What the College Board Really Ought to Do

[tokenadult]

<p>The College Board really ought to publish a complete bivariate plot (“scatter plot”) </p>

<p>[Introduction</a> to Bivariate Data](<a href=“http://cnx.org/content/m10949/latest/]Introduction”>OpenStax) </p>

<p>[Untitled</a> Document](<a href=“http://www.psychology.emory.edu/clinical/bliwise/Tutorials/SCATTER/scatterplots/scatt.htm]Untitled”>Untitled Document) </p>

<p>of SAT scores obtained by all test-takers plotted against test-takers’ self-reported family income bands. Right now, the College Board doesn’t do much to back up visually its true statement about the modest correlation between scores and family income. </p>

<p>

@2008-2009 SAT Registration Booklet

Relationship of SAT Scores to Family Income. The correlation of SAT scores and student-reported family income for two recent years was found to be about .23 and .29, which is consistent with general research findings on the relationship of educational achievement and family income level. Although average SAT scores tend to be higher for students from higher-income families, students from every income level, as reported on the SAT Questionnaire, obtain the full range of SAT scores. Furthermore, many students from low-income families do well on the test. For seniors who graduated in 2000, one-third of the students with reported family income at or below $30,000 obtained scores above the national average.

</p>

<p>Just once College Board published a table, which another researcher turned into a line graph, of income correlations to scores stratified by other variables. Checking the graph </p>

<p>[File:1995-SAT-Income2.png</a> - Wikipedia, the free encyclopedia](<a href=“http://en.wikipedia.org/wiki/File:1995-SAT-Income2.png]File:1995-SAT-Income2.png”>File:1995-SAT-Income2.png - Wikipedia) </p>

<p>suggests that income does not constrain scores as much as some suppose, but this graph suffers the defect of not showing the full scatter of scores at each income level. </p>

<p>Inasmuch as high school grade averages also correlate with family income, perhaps through even more and more nefarious mechanisms than do SAT scores, why is the weak correlation between SAT scores and family income even a problem? It’s enough if colleges find the students in the lowest income ranges who are well able to thrive academically in college, and recruit and admit those students. Alas, that is something that colleges haven’t done systematically over the last generation. </p>

<p>[BW</a> Online | July 7, 2003 | Needed: Affirmative Action for the Poor](<a href=“Businessweek - Bloomberg”>Businessweek - Bloomberg) </p>

<p><a href=“http://www.educause.edu/ir/library/pdf/ffp0621.pdf[/url]”>http://www.educause.edu/ir/library/pdf/ffp0621.pdf&lt;/a&gt; </p>

<p><a href=“http://www.educause.edu/ir/library/pdf/ff0615S.pdf[/url]”>http://www.educause.edu/ir/library/pdf/ff0615S.pdf&lt;/a&gt; </p>

<p><a href=“http://www.tcf.org/Publications/Education/carnrose.pdf[/url]”>http://www.tcf.org/Publications/Education/carnrose.pdf&lt;/a&gt; </p>

<p><a href=“http://www.tcf.org/Publications/Education/kahlenberg-affaction.pdf[/url]”>http://www.tcf.org/Publications/Education/kahlenberg-affaction.pdf&lt;/a&gt; </p>

<p>[A</a> Thumb on the Scale | Harvard Magazine](<a href=“http://harvardmagazine.com/2005/05/a-thumb-on-the-scale.html]A”>http://harvardmagazine.com/2005/05/a-thumb-on-the-scale.html) </p>

<p>[The</a> Best Class Money Can Buy - The Atlantic (November 2005)](<a href=“http://www.theatlantic.com/doc/200511/financial-aid-leveraging/4]The”>http://www.theatlantic.com/doc/200511/financial-aid-leveraging/4) </p>

<p>[The</a> Harvard Crimson :: News :: Recruiting a New Elite](<a href=“http://www.thecrimson.com/article.aspx?ref=510012]The”>http://www.thecrimson.com/article.aspx?ref=510012) </p>

<p>[Cost</a> Remains a Key Obstacle to College Access](<a href=“http://www.equaleducation.org/commentary.asp?opedid=1240]Cost”>http://www.equaleducation.org/commentary.asp?opedid=1240) </p>

<p><a href=“http://www.jkcf.org/assets/files/0000/0084/Achievement_Trap.pdf[/url]”>http://www.jkcf.org/assets/files/0000/0084/Achievement_Trap.pdf&lt;/a&gt; </p>

<p>[Legacies</a> of Injustice: Alumni preferences threaten educational equity–and no one seems to care. - Reason Magazine](<a href=“http://www.reason.com/news/show/123910.html]Legacies”>Legacies of Injustice) </p>

