Several reasons might lead technology to assist or impair human
capital attainment by students. Youths may employ the Internet in
educational matters such as writing papers, searches for answers to
questions and communicating with classmates on homework. However, time
spent in activities where "surfing the net" occurs could
substitute away from time allocated to reading, studying and completing
homework. This may hurt academic performance in the short term, which
might also diminish the ability or incentive to continue schooling over
the longer term.
Within the past decade, the Internet and WWW use have increased
substantially--for example, according to Pew Internet & American
Life Project Surveys, the percentage of U. S. online users has increased
from 40-45% in March 2000 to nearly 80% in April 2009 (Pew Internet
& American Life Project Surveys, 2009). Recent expansion of
adolescent use of the Internet is the result of an ongoing shift in
adolescents' daily behavior patterns. The majority of adolescents
from a sample in one study compared their online behaviors to the
phenomenon of placing telephone calls, which are typically mundane, the
purposes for which are both social and nonsocial (Gross, 2004). Hence,
adolescents' Internet use occurs without much thought or
consideration--it has become, in effect, just a normal daily activity.
Why is the potential impact of Internet use on educational outcomes
relevant for the discipline of economics? Human capital accumulation
bears directly and heavily on earning potential (see Grossman, 1972 and
Mincer, 1974) and it is widely accepted that strong and statistically
significant relationships link individual health and human capital
formation. Moreover, the impact of educational policies and factors that
affect learning continues to generate widespread public policy concern.
Thus, for economists and policy makers, gauging the relationship that
technology use has on educational outcomes is worthy of study.
Computer access and use among adolescents and other ages have grown
considerably over the past decade (Louge, 2006). In fact, more than 80%
of U.S. adolescents between the ages of 12 and 17 use the Internet, with
roughly half going online daily (Lenhart et al., 2005). The significance
of Internet use by children and adolescents has even spawned a new field
of inquiry in developmental psychology (Greenfield and Yan, 2006). With
the likelihood that Internet usage by adolescents will continue to
increase over time, concerns about the impact on high school
students' academic performance should be researched.
Stakeholders--parents, teachers, administrators, and the students
themselves--would benefit from knowing more about the digital
environment within which learning occurs. Regardless of whether academic
performance is positively or negatively impacted by Internet use, a
better understanding and greater awareness about such issues might
facilitate changes in pedagogy by educators, as well as learning on the
part of students and the support they receive from their parents.
In a conceptual context, we tacitly assume that students utilize
the Internet for both academic and non-academic purposes, with the most
intense users (which is described in the Data section) spending the most
time in non-academic pursuits (e.g. Facebook, downloading music). And
our general modeling framework is one of optimization, where there are
both educational benefits and costs to the Internet, and where the
primary benefit of Internet use is increased human capital accumulation
as evidenced by higher grades. At a basic level, Internet use denotes a
certain amount of technical savvy which emanates from a student actually
learning a new skill--this alone can translate into higher grades.
Benefits derived from Internet use usually come about at significant
costs, including deployment of the required infrastructure for providing
Internet access to students (which this study does not directly address)
as well as monetary and time costs devoted to the Internet that detract
from educational achievement (see Angrist and Lavy, 2002).
The central issue is to determine what, if any, level of Internet
use raises or lowers grades. This entails a quintessential marginal
benefit/ marginal cost analysis. This article begins the process by
examining quasi-defined levels of Internet utilization (where more
venues of use in a defined time period is assumed to equate to more
money and time devoted to use) and the resulting impact on student
The controversy over whether technology actually improves student
learning is one that stirs debate and motivates research. The articles
reported in the economics literature have been limited both in quantity
and scope with methods and results varying across studies. The
literature has focused primarily on the impact of technologies in
general on student learning; few studies have examined the direct link
between educational outcomes such as GPA and Internet use.
Gratton-Lavoie and Stanley (2009) compare undergraduate students
who opted to enroll in online microeconomics classes against those who
opted for the traditional in-class course. Results show a higher average
score on exams for students enrolled in online classes. However, after
accounting for selection bias, results indicate that age positively
affects students' average exam scores, with the online teaching
mode having a very small effect on average exam scores. Kubey et al.
