Ponitz graduation 2012
Cars and Parts Swap Meet
2012 senior class photos
Dayton Daily News education reporter Scott Elliott writes about schools, kids, teaching and learning.
Home > Blogs > Get on the Bus > Archives > 2006 > July > 16 > Entry
How much does income affect test results?
Household income is a factor in how kids score on standardized tests. I don’t think there’s much debate about that. The question is how big a factor is family wealth?
Many studies have shown family income is one of the most powerful predictors of student performance on standardized tests. Here’s one example of such a study that I was able to find quickly.
Recently, I wrote a story about the a study that showed Dayton was hit harder in the 2003 recession than other Ohio cities. For the story, I obtained tax return data, broken down by school district, that showed median income.
I thought it might be interesting to line up each school district’s median income with standardized test performance, based on each district’s “performance index” score.” The state-generated score reflects test performance across all tests taken.
I put the data together here so we could test the assumption that family income is a primary driver of student performance on standardized tests for the 60 school districts in the Miami Valley. If that assumption is true, the top 10 and bottom 10 districts for income locally should roughly match up with the 10 best and 10 worst scoring districts.
The numbers appear to back the theory that family income is highly correlated with test performance. Let’s look first at the top 10 districts by income, in order, and their test performance rank:
Springboro 8
Mason 2
Oakwood 1
Sugarcreek 6
Beavercreek 12
Centerville 5
Little Miami 18
Kings 4
Bethel 35
Lebanon 13
Six of the top 10 districts for income also rank in the top ten for test performance. That’s a pretty strong correlation. Now let’s look at the bottom 10 districts for income out of the 60 Miami Valley districts, with the lowest income at the top of this list and No. 51 at the bottom. The performance index rankings here would all be expected to be greater than 50:
Northride 57
Dayton 59
Mississinawa Valley 45
Trotwood 58
Piqua 50
Greenville 43
Mad River 44
Ansonia 46
Bradford 54
Sidney 40
The correlation, interestingly, is not as strong at the bottom as it is at the top, but it’s still pretty strong. Four of the bottom 10 districts for income also are in the bottom 10 for test performance.
So, generally, the Miami Valley data tends to show that median income rank is likely to correlate pretty strongly with standardized test performance in the school district. Following that rationale and using this data, we can probably determine the most underachieving and overachieving local school districts.
Follow the link to see those lists. It’s interesting to see what the school districts on each list have in common.
Permalink | Comments (8) | Categories: Testing
Comments
By Mary
July 18, 2006 8:39 PM | Link to this
Professional what? Old prof of what? Scientist of what? As an aero engineering and systems management major, I have been exposed to standard deviations, statistics, etc a time or two. We used quite a bit in predicting crack growth rates in critical aircraft structure to see how often inspections needed to be scheduled or when a crack would likely reach a critical size that could cause catastrophic failure. Am I current, no. I am under the impression you are an old prof from an educational college background, and therefore you would have some professional interest in the pay of teachers. Perhaps you were a teacher at one time.By Oldprof
July 18, 2006 12:02 AM | Link to this
Mary, we know you’re not a scientist, so please allow the professionals to discuss mathematical significance. In the meantime, note that I’m an old prof and that the rate of pay of public school teachers matters nothing to me—so your accusation that I’m greedy rings about as hollow as your incoherent dismissal of statistical significance. By the way—do you notice that the outliers on the data are (as Scott saw) almost all small rural schools districts where the students have—let’s be cognizant—fewer distractions, and where a lucky hire or two will have more impact? Let me know if I need to provide more explanation.By Mary
July 17, 2006 10:59 PM | Link to this
As one who seems to always be in the 25% or less fringe group, I object to a 65% correlation meaning very much except a 65% correlation. It still does not prove causes or what is right or wrong. Actually, I have read most drugs are marketed when they are known to work for only 35% of the patients.By Oldprof
July 17, 2006 10:31 PM | Link to this
In most sciences, 65% correlation is considered fairly strong. If 65% of patients respond positively to a particular treatment, it’s promoted as promising unless there’s another better one. But my data set was Scott’s, copied and pasted into Excel and duly correlated using their analytical function—and 78% “eyeballs” well from that data, for me. How’d you do yours?By Mike
July 17, 2006 3:24 PM | Link to this
Oldprof, maybe you used a different data set, but using Scott’s data I got a 65% correlation and .43 R-squared value. I wouldn’t consider that very strong. Also, I think the normal arguments about money are in regard to per-pupil expenditures not median incomes of the residents. Scott, could you link or cite your data sources for us?By Mary
July 17, 2006 1:06 PM | Link to this
Old prof, you are the one who sounds a little greedy. I suppose you think money also buys health and happiness. Some merchants go out of business, not necessarily for lack of revenue, but because their businesses are poorly managed. Schools can also be poorly managed and greedy.By Oldprof
July 17, 2006 10:01 AM | Link to this
Sure, there are some outliers in every correlation study, but let’s look at the real numbers. I ran them; the data correlates at 78%. Now, we aren’t seeing the third part of this equation—is the school district full of community-minded citizens who don’t mind paying a fair tax to support their schools, or are they full of self-serving greed-merchants who don’t want to pay for anything? Given that data, I’d expect an even higher correlation, but even so, the data doesn’t allow for much quibbling except among those who want to ignore the facts: income correlates to family educational achievement. Now, can we agree on that SO obvious point, supported by a STRONG correlation, and recognize that those who say “money is not important” are delusional? Then those of us who understand how to base decisions on real data can get on to the real issue: why don’t policy-makers understand this?By Mary
July 17, 2006 7:09 AM | Link to this
I think this shows that while there are some correlations to income and standardized test scores, income by itself does not “cause” higher test scores. Some small rural schools do better in score ranking than their income ranking and could reflect a value system and school size. Some of the suburban schools also achieve better test ranking than income ranking. As you pointed out Dayton schools also receive a slightly better ranking. I think a lot of studies show income of the community or money invested in the schools is not the real cause of achievment in education. The community cultural values and passion of the education professionals also play key roles.