-
10.1177/0013916504269665ENVIRONMENT AND BEHAVIOR / July
2005Meinhold, Malkus / ADOLESCENT ENVIRONMENTAL BEHAVIORS
ADOLESCENT ENVIRONMENTALBEHAVIORSCan Knowledge, Attitudes,
andSelf-Efficacy Make a Difference?
JANA L. MEINHOLD is an instructor with the Child and Family
Studies Programat Portland State University. She is a Ph.D.
candidate with the Department of HumanDevelopment and Family
Sciences at Oregon State University and received an M.A.and a B.A.
in human development from Washington State University, Pullman.
AMY J. MALKUS is an assistant professor of early childhood
education at EastTennessee State University, Johnson City. She
received an M.S. and a Ph.D. in childdevelopment/early childhood
education from Purdue University and a bachelorsdegree in
psychology from Washington College.
ABSTRACT: This article examines the relationships among
adolescent environmen-tal behaviors and self-efficacy, knowledge,
and attitudes. It was hypothesized thatadolescents who demonstrate
more proenvironmental attitudes are more likely todemonstrate
proenvironmental behaviors. It was further hypothesized that
perceivedself-efficacy would have a moderating effect on the
environmental attitude-behaviorrelationship in that the
relationship between proenvironmental attitudes and behav-iors
would be stronger among adolescents with high levels of
self-efficacy. Partici-pants were 848 students from three
academically achieving high schools on the Westcoast. Hierarchical
regression analyses were used for all subsequent analyses.
Resultsindicated that proenvironmental attitudes significantly
predicted proenvironmentalbehaviors and that environmental
knowledge was a significant moderator for the rela-tionship between
environmental attitudes and environmental behaviors. This
wasespecially true for males.
Keywords: adolescent environmental attitudes; behaviors;
knowledge; self-efficacy
The environment that supports mankinds survival has always been
aheated topic of discussion. Whether it is concern about the ozone
layer, the
511
ENVIRONMENT AND BEHAVIOR, Vol. 37 No. 4, July 2005 511-532DOI:
10.1177/0013916504269665 2005 Sage Publications
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
depletion of the rain forests, or air and water pollution, human
effects on theenvironment have historically been, and will continue
to be, of great concernto many individuals. Scientists, teachers,
children, and farmers each havetheir own stake in the environment.
The leaders of today are working globallyand locally to develop
programs and laws to protect many beautiful and valu-able areas of
the world. Laws, especially environmental laws, are importantfor
the sustainability of the earth and the resources important for
human sur-vival. Who is it that will make decisions about the
environment in the future?Who will create programs and fight for
policies and laws to protect theenvironment? Todays young
people.
It is clear that the attitudes, knowledge, behaviors, and
concerns that ado-lescents have about the environment will directly
and indirectly affect futuredecisions concerning our natural
resources and how they will be used andsustained. Youth of today
are the future policy and decision makers. There-fore, if we
examine adolescents attitudes, concerns, and knowledge aboutthe
environment around them and throughout the world, we can have a
betterunderstanding of the direction we are headed. It is important
to know whatadolescents think and feel about the environment, where
they learn about it,and how much they know. It is also important to
learn the direct contributionsthey are making to sustain and
support the environment. For this reason, it isvital that we
examine the possible connection adolescent environmental
atti-tudes, knowledge, and self-efficacy have on environmental
behaviors.
Self-efficacy is defined as the confidence that individuals have
in theirability to plan and execute a course of action and to
accomplish a task or solvea problem (Bandura, 1994). Self-efficacy
is an important topic to examinebecause of its close relationship
to self-esteem, locus of control, and pro-social development. Those
individuals showing both high levels of confi-dence and of control
in their abilities to execute and accomplish tasks arelikely to
show tendencies toward participating in prosocial behaviors,
thosebehaviors that are intended to help or benefit an individual
or group of people(Eisenberg & Mussen, 1989). Adolescents may
participate in prosocial actsfor many reasons: to gain approval, to
receive feelings of caring, or to showsympathy toward another.