(2001) uses a small survey of 572 students at a public university and
finds that heavy Internet use is highly correlated with poor academic
Angrist and Lavy (2002) argue that most studies covering
enhancements of learning through technology focus on qualitative
factors, such as participant perceptions. Thus, an empirical approach is
undertaken which compares outcomes between students who supplement
learning with computer aides against those students who do not. Their
results show that increased educational use of computers seems to have
little or no effect on students' test scores. Ordinary least
squares regression estimates demonstrate no relationship between
computer-aided instruction and academic achievement, with the exception
of a negative effect on eight-grade mathematics scores.
Ball et al. (2006) examine the effect of employing wireless
handheld technology by students on academic performance in undergraduate
principles of economics courses by way of a controlled experiment. One
group of students (experimental group) were equipped with wireless
handheld devices that allows interactive participation with standard
economics games, multiple choice tests, and communication with the
instructor during class time. The second group (control group) was not
given the devices. Course content, assignments, exams, and so on, were
identical between both groups. Results show that students in the
experimental group earned final grades that were an average of 3.2
points higher than did the students in the control group.
Anstine and Skidmore (2005) assess whether MBA students in online
economics classes learn as much of the material (measured by average
exam scores) as did their counterparts in the traditional economics
classes. Specifically, a small sample of MBA students was given the
option to enroll in either an online or traditional class. Accounting
for sample selection bias, regression analysis proffers that students in
the online classes did not learn as much, suggesting that the online
learning environment is less effective than the traditional classroom
Jackson et al. (2006) studies the impact of home Internet use on
academic performance of 140 low-income children between December 2000
and June 2002. The degree of Internet use is calculated using four
measures: minutes per day spent online, logins per day, number of
domains visited per day, and number of emails sent per day. Academic
performance of participants was measured by GPA and standardized test
scores on the Michigan Educational Assessment Program (MEAP). Results
suggest that children with greater Internet use had higher GPAs and
higher MEAP scores. However, the higher MEAP scores were only in the
reading portion, with Internet use having no effect on the mathematics
portion of the MEAP test.
It is worth noting that at least one study examined
adolescents' activities while online (Hunley, Evans,
Delgado-Hachey, Krise, Rich, Schell, 2005). Employing a logbook approach
whereby students documented their time for a seven-day period, Hunley et
al. (2005) found that at least 50% of the students (N = 101) logged the
following activities while online (hours per week indicated in
parenthesis): visiting web sites (1.27), playing games (4.43), reading
the news (0.73), researching information (1.22), and emailing (1.13).
Fewer than 50% of the students spent time chatting (2.12), word
processing (2.13), shopping (1.60), and "other" (2.00).
Many studies have limited sample sizes and education-related
variables. In contrast, our analysis employs a much larger sample size
of students for which there is substantially greater information on
demographics and household characteristics. Moreover, the number of
variables available in our dataset is large and generally exceeds the
number of variables found in the datasets in the above studies.
Since its inception in 1979, the National Survey on Drug Use and
Health (NSDUH), sponsored by the Substance Abuse and Mental Health
Services Administration (SAMHSA), is administered annually to
approximately 55,000 civilian, non-institutionalized individuals age 12
and over, chosen so that the application of sample weights produces a
nationally representative sample with approximately equal numbers of
respondents from the 12-17, 18-25, and 26 and over age groups.
Variables on Internet use are collected and compiled by SAMHSA
administrators only for the 2005 survey; hence these are the data we
analyze. Our sample consists of 12,184 enrolled high school students.
Data from the NSDUH allow for both breadth and depth of coverage on the
topic. Breadth comes from the ability to study aspects of educational
outcomes using data from an elaborate questionnaire administered to
12-17 year olds on a wide array of youth experiences. An assortment of
variables are observed, therefore, that have the potential to serve as
predictors for grades in the proposed model. Depth is provided by
variables on race, gender, family income, family composition, religion
A potentially problematic attribute of the data is non-random
measurement error emanating from the self-reported nature of responses.
