C) SEGÚN LA EXTENSIÓN
XII. TRAUMATISMOS OSTEOMUSCULARES
4. FRACTURAS Y FISURAS
cording to Student’s Age
The results from the previous section suggest that having at least one year of preschool
reduces the likelihood of grade retention in 1st and 2nd grade, but it would be interesting
to observe longer term impacts of the new compulsory education laws. As explained in
the data section, I use a second source of data, the Mexican Employment Survey (ENOE),
to measure another academic outcome: the number of grades completed relative to the
student’s age. As with the MxFLS, the use of these data is convenient because I observe
student’s birth date, and in consequence I can identify cohorts who were and were not
affected by the reform. In contrast with the MxFLS, information about preschool atten-
dance is included in this survey, but the first round was collected when the first phase of
the reform was already in place. This is the reason why preschool attendance is taken
from a third different database.
The relevant data from the ENOE is for children who must have completed one or more
years of education. The ENOE is collected every three months since the first trimester of
2005, with 20% of the sample being replaced every trimester, such that an individual can be
observed up to 1 year and three months. I use only individuals in the their last interview
years of education in the first round of the ENOE (which corresponds to the academic
year 2006) there is at least one year of data pre-reform, and data availability pre- and
post-reform varies with the education level.
For example, a student with one year of education in the first round of the survey must
have completed first grade in or before 2005, and so there is (only) one year of data pre-
reform. The same type of student in subsequent rounds (i.e. with one year of education)
can be observed for a very long time after the reform, in particular 10 years4. In a similar
way, a student with two years of education in the first round of the ENOE must have
completed second grade in or before 2005, and so there are two years of data pre-reform
and 9 years of data post-reform. For those with three years of education in the first round
of the survey, there are 3 years of data pre-reform and 8 years of data post-reform, while
for those with four years of education in the first round there are 4 years of data pre-
reform and 7 years of data post-reform. The last cohort analyzed here are students with
nine years of education, for whom there are 9 years of data pre-reform and 2 years of data
post-reform.
As with grade retention, I begin by visually examining the reduced form RD relation-
ship for number of grades completed relative to the student’s age. In each panel of Figure
5 I divide students according to the years of education they should have based on their
birth date, the academic year when they should be enrolled in first grade, and the exact
date of the interview. Then I plot the years of education they have completed relative to
the years of education they should have. Panel A, for example, plots the average years
of education among students who should have one year of schooling, while panel B plots
the average years of education among students who should have two years of schooling.
In all panels individuals born in the last quarter are excluded as they usually have more
years of education as a consequence of school entry laws.
4
The visual evidence of Figure 5 is quite noisy for those panels with either only a few
years of data pre-reform (Panels A and B) and only a few years of data post-reform (Panel
I). The rest of the panels show some evidence of an increase in years of education, although
such increase is not consistent across all levels of schooling. While there seems to be a
discontinuous jump for three and seven years of education, no significant jump can be
observed in the other panels.
Table 4 reports the estimated effect of having at least one year of preschool on grades
completed using being born after September 1, 1998 as the instrument. Again, my analysis
will focus on a 2-year bandwidth around the birth date cutoff. The results in Column (3)
show a plausibly longer term impact of the preschool policy across all years of education,
with individuals affected by the reform having 0.86 more years of education by the time
they should have completed elementary school (6 years of education) and 0.60 more years
of education by the time they should have completed middle school (9 years of education).
Except for what is observed at 5 years of education, the estimated coefficients suggest
that the effect of the policy is stronger the more years of schooling students accumulate
between 2nd and 4th grade, and starts fading out afterwards at a very slow rate, such that
a strong positive effect can still be observed at the end of 9th grade. Although there are
no other educational variables that can be explored in the Mexican Employment Survey,
the results reported in Table 4 are very suggestive of a longer term effect of mandatory
preschool on student achievement.
1.5.5 The Effect of Mandatory Preschool on Cognitive Outcomes
The results from Sections 5.2 and 5.3 above, although derived from two different data
sets, are two sides of the same coin: if students are effectively less likely to be retained as
a result of being forced to attend preschool, then you should expect them to accumulate
results. Information available in the MxFLS allows me to analyze whether students are
less likely to be retained in 1st and 2nd grade because attending preschool fosters their
cognitive development. Although Table 3 suggests there are no effects of preschool on
retention in 3rd and 4th grade, my analysis still includes students in these grades given
the longer term impacts observed in Table 4.
To measure the level of children’s cognitive ability I use the responses to Raven’s
Progressive Matrices that are part of the MxFLS. As described in the data section, the
proportion of correct answers is normed such that it has mean zero and standard deviation
one pre-reform. The analysis is conducted in two ways, using a within-age comparison
at the time students are supposed to be enrolled in 1st to 4th grade (6 to 9 years old),
and using a within-grade comparison for each grade between 1st and 4th grade. Neither
within-age nor within-grade comparisons are preferred; the former is just indicative of
whether potential differences in cognitive ability arise at a particular age, and the latter
identifies whether those potential differences arise while attending a particular grade.
Figure 6 plots the reduced form RD relationship for student’s test score by age, and
Figure 7 plots this relationship by grade. One particular concern in this case is the sample
size at each individual age/grade, and so Panel A in both figures uses the pooled sample
to increase precision5. If, as often hypothesized, preschool attendance positively impacts
children’s school readiness, test scores should increase discontinuously at the birth date
cutoff. In any case such discontinuous increase is actually observed, i.e. having completed
the grade previous to 1st grade does not have an effect on cognitive ability as measured
by the number of correct answers in the Raven’s Matrices.
The results from Table 5, which report coefficients on the effect of completing the
grade previous to 1st grade using being born after September 1, 1998 as the instrument,
confirm that there is no effect of preschool attendance on student’s cognitive ability. This
5
subsection suggests in consequence that preschool attendance can reduce grade reten-
tion and improve grade progression without affecting children’s cognitive ability. Such
finding is consistent with a previous study where the effect of preschool attendance on
cognitive skills disappears by the time children enter 1st grade (Magnuson, Ruhm, and
Waldfogel, 2007), as well as studies suggesting that progression through school may in-
volve more display of non-cognitive skills than test taking (e.g. Fitzpatrick (2008)). In this
study I cannot test whether there is an effect of the new compulsory education laws on
measures of non-cognitive outcomes, but the next section looks at another very plausible
channel through which improvements in educational achievement might arise: changes
in parent’s behavior.