There is a level of confidence adolescents may
512 ENVIRONMENT AND BEHAVIOR / July 2005
AUTHORSNOTE: This project was presented as a poster at the 2003
biennial meet-ing of the Society for Research in Child Development
in Tampa, Florida. Correspon-dence concerning this article should
be addressed to Jana L. Meinhold, Child andFamily Studies, Portland
State University, P.O. Box 751, Portland, OR 97207-0751;phone:
(503) 725-8562; fax: (503) 725-5430; e-mail: [email protected].
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
achieve when participating in prosocial acts; this should be
similar to feelingsof confidence they may have when participating
in prosocial acts toward theenvironment. Fogel, Melson, and Mistry
(1986) believe definitions of pro-social behavior should be
expanded. This expansion should include the dis-play of nurturance
toward and voluntary acts intended to benefit animals,plants, and
the environment in general, not just human beings.
Eisenberg and Mussen (1989) state that childrens feelings of
sympathy orcaring toward people can lead to prosocial behaviors on
their behalf. It hasbeen hypothesized that children and adolescents
who experience feelings ofconcern or caring about the environment
will most likely produce those sameprosocial behaviors. However, in
this instance, these prosocial behaviors willbe directed toward
different environmental causes as opposed to otherhumans (Malkus
& Musser, 1993, 1997; Szagun & Mesenholl, 1993).
There-fore, when using the expanded definition presented by Fogel
et al. (1986), itcan be seen that proenvironmental behaviors and
attitudes fit nicely into theframework of prosocial behaviors and
development.
Specifically, there is a connection hypothesized between
adolescentsprosocial acts toward the environment and their positive
feelings about them-selves and the contributions they can make in
the future. One can assume thatas adolescents engage in and feel
better about proenvironmental acts, theirlevels of self-esteem,
self-efficacy, and internal locus of control in their liveswill
increase as well. Some aspects of this assumption have been
demon-strated with younger children in previous research on the
connections amongproenvironmental attitudes, perceived competence,
and locus of control(Malkus & Musser, 1993, 1997). Researchers
have also looked at the rela-tionships among environmental actions
and variables such as locus of controlin adolescents, with similar
connections being found (Szagun & Mesenholl,1993; Szagun &
Pavlov, 1993).
It is important to identify the environmental attitudes,
behaviors, andknowledge of todays adolescents so researchers can
work to better under-stand adolescents actions within their
environment and what it means tothem. Self-efficacy is another
variable that would allow researchers to under-stand and gauge
levels of adolescent behavior, knowledge, and attitudestoward the
environment. Overall, adolescent populations are important to
thefuture sustainability of our environment both in this country
and throughoutthe world. High levels of general self-efficacy in an
individual may allowresearchers to identify the types of
individuals that will be more likely to havepositive attitudes and
behaviors toward and a greater knowledge about theenvironment.
Meinhold, Malkus / ADOLESCENT ENVIRONMENTAL BEHAVIORS 513
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
REVIEW OF LITERATURE
Environmental researchers and educators have focused on the
attitudes,knowledge, and behaviors of adolescents for several
decades. There has beensubstantial research compiled looking at
adolescents and the environment,and demographic variables, in
relation to their environmental attitudes andbehaviors (Bradley,
Waliczek, & Zajicek, 1997; Hausbeck, Milbrath, &Enright,
1992; Kahn & Friedman, 1995; Lyons & Breakwell, 1994;
Mohai,1992; Ostman & Parker, 1987; Riechard & McGarrity,
1994; Riechard &Peterson, 1998; Schahn & Holzer, 1990).
There have also been numerousprojects that address the
relationships environmental behaviors, knowledge,and attitudes have
with one another (Armstrong & Impara, 1991; Campbell,Waliczek,
Bradley, Zajicek, & Townsend, 1997; Lyons & Breakwell,
1994;Oskamp et al., 1991; Ramsey & Rickson, 1976).
Unfortunately, within thefield of environmental studies, there has
been little, if any, research publisheddiscussing self-efficacy and
its connection to environmental issues. Therehave been a number of
studies addressing self-efficacy and its connection toissues such
as smoking, drinking, pregnancy, and drug use in the lives of
ado-lescents (Denoff, 1987; Kaliski, Rubinson, Lawrance, &
Levely, 1990;Lawrance, Levely, & Rubinson, 1990; OLeary, 1985;
Roberts, 1982; Walker& Greene, 1987), but this has not yet been
extended to environmental issues.