However, studies on the quality of self-reported academic variables data
suggest that such reporting bias should be minimal. Cassady (2001) finds
that self-reported GPA values are "remarkably similar to official
records" and therefore are "highly reliable" and
"sufficiently adequate for research use." Hunley et al. (2005)
address concerns about self-reported survey data by way of demonstration
of the reliability of survey data as "appropriate" for
measuring accurately adolescents' Internet use. Specifically,
students provided estimates of their Internet use, and then logged their
actual daily Internet use for a one week period. Comparisons between
estimated Internet use and actual use showed reliability of the
self-reported estimates. Their conclusion is that researchers should
feel confident about self-reported survey data pertaining to Internet
RESEARCH METHOD AND EMPIRICAL SPECIFICATION
Consider the following equation, in which Grades is a function of
exogenous factors with Internet usage of prime importance,
Grades = [[beta].sub.0] + [[beta].sub.1]IU + X[[beta].sub.2] +
In the above equation, which applies to individual NSDUH
respondents (with the corresponding observation-level subscript
suppressed), IU represents venues of Internet usage in the past 30 days.
Vector X represents a set of other exogenous variables that conceivably
affect grades. The [beta]'s are parameters to be estimated and s is
the error term.
We investigate effects on grades by analyzing the probability the
student receives an 'A' or 'B' average or an average
of 'D' or below. Grades is measured using a 1-4 scale with
'4' representing A+, A, A- ; '3' representing B+, B,
B-; '2' representing C+, C, C- and '1' representing
D or below.
When the survey is administered, respondents are queried on venues
of Internet utilization in the past 30 days. We categorize Internet
users in three forms: Level 1; Level 2; and Level 3. For individuals in
Level 1, the Internet was utilized at home, at school, at a
friend's house, at a cafe with Internet access, over a cell phone
and some other place--this variable is "open" and does not
have specific options. For those in Level 2, the Internet was utilized
at home and at school. For those in Level 3, the Internet was utilized
only at school. We term those in Level 1 as intense Internet users;
those in Level 2 as moderate users; and those in Level 3 as light users.
For light usage, Internet access is subject to time constraints (i.e.
hours of operation for schools), whereas for intense and moderate usage,
there is virtual 24 hour access. To avoid the "dummy variable
trap" in the regressions, those that did not use the Internet (no
use) in the past 30 days is the omitted category and is used as the
category of comparison.
Several variables from the NSDUH data are considered explanatory in
equation (1): age indicators are included for whether the student is 14,
15, 16 or 17 years old with age 13 as the omitted category to avoid the
"dummy variable trap." Binary indicators are included for
whether the mother or father resides in the household, for whether
parents assisted the student with homework always or sometimes in the
past 12 months, with "never" as the omitted category, and for
whether the student is currently classified as a sophomore or junior/
senior, with "freshman" as the omitted category. We also
include a binary variable for school type (public or private). Potential
endogeneity (stemming from students' "self-selecting"
into certain learning environments by choosing to attend certain
schools) should be mitigated in that location of high school attendance
is largely determined by parental preferences in occupation, living
conditions, as well as other correlates.
To control for the possibility that a student subscribes to a
"work hard-play hard" ethos and therefore heavily utilizes the
Internet yet maintains high grades, a binary indicator is incorporated
for a student that heavily uses the Internet and also states that school
work is important/ meaningful, and is thus more likely to have good
grades. We term this a "high motivation" student.
Family income is measured in four categories: $10,000-$19,999;
$20,000-$49,999; $50,000-$74,999; and $75,000 or greater, with
$10,000-$19,999 as the omitted category. A measure for the number of
times the student moved in the past year is incorporated as is a binary
indicator for gender. For race, indicators are specified for Caucasians,
African Americans and Asians, with non-white Hispanics as the omitted
category. Further, student physical health is measured as follows: great
health, good health and fair health with "poor health" as the
omitted category. A factor for religiosity is also included given that
this may proxy for increased academic discipline. For this factor, a
binary variable is created and coded as '0' if religion does
not influence decisions and '1' if it does. Religiosity has
been linked to educational outcomes (Wolaver, 2002).
