent, certificates for 1 percent, and reported dropouts for 12, leaving the 6 percent lost-or-stolen-or-strayed.

    This post's Census Bureau results include a slightly earlier cohort, with non-public and GED completions included, and 82 percent with a high school credential.

    Both posts argue against claims, including yours, for public school graduation rates really only being in the low seventies.

    You've spent hundreds of words here saying the lower rates should be believed, but avoiding the specifics.


    1. Define the group of students you think was the legitimate 100% of students, with date and source.
    2. Specify what percent of those students you think actually graduated, including the diploma count you use to get that rate.

  8. Susan,

    You said, "You've spent hundreds of words here saying the lower rates should be believed, but avoiding the specifics."

    Well, this is important, and our kids are worth a solid discussion. If we are losing kids, which even you say is a concern, we need to know the dimensions of that. I know you agree.

    As far as specifics, the results of the NCES study reported in "User’s Guide to Computing High School Graduation Rates" are as specific as you are going to get given the poor quality of data currently available. Those results supporting the AFGR held up in the two states used in the study and also seem to agree with rates coming out of California, which transitioned to a high quality student tracking system a couple of years ago. Maybe this study's conclusions won't hold up for Kentucky, but so far its findings seem credible in those states that have high quality tracking systems that can be used to check the accuracy of the AFGR.

    I do have one more question.

    Why do you think the US Census gets accurate answers to its obviously sensitive questions about who in each household has a diploma? Most of the Census data is collected by telephone interview. If your kids dropped out of school and some total stranger from Washington started asking about their educations on the phone, what might you say?

    And, how could the Census audit their data to detect errors?

  9. Please:

    1. Define the group of students you think was the legitimate 100% of students, with date and source.

    2. Specify what percent of those students you think actually graduated, including the diploma count you use to get that rate.


Updates and data on Kentucky education!