Self-efficacy, as described by Bandura (1977), tells us that an
individualsbeliefs have a profound effect on actions and intended
outcomes. Banduras(1977, 1986) theory of self-efficacy suggests
that peoples ability to success-fully execute an action is
determined by their belief in their ability to do so. Ifpeople have
strong beliefs in their abilities to change the world around
them,they will produce more effective coping strategies and higher
levels ofachievement than those showing lower levels of belief in
their abilities.Banduras theory of self-efficacy gives researchers
a framework to predictand more accurately explain behavior and the
changes in behavior humansexperience. Self-efficacy is not only a
perception of ability; it also affects anindividuals motivation.
Bandura (1986) believed that an individual producesself-appraisals
(judgments about abilities), and these self-appraisals can havea
powerful effect on motivation. When people believe they are
proficient at atask, there is a greater amount of motivation to
work on those tasks with littledifficulty.
The current project will address issues of self-efficacy both in
a generalsense and in the sense of adolescents environmental
self-efficacy. Littleresearch has focused on the general
self-efficacy of normal to high achievingadolescents; the current
project hopes to address this hole within theresearch. Not only are
there few studies addressing adolescent self-efficacy,
514 ENVIRONMENT AND BEHAVIOR / July 2005
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
there is little research looking at the connection self-efficacy
has to environ-mental actions, behaviors, and knowledge. The
current project will, in part,be an exploratory study designed to
address these gaps in the field, bothenvironmental and
efficacious.
LINKING SELF-EFFICACY AND THE ENVIRONMENT
When examining the environmental attitudes, behaviors, and
knowledgeof adolescents in connection with their perceived
self-efficacy, there has beenlimited research available. Concepts
related to self-efficacy (i.e., locus ofcontrol, self-concept, and
self-esteem or perceived competence; three con-cepts which have
definite parallels) have been addressed within the field
ofenvironmental research. These three concepts, although not
identical to self-efficacy, are related enough to provide a
tentative link between environmen-tal behaviors and perceived
self-efficacy.
Locus of control is expressed along a continuum, with an
internal locus ofcontrol at one end and an external locus of
control at the other. An internallocus of control indicates that an
individual perceives that his or her actionswill produce the
desired outcome, whereas an external locus of control repre-sents
the individuals belief that life outcomes, such as the one in
question,are due to outside forces (Bandura, 1986; Gecas, 1989).
Locus of control canbe linked to self-efficacy using the parallels
seen within ones belief or per-ceived abilities to change or
control ones life. Self-esteem is another conceptthat is often
associated with self-efficacy and locus of control.
Self-esteemdeals with the judgment of ones personal self-worth
(Bandura, 1986; Stan-ley & Murphy, 1996). Self-worth is not
associated with actions and controlissues the way self-efficacy and
locus of control are but instead deals with theindividuals
perceptions and beliefs of the self. Self-concept, the third
vari-able that holds similarities to self-efficacy, is defined as
the self-appraisal thatis formed through the experiences and roles
the individual ascribes (Bandura,1997; Solberg, OBrian, Villareal,
Kennel, & Davis, 1993). An individualwho ascribes to the team
player role when involved in sporting events willproduce a
self-concept that is reflective of the role he or she is playing at
thetime. Self-concept is different from self-esteem in that
self-concept places avalue on the concept or role, whereas
self-esteem only refers to the individ-uals sense of worth.
Each of these concepts of self are related to prosocial
behavior, the volun-tary action that an individual produces with
the intention to help or benefitanother individual or group of
individuals (Eisenberg & Mussen, 1989).