Table 1 presents select summary statistics. Intense Internet use is
0.047 and moderate Internet use is 0.491 while light use is lower with a
mean of 0.350--all indicating abundant exposure to the Internet.
Approximately eight percent of students attend private schools. Fathers
are less likely to be present in the household than are mothers and the
proportion of parents that always help with homework is also quite high
(0.54). Caucasians comprise approximately 63 percent of the sample,
African Americans about 14 percent, while non-white Hispanics and Asians
account for about 15 percent and three percent, respectively. About one
third of students report being in excellent health, with 41 percent
reporting good health, and a large proportion (0.651) state that
religion influences decision making.
The Effects of Internet Use on the Probability of Obtaining an
'A' or 'B'
As shown in Table 2, intense Internet use is significant and lowers
the probability of earning an 'A' or 'B' versus
lower grades; light Internet use also lowers the probability while
moderate use elevates the probability of an 'A'/
'B'. The Log Pseudolikelihood is -6707.84. Intense Internet
use reduces the probability of achieving an 'A'/ 'B'
by 0.03--for students that are intense Internet users, the probability
of having an 'A'/ 'B' average is undercut by
approximately 5 percent compared to students who did not use the
Internet at all in the past 30 days (to which, for parsimony, we refer
to as 'no use' for the remainder of the section). If a student
reports moderate usage, the probability of having an 'A'/
'B' increases by 0.08 compared to no use--moderate users have
a roughly 12 percent increased probability of earning this average
compared to no use. Light internet users have about a 6 percent lower
probability of earning an 'A'/ 'B' versus no use.
The negative effects associated with intense Internet utilization
may indicate that this level of usage actually impairs the learning
process (perhaps by lowering attention span) which, in turn, reduces the
capability of the student to earn top grades. Also, students using the
Internet at a friend's house or cafe may be distracted by
non-academic conversations even when using the Internet for academic
purposes. In addition, intense use may translate into less time spent on
and homework and studying, compared to no use; hence, grades are lower
for those in the intense use category versus no use.
Interestingly, light users have a diminished probability of an
'A'/ 'B' versus no use. This may provide evidence
that when students have Internet access only at school, that time is
utilized "surfing the net" for recreational purposes (e.g.
Facebook), which is time subtracted from studying; therefore, grades are
actually lower for those in the light use category compared to no use.
Overall, moderate use (which includes home use as a major component) has
the most positive impact on grades, which could indicate that home
Internet use by students is more focused on academic pursuits compared
to other venues.
As stated in our Motivation section, there is an opportunity cost
involved in using the Internet, which includes reduced study time and
possibly increased devotion of the students' monetary resources to
Internet services that detracts from the prospect of receiving an
'A'/ 'B' average. These results imply that those
costs are salient. This is an interesting contrast to the study done by
Jackson et al. (2006), which (as discussed earlier) found that
adolescents who used the Internet more had higher grade point averages.
An additional contrast to our results and the results of the Jackson et
al. (2006) study are the results of Hunley et al. (2005), which did not
show a significant relationship between time spent on the computer at
home and grades.
The Effects of Internet Use on the Probability of a 'D'
or Lower Average
Table 3 presents the regression estimates for the probability the
respondent has a 'D' or lower grade versus other grades. The
Log Pseudolikelihood is -6707.84. Intense Internet use elevates the
probability of achieving a 'D' or lower grade by almost 0.02.
If a student reports moderate usage, the probability of having a
'D' or lower average falls by 0.03 compared to no use, but
rises by 0.01 for light use (compared to no use). Intense users have a
higher probability of a 'D' or lower grade (about 25 percent),
while moderate users have a decreased probability (approximately 28
percent) of having this average, compared to students who report no use.
Light users have a roughly 13 percent increased probability of a
'D' or lower average compared to no use.
