Test Scores vs Kids in Poverty

Two sets of headlines in January and February of 2002 oblige NARPAC to turn back to the subject of public education, and to point out that DC is not the only jurisdiction in the Washington metro area to be struggling with the problems of educating the next generation. The Prince George's County School Board has been openly feuding with its school superintendent to the point where students are embarrassed by performance of their elders. The School Board decided to fire the superintendent, and State authorities reversed their decision and fired the School Board. And the elite Montgomery County School System is aghast to find that Maryland's standardized test scores (MSPAP's) indicate disturbingly erratic changes in student scoring which suggests that all is not well somewhere, but they cannot yet tell whether it is the tests or the kids.

In each case, it is clear that both authorities and parents continue to be convinced that their kids' education can (or should) be taken care of within the school system itself. While there surely are better schools and worse schools within any one system, there are indisputably better and worse kids, often shaped by better and worse parents and better and worse community environments in which they are marinated. One of NARPAC's very first graphic analyses five years ago lined up DC's school test scores by the level of poverty (subsidized school lunches) in each. Although there was substantial difference in individual school performance (i.e., school system opportunities), there was an underlying trend line that essentially assured that the best school in a very poor neighborhood would never match the very worst school in a very rich neighborhood (i.e., environmental opportunities). This is discussed under school system scores.

Publication of the school-by-school MSPAP scores for Montgomery County Schools makes possible a similar analysis, this time clustered around the "good" end of the test scale and the "rich" end of the wealth scale. For both Frederick and Montgomery Counties in Maryland, the same conclusion applies: The worst of the well-above-poverty student schools will do better than the well-below-poverty schools. This is shown on the scatter charts to the right combining elementary and middle schools. Virtually all data points are included between two straight lines: a minimum dropping from a score of 40 to a low of 10, and a maximum dropping from 80 to 20. This suggests a 2:1 variation between individual schools, but a 4:1 variation by poverty level. But why does poverty impact on school performance at all? After all, the parents do not pay the bills in public schools. In fact, the majority of school revenues are provided by taxpayers that do not use the public school system. One could argue that school performance should be independent of family income.

Primary Causes of Household Poverty

One basic question, then, must involve the causes of poverty in the US. There are three common measures easily derived from Census data, and household data are more relevant than individuals data because a good portion of households have kids in them (35% in the DC metro area) First is the variation in household income with family composition. Two earner households, married or not, are bound to be better off than one-earner households, and, right or wrong, lone male householders are wealthier than lone female householders, even though the female is almost twice as likely to have kids at home. In 2000, lone female householders earned just under half as much as couples.

The second common way of looking at household wealth is by basic race: once just black vs white, but now a broader contest between Asians, whites, blacks, and Hispanics. As shown on the charts to the left, Asians are now at the top of the pile averaging 19% higher household income than whites, who, in turn, bring home on average of 43% more than black or Hispanics, The details of earning power of blacks and whites is treated elsewhere. Furthermore, family households (2-adults) bring home 36% more than lone male householders, who in turn earn 57% more than lone female householders. Lone females struggle along on less than half the income of 2-adult families, although they average nearly the same number of kids per household! The importance of this factor was stressed in an earlier NARPAC analysis of the "education gap".

It is also of interest to look specifically at the composition of those below the poverty line by race. These data, reported previously in an analysis of families in poverty by race, are shown here on the right rearranged by household type. Clearly, the bulk of all poor households are those held together by a lone female, most frequently never married. Very few are hosted by a lone male. Among 2-adult households, most are white (and a few Asians), a lot are Hispanic, and only a few are black. The "Million Man March" on Washington a few years ago had a noble purpose: black males do not appear oriented towards family responsibilities. This trend is amplified by showing the number of all poor families that are headed by a lone female over the years. In the last 40 years, they have grown from less than 24% to a peak of almost 54% before dropping somewhat under the influence of the recent (pre- recession) welfare-to-work programs.

The third, and perhaps most interesting determinant of household income is by household education level. This next chart shows the relative income of those who did not complete grade school (GS) those who did not complete high school (HS), those that did complete high school but little more (HS); those than had some college time (Col 4), those that completed 4 years (College 4) and those that went further (Col 4). Those at the top as a group (including lone and dual-providers) make over $96,000; while those stuck below grade school (even in pairs) exist on an average of $26,000 well below twice the poverty level if there are kids, and therefore eligible for school lunches.

