Friday, October 4, 2013

Some more on Kentucky graduation rates

See that relatively good-looking 2013 graduation rate and the relatively weaker ones from years past? This post is about why they differ and why the new one is such a great step up.

The 2013 rate comes from tracking a cohort of students from entering high school to leaving, even if they changed schools. That's possible with the new data system we put in place a few years ago.  At last, we can divide our number of graduates by the true number of kids entering high school, rather than by an estimate.   It's the smart way to show what proportion of kids graduate.

The 2011 and 2012 numbers are different.  AFGR is short for Averaged Freshman Graduation Rate, because that method averages together the number of ninth-graders four years back and the number of tenth-graders three years back, and then divides the number of graduates by that.

Scratching your head?  Good.  Here comes some explanation.

Grade 9 is always every state's largest single class, because more kids repeat that grade than any other.  Only, no matter how many times a kid is in that grade, he or she can only graduate once. So every graduation rate has to have some plan for making sure the repeaters aren't counted repeatedly.

AFGR was a way to estimate first-time freshmen when data systems were not able to track individual kids.  In its original form, AFGR was used for whole states, and it included a grade eight count from five years before the graduation count.  Serious research showed those estimates to be pretty good.  To use AFGR for individual schools, the grade eight data had to be dropped--and I have not found any research explaining why that produced close estimates.

In fact, I'm pretty sure that a two-grade AFGR consistently yields too high a number in grade nine,  When you divide by a number that's too big, you get a graduation percent that's too low.

That's why the 2011 and 2012 numbers are so much lower. Each was calculated using a sound number of graduates and dividing by too many ninth graders.

That's also why it's a mistake to compare this year to past years.  The comparison wouldn't be apples to oranges, but apples to marshmallows, because the old numbers were unhelpfully soft.  Again, AFGR was probably the best option with the old data system, but it wasn't as good as we needed or as good as the cohort method we now have.

In short, the new cohort rate is a better way to figure out who starts high school and who leaves with a diploma.  The 86.1% result is more accurate and having it come in higher than AFGR is reason to smile for a minute before we get back to work on moving the rate even higher for future years. 


Updates and data on Kentucky education!