The estimated effect for intense use is rather large, even
accounting for the fact that the outcome incorporates grades of
'D' and 'F'. Again, there may be large opportunity
costs associated with such rigorous Internet use which undermines
academic achievement. Thus, grades are lower and higher failure rates
may account for some of the largeness. Moreover, moderate users fare
better academically compared to no use: moderate users have a decreased
probability of earning a 'D' or less versus those
students' that report no Internet use. For light users, the
probability of earning 'D' or lower is higher compared to no
use, again potentially indicating that students who only have Internet
access at school spend this time in recreational use and hence suffer
lower grades as study time falls.
The Effects of Other Explanatory Variables on Grade Probabilities
Many of the other explanatory variables have a significant impact
on grades. Interestingly, "High Motivation" students have a
greater probability (0.12) of earning an 'A'/ 'B'
average but the probability of earning a 'D' or lower is
reduced by 0.06. The presence of mothers in the households generally has
a favorable impact on 'A'/ 'B' grades, while the
presence of fathers is not significant. However, parental involvement
does have profound effects as assisting with homework raises student
grades. For example, if a parent always helps with homework, the
probability of an 'A'/ 'B' rises by approximately
0.06; the probability of 'D' or lower falls by 0.02.
Those that attend private schools have a 12 percent greater
probability of earning an 'A'/'B' and a 27 percent
lower probability of having a 'D' or lower average. In
addition, Caucasians and Asians have higher probabilities of achieving
an 'A'/ 'B' average versus African Americans, while
females enjoy a higher probability of 'A'/ 'B' and
versus males. Higher levels of income are also significant in some
instances. Students in families earning $20,000-$49,999 and
$50,000-$74,000 a year have a greater probability of obtaining an
'A'/ 'B' average (0.037 and 0.197 respectively) and
lower probability of having a 'D' or less (-0.008 and -0.017
respectively), compared to families earning $10,000-$19,999.
As students advance in age, the probability of having an
'A'/ 'B' mildly decreases and the probability of a
'D' or lower increases. Of course, this may indicate an
increasing opportunity cost involved in studying and in other
educational activities as students learn to drive, enjoy more personal
freedom and possibly rebel against parents. The effects are opposite for
class standing where students that are juniors/ seniors have enhanced
probabilities of earning an 'A'/ 'B' and lower
probabilities of earning a 'D' or less. This could imply that
at least some students study more in an effort to "drive-up"
GPA's for approaching college entrance.
In keeping with broader literatures on human capital, students that
are in better health also earn higher grades (higher probability of
'A'/ 'B'; lower probability of 'D' or
less), while those that relocate more often have lower 'A'/
'B' probabilities and higher 'D' and below
probabilities. In addition, religiosity impacts grades: students who
state religious beliefs influence decisions have a 0.064 greater
probability of having an average 'A'/ 'B' average
and a 0.025 diminished probability of having a 'D' or less
than 'D' average. For the most part, our results demonstrate
that the number of venues of Internet use have an impact on the academic
achievement of high school students even after controlling for a host of
For this study, there is evidence that the grades of high school
students are lowered when additional venues of Internet access are
utilized. Specifically, when all venues of Internet use are exhausted,
which we refer to as intense use, grades are lower when compared to
students that report no Internet use. Moreover, students that only use
the Internet at school, which we term light use, also suffer from lower
grades compared to those that did not utilize the Internet. Conversely,
students that used the Internet at school and at home, which we term
moderate use, enjoy higher grades versus those that did not use the
Internet. Our model supports a hypothesis of "optimal"
Internet use. Results indicate that grades are higher when students
undertake moderate Internet use; however, grades decline when students
are below or surpass a certain threshold (i.e. optimum). Potentially
large opportunity costs of Internet use (in the possible form of
detractions from time spend studying and engaging in other activities
that enhance grades) may be present for intense and light Internet
The results provide useful information to high school
administrators, teachers, counselors, parents, and students, when they
consider implications for use of the Internet in an educational setting.
Moreover, university administrators and faculty will find the results
helpful, since many high school graduates continue their education by
way of college and university studies. From a policy perspective, high
school administrators may wish to consider guidelines that curtail
non-academic Internet use in schools.