<p>[Promise</a> Lost: College-Qualified Students Who Don?t Enroll in College (IHEP)](<a href=“http://www.ihep.org/publications/publications-detail.cfm?id=117]Promise”>http://www.ihep.org/publications/publications-detail.cfm?id=117) </p>

<p>[Colleges</a> reach out to poorer students - The Boston Globe](<a href=“http://www.boston.com/news/education/higher/articles/2008/11/10/colleges_reach_out_to_poorer_students?mode=PF]Colleges”>http://www.boston.com/news/education/higher/articles/2008/11/10/colleges_reach_out_to_poorer_students?mode=PF) </p>

<p><a href=“http://www.google.com/hostednews/ap/article/ALeqM5hkBGMsvJKRKaL67qxkOCaDByDJFAD94R70G02[/url]”>http://www.google.com/hostednews/ap/article/ALeqM5hkBGMsvJKRKaL67qxkOCaDByDJFAD94R70G02&lt;/a&gt;&lt;/p&gt;

[chaospaladin]

<p>Nice post.</p>

[brassring]

<p>There must be a reason the College Board does not provide data showing higher SATs are correlated nicely with income across ethnic demographics.</p>

<p>If the College Board truly had evidence that there was little or no correlation they would be touting the results loudly – to the accompaniment of jungle drums…</p>

[skrlvr]

<p>What is interesting about the graph of 1995 scores is that it really doesn’t get beyond 1100 (out of 1600).</p>

<p>What I would be curious about is the income levels of those SAT scorers at the very upper end of the scale, which seem to be the scores you want to hit for the Ivy League and comparable institutions.</p>

<p>For all those folks who feel that a 2200+ is needed for these types of schools, I wonder what the income level of the 2200+ crowd is (or for that matter the 2000+ crowd).</p>

<p>But your absolutely correct about the CB giving out more data here.</p>

[tokenadult]

<p>

What is interesting about the graph of 1995 scores is that it really doesn’t get beyond 1100

</p>

<p>What is being graphed there is either median (most likely) or mean scores in each broad income band. There isn’t ANY income band in which all the students score at 1600. </p>

<p>[SAT</a> Data Tables](<a href=“http://professionals.collegeboard.com/data-reports-research/sat/data-tables]SAT”>SAT Suite of Assessments - College Board Research) </p>

<p>Peak scores are rare. But they occur among students from a wide variety of family incomes.</p>

[Cali_Trumpet]

<p>I dont understand. Collegeboard claims the correlation is around .23 between SAT scores and family income. This correlation is practically meaningless. It basically says it is hard to predict the family background (the dependent variable) based on the SAT score, and vice versa.</p>

<p>That graph for the 1995 correlation is interesting, but is that done using averages? If so, averages show much stronger correlation than usually exist, which might be misleading.</p>

[tokenadult]

<p>Yes, a correlation of .23 is actually rather weak. </p>

<p>The graph for 1995 shows a plot of just one point for SAT score for each income band, and then a spurious line (what Tufte would call chart junk) connecting each point. The point is, I’m pretty sure, showing the median score for all students who self-reported being in that income band. What the chart should have done is shown the full scatterplot (not published by College Board) for all of the scores.</p>

[xiggi]

<p>

Inasmuch as high school grade averages also correlate with family income, perhaps through even more and more nefarious mechanisms than do SAT scores, why is the weak correlation between SAT scores and family income even a problem?

</p>

<p>What continues to amaze me is the ink spilled on this subject, and the equally moronic STUDENT-REPORTED GPA. </p>

<p>Isn’t the answer CLEARLY provided in he quoted text? Consider this, “The correlation of SAT scores and** student-reported **family income for two recent years was found to be about .23 and .29, …”</p>

<p>Yes, the data is culled from an utterly asinine survey that is forced upon millions of students who do NOT have the slightest clue about their parents’ real income, or do NOT take the survey seriously. After all, aren’t the students already annoyed at having to spend close to four hours filling in little oval bubbles.</p>

<p>Yet, building another monument to the garbage in-garbage out theory, the College Board keeps on compiling the data. What the College Board Really Ought to Do? Stop relying on self-reported data by frustrated or clueless teenagers!</p>

[tokenadult]

<p>It’s true that there is much to be desired in the data-gathering on this subject. One different way to gather the data would be to get a group of researchers, with suitable guarantees of anonymity, access to FAFSA and other financial aid filings for students so that the officially reported incomes of the families could be matched with entrance test scores. That would also allow finer resolution in the view of income. The big data gap in that instance, of course, would be the high-income families who don’t apply for any form of financial aid. </p>

