Lately, my wife and I have been staring slack-jawed at elementary school options, little ropes of drool hanging zombie-like from the corners of our mouths  – and so we’ve decided to cede our choice to the numbers.

But when you peel back the data, things like high test scores mean next to nothing about school quality – isn’t it likely that socioeconomics and not the school itself created these high test scores? My wife and I want education causation and not just correlation – a school that creates more education than should be predicted by our (reasonable) genetics and (low) income.

Luckily the layer you uncover when you poke the data is, in fact, rich with meaning. Unlike test scores alone, here are three numbers that actually DO mean something about school quality:

1. Test Scores divided by Parent Education

Thanks to friend and mommy/food blogger @yanthomas for pointing this out: by comparing test scores to years of parent education, you can see which schools’ scores outpace expectations.

For example the California DOE captured this data showing that 34 percent of parents at San Antonio Elementary in Ojai, CA, (where we used to live) attended graduate school. And GreatSchools.com reports that the average of the school’s 2^nd grade Language Arts and Math proficiency on the California Standards Tests was 53 percent.

So that’s 53/34=1.56.

Next door at Topa Topa Elementary, 26 percent of parents attended grad school and 62 percent of 2nd graders were proficient at Language Arts and Math – that’s 62/26=2.38. Not to diss San Antonio, but wouldn’t you rather send your child to the school whose test scores are higher than parent education predicts?

Unfortunately (or fortunately) we moved to Boulder, CO, which doesn’t gather parent education data. But we got a similar result by comparing test scores to the percentage of students qualifying for free and reduced lunch. It’s all about achievement outpacing socioeconomic expectations. In fact, this statistic was even a bit more persuasive for us, as it tends to weight economic and other diversity – and all else equal we’re willing to give a little in test scores to get a little diversity (and a little diversity is about all we’re likely to get in Boulder…).

2. Test Score Growth

Sure, you can blame any single test score on demographics and not on the school – but if second-graders are in the 42nd percentile while fifth graders are in the 65th, you’d have to chalk the growth up to the school. In other words, kids across grades should score in roughly the same percentiles on standardized tests. Learning at a good school will outpace learning at a bad school and so at a good school, scores should grow across grade levels. (A bad school will drop percentiles across grades.)

For example, take a look at San Antonio above. Sure, their 2nd grade competency is 53 percent, but their 6th grade competency is 69 percent. The school is doing something to take kids from middling achievers at matriculation to high achievers at graduation. Bravo.

3. Teacher Salary times Percent Highly Qualified divided by Average Age

Okay, so you want to guess teacher quality? You can’t just look at experience: some experienced teachers are burned out while some first-year teachers are energized and highly prepared. Likewise, it’s good to have “highly qualified” teachers (each state has its own definition), but what if the reason a school’s teachers are “highly qualified” is that they’re all 72 years old and burned out? Finally, it’s tempting to look at salary: if there’s any absolute in labor economics, it’s that higher pay attracts better candidates. But again, and especially in teaching, high pay may be due to seniority and not to performance.

What you want is these three factors taken together: teachers who are highly qualified and highly paid, early in their careers. This quick advancement predicts teacher quality. Again, this isn’t saying that older or younger teachers are better or worse, just that a good, motivated teacher will become “highly qualified” quickly, and a school that pays younger employees more will attract better candidates. (You don’t want teachers who rest on their low-paid laurels with naught but a four-year teaching certification framed on the wall.)

Now armed with the statistical tools of success, good luck and happy data-riffic school hunting!

* A Somewhat Extended Disclaimer:

My wife spent six years as a public school social studies and language arts teacher, and each of those six years I made the chicken-wire frame of a 7-foot tall Egyptian mummy that students then papier mached and painted with hieroglyphics. In fact, before interviewing Nobel winners to write books like Brain Trust, I got my writing start by typing up my wife’s innovative curriculum. In my opinion, the creativity, excitement for learning, and ability to work as a team that students carried forward from her class were far more important than knowing which direction to read hieroglyphics based on the way the symbols face. In other words, I’m very aware that the most important part of a classroom has little to do with test scores and the capitol of South Dakota. (Still, as much as I like to think I’m a gluten-free Waldorf hippie, I’m just not – I want a mix and numbers are an important part.)

Also, there are elements besides achievement and teacher quality that I hoped to define with numbers, such as school culture and extra-curriculars. But I either couldn’t find consistent data or data was only available on a state or district level – not much use when picking a school within a district.