Our data did not explicitly outline whether students' Internet
use was for academic or social purposes; therefore, future research that
incorporates this data would provide more information. In addition, the
costs of deploying the required infrastructure needed to provide
Internet access to students would prove useful in continued analyses of
the benefits and costs of the Internet.
Angrist, J. & V. Lavy (2002). New evidence on classroom
computers and pupil learning. The Economic Journal, 112(482), 735-765.
Anstine, J. & M. Skidmore (2005). A small sample study of
traditional and online courses with sample selection adjustment. Journal
of Economic Education, 36(2), 107-127.
Ball, S. B., C. Eckel, & R. Christian (2006). Technology
improves learning in large principles of economics classes: Using our
WITS. The American Economic Review, 96 (2): 442-446.
Cassady, J.C. (2001). Self-reported gpa and sat: A methodological
note. Practical Assessment, Research & Evaluation, 7(12). Retrieved
December 13, 2010, from http://PAREonline.net/getvn.asp?v=7&n=12.
Gratton-Lavoie, C. & D. Stanley (2009). Teaching and learning
principles of microeconomics online: An empirical assessment. Journal of
Economic Education, 40(1), 3-25.
Greenfield, P. & Z. Yan (2006). Children, adolescents, and the
internet: A new field of inquiry in developmental psychology.
Developmental Psychology, 42(3), 633-649.
Gross, E. (2004). Adolescent internet use: What we expect, what
teens report. Applied Developmental Psychology, 25(6), 633-649.
Grossman, M. (1972). On the concept of health capital and the
demand for health. Journal of Political Economy, 80(2), 223-255.
Hunley, S.A., J. H. Evans, M. Delgado-Hachey, J. Krise, T. Rich,
& C. Schell (2005). Adolescent computer use and academic
achievement. Adolescence, 40(158), 307-318.
Jackson, L.A., A. von Eye, F.A. Biocca, G. Barbatsis, Y. Zhao &
H.E. Fitzgerald (2006). Does home internet use influence the academic
performance of low-income children? Developmental Psychology, 42(3),
Kubey, R.W., M.J. Lavin & J.R. Barrows (2001). Internet use and
collegiate academic performance decrements: Early findings. Journal of
Communication, 51(2), 366-382.
Lenhart, A., M. Madden & P. Hitlin (2005). Teens and
technology: Youth are leading the transition to a fully wired and mobile
station. Pew Internet & American Life Project, 27 July 2005.
Retrieved June 10, 2009, from
Louge, N. (2006). Adolescents and the internet. ACT for Youth
Center of Excellence. Retrieved June 15, 2009, from
Mincer, J. (1974). Schooling, experience and earnings. New York:
National Bureau of Economic Research.
Pew Internet & American Life Project Surveys (2009). Internet
adoption. Retrieved August 4, 2009, from
Wolaver, A M. (2002). Effects of heavy drinking in college on study
effort, grade point average, and major choice. Contemporary Economic
Policy, 20(4), 415-428.