Meinhold, Malkus / ADOLESCENT ENVIRONMENTAL BEHAVIORS 515
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
Children may participate in prosocial acts for many reasons, but
mostprosocial acts are to gain approval from others. Recently,
researchers havebeen addressing the link between prosocial
development and environmentalattitudes and concerns (Malkus &
Musser, 1993, 1997; Szagun & Mesenholl,1993; Szagun &
Pavlov, 1993). When environmental attitudes and behaviorsare
applied to the variables related to prosocial development (i.e.,
locus ofcontrol, self-esteem, self-efficacy), the definition takes
on an expanded form.This expanded definition includes voluntary
behaviors and actions directednot only at others but also toward
pets, animals, and the protection and con-servation of the
environment (Fogel et al., 1986). Researchers have foundsignificant
positive correlations among perceived competence and
globalself-worth, internal locus of control, and proenvironmental
attitudes and con-cerns (Malkus & Musser, 1993, 1997). Overall,
findings show a connectionbetween childrens positive feelings
toward the environment and their levelof prosocial development. It
can be inferred that childrens attitudes aboutand direct actions on
the environment can have an impact both ecologicallyand in the
positive feelings children have about themselves (Malkus
&Musser, 1993, 1997).
When individuals show concern and empathy toward the
environment,and in turn act on these feelings, it can be seen as a
form of prosocial develop-ment. Previous research (Malkus &
Musser, 1997; Szagun & Mesenholl,1993) has shown a relationship
between environmental attitudes and pro-social development in the
form of concern. When an individual comes in con-tact with a
distressing situation concerning the environment, and when
theresponse is one of concern and positive action, the individual
is engaging inprosocial behavior.
In summary, it has been tentatively demonstrated that
environmental con-cern is closely related to prosocial development
through the common vari-ables of self-esteem, self-efficacy, and
locus of control. Individuals whoshow concern for the environment
and act on these concerns demonstrate ahigh level of prosocial
development. This, in turn, may allow individuals togenerate
positive feelings about themselves. When experiencing
positivefeelings, individuals may also experience an increase in
the degree to whichthey feel the future execution of actions and
behaviors will consistently pro-duce positive outcomes. At the same
time, this increases the likelihood offuture incidences of
prosocial behavior, whether directed at the environmentor at other
people, and so it is hypothesized to continue in an
ever-expandingcircle.
Based on the research and theoretical explanations provided
above, thefollowing hypotheses were produced. First, adolescents
who demonstratemore proenvironmental attitudes will be more likely
to demonstrate proenvi-
516 ENVIRONMENT AND BEHAVIOR / July 2005
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
ronmental behaviors compared to adolescents who demonstrate
lower levelsof proenvironmental attitudes. Second, adolescents who
demonstrate moreproenvironmental attitudes and high perceived
self-efficacy will demonstratemore proenvironmental behaviors than
will adolescents with less proenvi-ronmental attitudes and lower
self-efficacy. And third, adolescents whodemonstrate more
proenvironmental attitudes and greater environmentalknowledge will
demonstrate more proenvironmental behaviors than willadolescents
with less proenvironmental attitudes and lower
environmentalknowledge.
METHOD
PARTICIPANTS
Participants for this project were 848 adolescents (377 males,
469 females,2 unknowns), between 14 and 18 years old, from high
schools located inthree large cities on the West coast of the
United States (Seattle, Portland, LosAngeles). A single high school
from each city was selected to participatebased on the criteria
outlined below. Schools were selected from three differ-ent
geographic locations to allow for a more diverse population. This
resultedin a range of different ethnicities represented (White =
70%, Asian = 11%,Hispanic = 8%, Black = 5%, Other = 7%). The sample
population includedall public schools within chosen counties that
fall in the three cities.Researchers utilized state assessment
scores from the 1997-1998 or 1998-1999 (depending on availability
of data) school year in the selection process.
DESIGN AND PROCEDURE
In the fall of 1999 and spring of 2000, data were collected in
one highachieving school from each of the cities. In selecting the
sample, a list of allpublic schools in each of the three cities was
obtained from the appropriateState Board of Education Internet
sites for Washington, Oregon, and Califor-nia. Distinct but
complementary sampling strategies were used in each state.Sampling
procedures among the three states were kept as similar as
possible,but the ranking method used to determine the highest
achieving schools var-ied by state, due to differences in the
assessments used, the year for whichdata were available, and the
grade level of students assessed. Therefore, thesampling strategy
for each state is described separately. In Washington, thedata from
the Iowa Test of Educational Development (Iowa) for the 1999
Meinhold, Malkus / ADOLESCENT ENVIRONMENTAL BEHAVIORS 517
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
school year were used. In Oregon, the data for the Oregon
Statewide Assess-ment (OSA) from the 1998 school year were used.