Clearly, Americans are not poorly educated because they are poor, they are poor because they were poorly educated.

Creating an "Aggregate Education Score" (Ag/Ed)

The trend over the past 30 years has been towards more and more education. As shown on the next two charts, fewer and fewer Americans are settling for less than high school (or grade school) education, though drop-out rates are still far from minimal. At the other end of the scale, those with some college training have more than tripled from 15 million to 48 million.

To track this increase in education level, NARPAC has conjured up an "aggregate education (Ag/Ed) score" that gives 1 point for 'HS-', 2 points for 'HS', 3 for 'Col4', and 4 for 'Col4+'. That combined score has risen from 2.03 to 2.56 in the past 20 years, some 26%. And, in fact, mean household income in 2000$ has risen from $41.9K to $57.1K, some 36%, while median income rose from $35.2K to $42.2K, almost 20%. Take your pick, rising education and rising income are moving together.

One can then hypothesize a set of schools within a school system, each with 1000 kids: at the high end of the spectrum the student body is all-white with none in poverty; and at the other extreme it is all black kids below the poverty line. Two factors seem particularly germane: as the level of poverty increases, the number of parents behind each 1000-kid school drops from just under 1000 to just under 600, and the school's Ag/Ed score drops substantially from 5.86 to 2.25: less parental influence, and far less parental education (upper chart, right)

Stretching this analogy further, the American Housing Survey makes it possible to calculate Ag/Ed scores for 18 very different population subgroups within each of many US metro areas, including the Washington metro area, and to compare each with a median household income for that subgroup. These subgroups include:

Household Income vs Household Education

Two things stand out on that lower chart above: a) there is a very linear relationship between Ag/Ed score and household income, and b) there is no discontinuity between the all-race data points (about 25% black at most) and the black-only data points. Out of curiosity, NARPAC repeated the calculations for both the Philadelphia and Detroit metro areas, with essentially identical results, and plotted them all together below:

If there is any differentiation between the three groups, it is that the Detroit area seems to have somewhat lower income for the same education level, regardless of race. (Such differences do exist in different parts of the US.) In each case, however, the black education levels are at the lower end of the scale. To make the future for its children more competitive, NARPAC can offer the black community only two bits of free advice: get more education and get married!

NOTE: It should be noted that it is not possible to take the next seemingly obvious analytic step to combine the trend line of the test scores vs kids in poverty and the Ag/Ed vs household income to derive a summary chart of kids' test scores v parents' AgEd. This is because neither trend line has a "proper zero". First, the MSPAP test scores are relative to some passing expectation (not zero). Second, using the fraction of kids given subsidized lunches as an indicator of poverty does not leave room for variations in wealth in schools providing either no one or everyone such assistance. And third, NARPAC's AgEd scores give no credit for grade school drop-outs' education even though they are still capable of some household income. Further direct research (if it hasn't already been done) could lead to a direct correlation between cause and effect of parents' smarts vs kids' smarts: a fundamental dependency over which the school board has no short term control.

Local Scores vs National Norms

Recent MSPAP scores were published by race for both Maryland State as a whole, and Montgomery County in particular (one of the 'elite' counties in the Washington metro area). It is also interesting because of Montgomery County's higher than average share of Asians and Hispanics and smaller share of blacks compared to Maryland (or DC for that matter), as indicated on the chart to the left. The scores show some interesting variations as well. The County has higher than the state average scores for its black and white kids, but virtually the same scores as the average for Asians and Hispanics. Furthermore, the scores for Asians and Whites are essentially twice those of blacks and Hispanics. Can this be confirmed using NARPAC's Ag/Ed scores?

The Census provides a separate data base which tracks educational achievement by race and family composition, which NARPAC converts to Ag/Eds. First, as an interesting aside, it is possible to separate the education level of households with and without kids under 18. As shown on the attached chart, families of all races with kids are somewhat better educated that those without school-age kids. NARPAC believes that this effect results primarily from the fact that parents with kids still at home are likely to be younger (and therefore better educated) than those who no longer have kids at home, and than all those couples who never had kids.