Wesley Austin, University of Louisiana at Lafayette
Michael W. Totaro, University of Louisiana at Lafayette
Table 1. Descriptive Statistics
Variable Mean Deviation
Probability of an 'A' or 'B' grade 0.684 0.465
Probability of a 'D' or lower grade 0.070 0.256
Intense Internet Use (past 30 days) 0.047 0.213
Moderate Internet Use (past 30 days) 0.491 0.499
Light Internet Use (past 30 days) 0.350 0.407
No Internet Use (past 30 days) 0.112 0.315
High Motivation Student: heavy internet use/
positive school attitude 0.713 0.452
Mother in household 0.918 0.275
Father in household 0.732 0.443
Respondent is female 0.501 0.500
Attending private school 0.082 0.274
Age of student (13 years old) 0.134 0.340
Age of student (14 years old) 0.215 0.410
Age of student (15 years old) 0.228 0.420
Age of student (16 years old) 0.222 0.415
Age of student (17 years old) 0.192 0.394
Race (Caucasion) 0.631 0.483
Race (African American) 0.136 0.342
Race (Asian) 0.030 0.170
Race (non-white Hispanic) 0.152 0.359
Sophomore 0.220 0.414
Junior or Senior 0.324 0.468
Family income (less than $20,000) 0.180 0.344
Family income ($20,000-$49,999) 0.345 0.475
Family income ($50,000-$74,999) 0.202 0.402
Family income ($75,000 or more) 0.286 0.452
number of times moved (past year) 0.322 0.696
Parents help with homework (always) 0.547 0.498
Parents help with homework (sometimes) 0.230 0.421
Student health status (great) 0.331 0.471
Student health status (good) 0.418 0.493
Student health status (fair) 0.213 0.410
Religion influences decisions 0.651 0.477
Table 2. Probit estimates for the probability of an 'A' or 'B'
(n = 12,184)
Log Pseudolikelihood = -6707.84
Explanatory variables Coefficient Error
Intense Internet use -0.034 *** (0.021)
Moderate Internet use 0.082 * (0.014)
Light Internet use -0.039 * (0.014)
High Motivation Student 0.116 * (0.011)
Mother in household 0.057 * (0.016)
Father in household 0.012 (0.011)
Respondent is female 0.145 * (0.008)
school type (private) 0.082 * (0.015)
Age of student (14 years old) -0.047 * (0.016)
Age of student (15 years old) -0.127 * (0.019)
Age of student (16 years old) -0.193 * (0.024)
Age of student (17 years old) -0.191 * (0.028)
Race (Caucasian) 0.089 * (0.020)
Race (African American) -0.011 (0.022)
Race (Asian) 0.198 * (0.018)
Sophomore 0.073 * (0.014)
Junior or Senior 0.137 * (0.018)
Family income ($20,000-$49,999) 0.006 (0.013)
Family income ($50,000-$74,999) 0.037 ** (0.015)
Family income ($74,999 and over) 0.097 * (0.014)
number of times moved (past year) -0.035 * (0.006)
Parents help with homework (sometimes) 0.021 ** (0.006)
Parents help with homework (always) 0.057 * (0.008)
Student health status (great) 0.217 * (0.019)
Student health status (good) 0.164 * (0.021)
Student health status (fair) 0.062 * (0.022)
Religion influences decisions 0.064 * (0.009)
* statistically significant at 1%
** statistically significant at 5%
*** statistically significant at 10%
Table 3. Probit estimates for the probability of a 'D' or lower
(n = 12,184)
Log Pseudolikelihood = -2697.80
Explanatory variables Coefficient Error
Intense Internet use 0.018 ** (0.010)
Moderate Internet use -0.021 * (0.005)
Light Internet use 0.009 *** (0.005)
High Motivation Student -0.053 * (0.006)
Mother in household -0.007 (0.007)
Father in household 0.001 (0.004)
Respondent is female -0.025 * (0.003)
school type (private) -0.019 * (0.006)
Age of student (14 years old) 0.011 (0.007)
Age of student (15 years old) 0.033 * (0.009)
Age of student (16 years old) 0.065 * (0.014)
Age of student (17 years old) 0.053 * (0.016)
Race (Caucasian) -0.012 -0.008
Race (African American) -0.013 *** (0.007)
Race (Asian) -0.037 * (0.006)
Sophomore -0.002 * (0.005)
Junior or Senior -0.048 * (0.006)
Family income ($20 ,000-$49,99 9) 0.005 (0.005)
Family income ($50,000-$74,999) -0.008 (0.006)
Family income ($74,999 and over) -0.017 * (0.006)
number of times moved (past year) 0.011 * (0.002)
Parents help with homework (sometimes) -0.001 * (0.002)
Parents help with homework (always) -0.020 * (0.003)
Student health status (great) -0.057 * (0.006)
Student health status (good) -0.051 * (0.007)
Student health status (fair) -0.021 * (0.006)
Religion influences decisions -0.025 * (0.004)
* statistically significant at 1%
** statistically significant at 5%
*** statistically significant at 10%