And in California, scoreson the SAT for the 1997-1998 school year
were used.
Once top ranked schools were identified for each list, school
names werechosen at random and placed on the lists in the order
they were selected.Schools were contacted in the order they were
chosen until participationfrom one school on the list was secured.
Once a school agreed to participate,parental letters and parental
consent forms were distributed by a member ofthe research team or
by school personnel, depending on school preference, toall
sophomores and juniors at the school. Students were given about 1
weekto return parental consent forms depending on the school
schedule.
Approximately 1 week after students received parental forms,
theresearch team administered the survey during school hours to all
studentswith parental permission in a classroom setting or in a
location determinedsafe by teachers and school administrators. Once
students were briefed, theywere asked to fill in the demographic
portion of their survey, and a specialcode number was assigned to
each school for identification purposes. Thesurvey was
confidential, and names were not gathered on the data
collectionsheet. As a student completed the survey, it was placed
in a pile at the front ofthe room, and the student was either asked
to wait quietly or to return to par-ticipation in the regularly
scheduled classroom assignment. All schools par-ticipating in the
project were formally thanked with a letter.
MEASURES
A 142-item survey, Young People and the Environment, was used to
mea-sure adolescent attitudes, knowledge, goals, self-efficacy, and
behaviors rel-ative to the environment (Pursley, 2000). The
original 29-item questionnairedeveloped by researchers in Australia
(Fien, Yencken, & Sykes, 2002) wasadapted by the researchers
for use in a larger international study and for use inthis project
to be certain that is was culturally appropriate for a U.S.
audienceand to allow for the use of Scantron sheets during data
collection.
Dependent variable. Environmental behaviors were assessed using
a 10-item subscale of the Young People and the Environment survey.
Self-reported behavior is a continuous variable with a score
ranging from 0 to 20based on the sum of answers to 10 questions,
with 20 being the most environ-mental action and 0 being no
environmental action. Questions addressedwhether students have
engaged in or would engage in any environmentalactions (e.g.,
choosing household products that are better for the environ-ment).
Responses included, I have done this in the last 12 months, I
would
518 ENVIRONMENT AND BEHAVIOR / July 2005
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
consider doing this, and I would not consider doing this. Items
were codedas 2 (have done), 1 (consider doing), and 0 (not consider
doing). Cronbachsalpha for this subscale was .78 with scores
ranging from 0 to 20 and a mean of11.25 (skewness = .58, kurtosis =
.42).
Independent variable. Environmental attitudes were assessed
using anearly version of the New Environmental Paradigm (NEP). This
instrument,reported in Olsen, Lodwick, and Dunlap (1992), adapts
Dunlap and VanLieres (1978) original NEP scale to a bipolar format.
The NEP scale used inthis project is a 12-item scale addressing
both environmental and technologi-cal issues relative to the
environment. Students were asked to rank where theyfall on a
5-point scale that measured the degree to which they were more
envi-ronmental or less environmental. For each question, students
indicated theirdegree of agreement to a pair of opposing
statements. An example of an envi-ronmental statement contrasts the
view that people should adapt to the envi-ronment whenever possible
with the view that the environment should bechanged to meet peoples
needs. Students indicated that they strongly agreeor agree with the
first statement, are undecided, or agree or strongly agreewith the
contrasting statement.
The dimensionality of the 12 items from the NEP measure were
analyzedusing factor analysis. Discussions with the creator of the
NEP (R. Dunlap,personal communication, March 30, 2000) indicated
the possibility that themeasure was not unidimensional and would
break into two factors: an envi-ronmental factor and a technology
factor. Two factors were forced androtated using a varimax rotation
procedure. The rotated solution confirmedthat the measure broke
into the two previously predicted factors: technologyattitudes and
environmental attitudes. The environmental attitudes
factoraccounted for 24.6% of the variance, and the technology
factor accounted for14.8% of the item variance. Two of the items
did not clearly load on one fac-tor over the other, leading the
primary researcher to use the factor score insubsequent analyses
(Tabachnick & Fidell, 1989) rather than to split the mea-sure
into separate subscales using raw data only. After factor scores
wereobtained, a theoretical decision was made to utilize only the
environmentalfactor score and to drop the technology factor score
from the analysesbecause those four items used to create that score
had limited usefulness tothe construct of environmental attitudes.