<p>It’s important to note here that any such correlational data-gathering scheme proves nothing about causation. It would take a different kind of study design to figure out just what it is about a student’s life circumstances that may be most influential in helping a student obtain a high SAT score.</p>

[skrlvr]

<p>I don’t know that much about statistics, but I guess I was trying to see what the graph would look like if the SAT score was on the X axis, and the the mean or median income was on the Y axis. </p>

<p>Given that so few people score at the upper end, if it skewed significantly toward one income group, it might not affect the overall mean score of that income group.</p>

[MarathonMan88]

<p>I’m not sure that such scatter plots would help answer the questions that matter: to what extent the correlations between income and test scores are due to some feature of the tests; to the greater availability of coaching for higher income students; or to real differences in academic preparation. (I don’t think there’s much disagreement that part of the income/score correlation is a reflection of the undeniable fact that there’s unequal access to education in the U. S. based on socio-economic status).</p>

[zoosermom]

<p>

One different way to gather the data would be to get a group of researchers, with suitable guarantees of anonymity, access to FAFSA and other financial aid filings for students so that the officially reported incomes of the families could be matched with entrance test scores.

I would find that completely and utterly unacceptable. There is no reason to invade the privacy of countless families to satisfy some curiousity.</p>

[Hanna]

<p>“I would find that completely and utterly unacceptable. There is no reason to invade the privacy of countless families to satisfy some curiousity.”</p>

<p>Census data about household income by ZIP code might serve as an equally effective data set.</p>

[zoosermom]

<p>

Census data about household income by ZIP code might serve as an equally effective data set

That seems reasonable and respectful.</p>

[sorghum]

<p>High-income parents are on average more intelligent than low-income parents, and would be expected to have more intelligent children. The weak correlation observed (0.23 or whatever) would be consistent with this.</p>

<p>Of course there are millions of exceptions, and society isn’t fair, etc., etc. but this factor must account for some of the variance.</p>

[tokenadult]

<p>If there is no general interest in the issue, no one will do the research. But I think there is general interest in the issue of how much, if at all, family income influences student SAT scores, because I see discussion of that issue here on CC. Currently, colleges see lots of very private information for each individual applicant. They seem to be reasonably good about not blabbing that information to people who have no need to know. Social science researchers have a lot of ways of protecting anonymity of research subjects (those begin with dealing with numbered files, that have no names on them), and I suppose that there is a way to gather detailed information on this subject without violating individual family privacy. I wouldn’t want research on the issue to be done without strong privacy protections, as I wrote above.</p>

[zoosermom]

<p>I’m not sure that there is general interest outside of CC or academia. I’m really not.</p>

<p>I also think that there needs to be a very, very good reason before invading the privacy of citizens. I, for one, would never consent to my data being used in that way. It’s simply nobody’s business for very little benefit. Protecting privacy by researchers is well and good, but I want my privacy protected FROM researchers.</p>

[bluebayou]

<p>To paraphrase Jack Nicholson: colleges don’t want the truth. Or, they can’t afford the massive embarrassment if a real study shows an extremely strong link to income. If such link was shown, CB would be out of business due to the political uproar led by the NYT and Congress.</p>

<p>Of course, xiggi is spot-on (as usual): self-reported data is close to worthless. Even if HS kids think they know their parent’s income, do they really? Is it Gross, AGI? Does it include profit-sharing? How about all those businesses who run ‘off-the-books’? Or, with a 50% divorce rate, does kid really know the non-custodial spouses $?</p>

<p>Nevertheless, such self-reported data is Golden for CB. It shows a (logical) connection between test scores and income that passes the smell test, but it is not strong enough to make the politicians grandstand. CB would not have it any other way.</p>

<p>Better to use the mushroom theory theory…(keep 'em in the dark and feed 'em…)</p>

[brassring]

<p>Colleges know. Ivy League universities know. Public universities know. So, why don’t we know. Because we don’t seem to demand to now.</p>

<p>Bluebayou is correct: “The colleges don’t want the truth”. More precisely, they know the truth but are silent conspirators with CB in not talking about it. This should not prevent us from demanding the truth. More precisely, demanding that data be released to a group of researchers, perhaps mandated by Congress, to analyze and reveal the truth.</p>

<p>Unless there is a well-organized constituency (parents, teachers etc) demanding information, there is no pressure on Colleges and CB , beyond self-serving generalities claiming low correlation. And since, parents are never united (those whose kids get admitted are in utopia and the others simply move on), there will never be a clamor for the truth…</p>

[zoosermom]

<p>But why would anyone clamor for THIS information? It’s pretty clear that there is a connection, but, really, who cares? Or better yet, what is the reason for anyone to care outside of college folks?</p>