students underreporting, matching and overreporting
were computed. Furthermore, the percentage of students
whose self-reported HSGPA was within +/– three grades
of their school-reported HSGPA was also computed.
Among all students, the correlation between self-
reported and school-reported HSGPA was 0.74. With
regard to gender, the correlation was slightly lower
for females (r = 0.73) than for males (r = 0.75). For
race/ethnicity, these correlations ranged from 0.65 for
Asian students to 0.76 for white students. For parental
education level, there was a correlation of 0.73 between
self-reported and school-reported HSGPA for students
whose parents have earned less than a bachelor’s degree,
while there was a correlation of 0.75 for students whose
parents have earned higher than a bachelor’s degree. For
parental income level, the correlation for students in the
lowest income bracket was 0.70, while the correlation for
students in the highest income bracket ($100,000+) was
0.77. The correlations by SAT score band show a large
variation with a correlation of 0.62 for the lowest score
band (600–1200) and a correlation of 0.71 for the highest
score band (1810–2400).
Next, the percentage of students whose self-reported
HSGPA exactly matched their school-reported HSGPA
was computed along with the percentage of students
who underreported their HSGPA by three grade points
and those who overreported their HSGPA by three
grade points (see Table 4). If one thinks of the 11-point
HSGPA scale ranging from 0.00 to 4.00, a student
reporting to have a B+/3.33 when their college also
reported their HSGPA as a B+/3.33 is considered to be
an exact match. A student reporting to have a B/3.00 but
who overreported by one grade point from the school
would have a B–/2.67 school-reported HSGPA. If the
same student had underreported by two grade points, he
or she would have an A–/3.67 school-reported HSGPA.
By summing the percentage of students underreporting
and overreporting within plus or minus three grade
points, one can determine a more comprehensive are
HSGPA match (e.g., all grades from A+ through B+
considered to be matching HSGPA).
For the total sample, 52 percent of students reported
an HSGPA that precisely matched their school-reported
HSGPA. Of the remaining 48 percent, 29 percent
underreported their HSGPA and 19 percent overreported
their HSGPA. Figure 2 depicts the distribution of exact
matches, underreporting and overreporting for the
sample. The percentage of students with a self-reported
HSGPA that was within one full grade (e.g., range of
B+ through B–) of the school-reported HSGPA was
89 percent (22 percent + 52 percent + 15 percent).
These findings suggest that when inaccurate reporting
does occur, the inaccuracy tends to be minimal in
magnitude. That is, for students who do not report an
exact match, they are likely to be off by only one grade
point, which translates to a 0.33 on the 0.00 to 4.00
scale. This is in keeping with the mean difference found
between self-reported HSGPA and school-reported
HSGPA (–0.04). The same analyses were conducted by
gender, race/ethnicity, parental education and parental
income subgroups and are also presented in Table 4.
With regard to gender, females more accurately
reported their HSGPA, with 54 percent of females having
an exact match compared to males with a 50 percent
match rate. Interestingly, males and females overreported
at relatively the same rate (20 percent and 19 percent,
respectively) as well as underreported at relatively the
same rate (29 percent and 28 percent, respectively). Using
the more liberal criteria of matching within one grade
level, the match rate increased to 89 percent for males
and 91 percent for females. For race/ethnicity, African
American students had the lowest exact match rate (42
percent), whereas Asian/Asian American/Pacific Islander
students had the highest exact match rate (55 percent).
These percentages increased to 85 percent and 90 percent,
respectively, when self-reported and school-reported
HSGPAs were matched within one grade level of each
other. As for parental education and income, the match
rate tended to increase with education and income level.
Finally, the largest discrepancies of match rates occurred
in SAT score band groups. For students in the lowest
SAT score band, their exact match rate was 30 percent
(76 percent matched within +/– one grade) compared to
64 percent for students in the highest SAT score band
(94 percent matched within +/– one grade). That is,
higher-ability students were more than twice as likely
to accurately report their HSGPA. These results led to
an investigation of the match between self-reported and
school-reported HSGPA by students’ self-reported HSGPA
(see Table 5). The darkest shaded diagonal in Table 5
shows the percentage of HSGPAs that exactly match.
The lighter shades indicate the percentage overreporting
and underreporting within one grade by self-reported
HSGPA, with the lower row indicating overreporting and
the higher row indicating underreporting by the student.
The percentage of self-reported HSGPAs that exactly
matched the school-reported HSGPAs steadily decreased
as the school-reported HSGPA decreased, ranging from
a 78 percent match for an A to a 10 percent match for a
C.
3
The number of students reporting to have an HSGPA
equivalent to an A is 13,658 and decreases down to 48
for students reporting to have an HSGPA equivalent to
a C–. The percentage of students overreporting their
HSGPA steadily decreased as the HSGPA decreased, and
the percentage of students underreporting their HSGPA
increased as the HSGPA decreased.
4
3. Self-reported grades with less than 15 students in the group were not reported.