The Performance of Repeat Test Takers on the Law School
Admission Test: 2003–2004 Through 2009–2010 Testing Years (TR 11-01)
by Laura A. Marcus, Andrea Thornton Sweeney, and Lynda M. Reese
Executive Summary
The purpose of this report is to provide an update of summary information about the
number, percentages, and performance of repeat test takers on the Law School
Admission Test (LSAT). The number and percentages of repeat test takers as well as
their LSAT performance (mean LSAT scores and mean score gains) are summarized
for the 2003–2004 through 2009–2010 testing years and compiled into a single report,
allowing trends to be tracked and monitored.
Summary information is reported first across testing years to show general trends,
and then by individual test administrations (June, September/October, December, and
February) to show finer distinctions and within-year trends. Finally, the percentages and
performance of repeat test takers are summarized by gender and race/ethnicity. The
primary results covered in this report are summarized below.
The average percentages of first-, second-, and third-time test takers over these 7
testing years were about 75%, 22%, and 4%, respectively. Within testing years, the
percentages of first-time and repeat test takers followed a cyclic pattern. On average,
the percentage of first-time test takers was about 86% in June, 80% in
September/October, and 66% in both December and February.
In 3 of the 7 testing years, there were more male than female first-time test takers.
There were more female than male second- and third-time test takers in all of the
testing years in this study.
Caucasian test takers made up the largest percentage of first-, second-, and third-time
test takers, followed by African American, Asian, Hispanic/Latino, Other, Puerto
Rican, and Native American test takers. However, the percentage of Caucasian test
takers decreased as the number of tests taken increased. The percentages of most of
the other racial/ethnic subgroups increased as the number of tests taken increased.
Across testing years, mean LSAT scores were highest for second-time test takers
(151.2), followed closely by first-time (151.1) and third-time (148.3) test takers. In the
testing years 2003–2004 through 2006–2007, first-time test takers had the highest
mean LSAT score. Second-time test takers had the highest mean LSAT score for the
last 3 testing years. Third-time test takers consistently had the lowest mean LSAT score
(of first-, second-, and third-time test takers). The same trend also held in most cases
across the male and female gender subgroups.
Test takers who repeated the LSAT gained an average of 2.8 points the second time
they took the test and 2.1 points the third time they took the test (compared to the
second time). Mean score gains for male test takers were 0.2 points higher on average
than mean score gains for female test takers (2.9 points vs. 2.7 points). Of the largest
racial/ethnic subgroups, the mean score gains in descending order were as follows:
Caucasian, 3.0 points; Asian, 2.8 points; Hispanic/Latino and Puerto Rican, 2.5 points
for each; and African American, 2.0 points.
In evaluating the results reported here, especially regarding gender and racial/ethnic
results, the reader should bear in mind that the test takers were self-selected. That is,
these test takers chose to take the LSAT themselves, possibly more than once; they
were not randomly chosen to be assessed (or reassessed). Also, test takers voluntarily
self-reported their gender and race/ethnicity. That is, individuals chose whether to
respond to these classification questions and decided how they would respond
(especially with regard to race/ethnicity). As a result, differences in LSAT performance
across gender or racial/ethnic subgroups cannot be attributed to these subgroups in
general, but merely to those who chose to take the LSAT and identify themselves as
belonging to those subgroups.
Also note that summary statistics across gender or race/ethnicity describe subgroup
differences, not individual differences. Thus, for example, a repeat test taker from one
racial/ethnic subgroup may outperform 90% of the repeat test takers from another
racial/ethnic subgroup, even though the subgroup mean differences might suggest
otherwise.