Although they are not a very robust predictor (even ACT, Inc., will say that a better predictor of college success is obtained when combining the ACT scores together with high school GPA), and they are only one piece of the admissions puzzle, the yearly numerical scores are easily analyzed, make for a good news story, and result in many diverse opinions.
In looking at the ACT test scores, you can find many correlations, such as lower average scores for children of single parents versus married or divorced parents, regional differences, and differences by race. This leads to many claims of unfairness in the design of the test, intentional or not. So the question is this: does the ACT provide a gateway to opportunity by identifying the best and the brightest to attend the "best" schools", or are they an artificial barrier to entry that can be used for political (or other) purposes by leaving the disadvantaged behind?
Clearly, the strongest correlation you will find in the ACT scores is increasing scores with increasing family income. Using data for all undergraduates from the 2008 National Postsecondary Student Aid Study (NPSAS), the distribution of ACT composite scores by family-income for dependant students attending four-year schools, as shown below, demonstrates this (note that the data is incomplete for ACT scores less than 10).
To answer this, I looked at NPSAS data for net tuition and fees (after all grants, veterans benefits, and tax benefits) for dependant students, to see if the advantages of an increased ACT score (possibly a benefit of well-off parents) correlates to a reduced cost of college due to merit-based aid. The result shows that the higher the ACT score, the greater the net cost of college (which must be covered by savings, income, or loans). That is, even though they qualify for greater merit-based aid, these students tend to attend higher-cost schools, resulting in a greater net cost. Also, since they tend to have higher-income families they would receive less need-based aid.
The net result of this is that there is a tendancy to segregate college choice by family income, where children of high-income families tend to go to higher cost schools (whether better academically or not, though they are typically better in terms of quality of the facilities and surroundings), and children of lower-income families go to lower cost schools (better academically or not). Often, this means that wealthy families send their children to private schools, and lower-income families send their children to public schools (another, "DUH!" conclusion). And in those careers where the school you graduated from matters (business and law, for example, as compared to engineering or nursing), it reinforces the inter-generational advantages for graduates in those fields.