It is also possible to sort out the Ag/Ed of two-parent and one-parent households, and by the number of kids in the families of each. While NARPAC has not bothered to present these data here, parents with two kids turn out to be a little better educated, and those with four, significantly less, all else being equal. But there is a huge difference in the national education level of two-parent and single-parent families. In each case, Asians turn out the highest scores with whites close behind, and the blacks and Hispanics and considerably further behind. More important, however, the variation between one- and two-parent families is far greater than the differences between the races.

The bottom chart in this series brings this last point home. Using the demographics of the local area, the scores can be combined by which share comes from the 2-parent families (in red), and the one-parent families in blue. This can adjust the household education level scores significantly. To the extent that more Hispanic and Asian kids have both parents supporting them, they do relatively better than either the whites or blacks that do not. It also explains why Hispanic kids do as well or better than black kids in Maryland (earlier charts above): 93% of Hispanic kids come from families with two adults present while only 60% of black kids are so blessed. An almost identical pattern emerged from another Maryland State Dept of Education dealing with readiness of kids to enter kindergarten: whites, 56%; Asians, 55%; Hispanics, 39%; Blacks, 37%.

National Trends in Education Level

Census 2000 data on education also makes it possible to confirm how education levels vary with several basic parameters as shown on the chart to the left. For the entire US, the NARPAC Ag/Ed value is 2.07 (better than completing High School, but well below a college degree). The difference between men and women is now at most a 2% advantage for the males: a remarkable achievement for women over the past half century. The difference with age is smaller than expected, with only those above 65 showing a significantly lower level. Inner city education levels are also surprisingly close to those for the suburbs (metro areas excluding the inner cities).

Among the four major races, the results are similar to those shown in the Washington Metro Area. Asians have a noticeable advantage over whites who have a more significant advantage over blacks, who in turn are way ahead of Hispanics. Asian women in inner cities are somewhat less well educated than their suburban counterparts, whereas suburban Asian males are now at the top of the scale, presumably because a larger share are professionals. It will be interesting to see how these standings shift as the Asian and Hispanic second/third-generation American populations grow: will the Asians pull further ahead, and will the Hispanics catch up?

Just Too Many Kids

There was a time when the conventional wisdom held that the major reason for poverty, and particularly among the minorities, was just too many families with too many kids. This appeared to be the case in the 1960's in DC when there were close to 200,000 kids under 18 on the black side of town. But by the 1980s, this number had been drastically reduced and has somehow given rise to the myth that thousands of middle class families left DC in the '80s, even though the number of households and taxpayers stayed relatively constant. If fact, that initial baby boom matured and adopted a different life style.

Using the same American Housing Survey Census Data, NARPAC has taken a look at the number of households with kids, and the number of kids in each of those households, using the same subdivisions as used above to develop trends lines of education vs income. The results are shown in the two graphics below for various demographic subsets of the Washngton, Philadelphia, and Detroit metro areas:

The overall share of households that have kids in them is now around 37%, with blacks slightly higher at 40% and Hispanics (a majority of whom are first generation immigrants) still slightly over 50%. Perhaps more interesting, however, is the fact that there is virtually no difference between those above and below the poverty line, or in the suburbs or the inner city.

Much the same applies to the average number of kids in those households that do have kids. The norm is now between 1.8 and 1.9, with Hispanics somewhat higher at 2.3. However, in this case, the poor, including poor blacks, do have well over 2 kids per household. It should also be noted that there is significant variation between the three chosen metro areas, with Washington DC consistently somewhat more "kid-averse" than Philadelphia, while inner city blacks in Detroit are in a class by themselves, seemingly somewhat behind the changing times. But NARPAC concludes from the above that it is not an excess in offspring that causes poverty, since there is virtually no difference in family size for cohorts with substantially different means and opportunities.

In summary. according to this NARPAC analysis of Census 2000 data, there are two primary determinants of kids' school performance: how well educated their parents are, and whether one or both are helping their kids succeed, and one secondary determinant: how well the schools do their job.

NOTE: Additional analysis on this key subject is contained in NARPAC's 2005 summary and commentary on DC's latest DCPS Strategic Plan

Related Sections
• For general discussion of issues concerning public education in DC, and DCPS facilitiesplease refer back to the previous chapters.

• For more detailed information on relevant reports on DC public education and related topics please refer ahead to the next chapter.

Did you find this of interest?
Please give us your FEEDBACK

This page was updated on Jun 5, 2005