The environmental factor score wasused in all subsequent
analyses.
Moderating independent variables. Self-efficacy was assessed
using asingle 10-item scale. Total self-efficacy is a combination
of the 7-item PearlinMastery Scale (Pearlin, Menaghan, Lieberman,
& Mullan, 1981) and three
Meinhold, Malkus / ADOLESCENT ENVIRONMENTAL BEHAVIORS 519
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
environmental self-efficacy questions that were created by the
researchersfor use in this study. Total self-efficacy is a
continuous variable with scoresranging from 1 to 5 based on the
average of answers to 10 questions. Partici-pants responded to
statements on a 5-point Likert-type scale ranging fromstrongly
agree to strongly disagree with a middle option of undecided.
Itemswere coded from 5 (most self-efficacy) to 1 (least
self-efficacy). Low scoresrepresent low self-efficacy, whereas high
scores represent high self-efficacy.Cronbachs alpha for this scale
was .78. Scores for this scale ranged from1.50 to 5.00, with a mean
of 3.62 (skewness = .36, kurtosis = .08).
Environmental knowledge was assessed using a combined score from
twoknowledge subscales: actual knowledge and familiar knowledge.
Actualknowledge is a continuous variable with a score ranging from
0 to 11 basedon the sum of correct answers to 11 questions given in
the form of a multiple-choice environmental test. Knowledge
familiarity was also a continuousvariable with a score ranging from
0 to 11 based on the sum of responses to 11questions. Students were
asked to identify how familiar they are with theenvironmental terms
provided. Both familiar knowledge and actual knowl-edge were
combined to create a 22-item environmental knowledge measure.A sum
score was calculated by adding the overall scores of the two
subscalestogether. Cronbachs alpha for the combined knowledge scale
was .69.Scores for this scale ranged from 3 to 22, with a mean of
13.79 (skewness = .15, kurtosis = .31).
Demographic and control variables. Sex was a categorical
variable with 0being male and 1 being female. Question two of the
survey asked to whatracial or ethnic group the adolescent belongs.
The race variable is categoricalwith answers being coded 1 through
5: 1 = Asian, 2 = Black, 3 = Caucasian,4 = Hispanic, and 5 = Other.
Socioeconomic status (SES) is a continuousvariable ranging from 1
to 5 based on the average of answers on two ques-tions. Questions
three and four ask students their mother and fathers highestlevel
of education. By combining these questions, researchers were able
toidentify a more accurate level of SES for each student. This
continuous vari-able ranged from 1 to 5, with 5 being the highest
level of SES and 1 being thelowest. Father and mothers education
levels were used as a proxy measurefor SES (as per Miller, 1991).
Those students who had either father ormothers education missing
were given a score based on the education of theremaining parent.
Scores for this two-item measure ranged from 1 to 5, with amean of
3.72. Perceived grade point average was a continuous variable
rang-ing from 1 to 5 based on the answer to a single question that
identified the per-ceived grades a student received in school, with
5 being mostly As and 1being mostly below Ds. Scores ranged from 1
to 5, with a mean of 3.35.
520 ENVIRONMENT AND BEHAVIOR / July 2005
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
RESULTS
This study examined adolescents environmental attitudes in
relation totheir self-reported environmental behaviors. The
moderating effects thatself-efficacy and environmental knowledge
have upon the relationshipbetween proenvironmental attitudes and
behaviors were also addressed.Descriptive statistics were run
followed by a series of hierarchical regres-sions. A correlation
matrix for all relevant variables for the total sample canbe seen
in Table 1.
The first hypothesis addressed the relationship adolescents
environmen-tal attitudes have on adolescents environmental
behaviors. It was predictedthat those adolescents who demonstrate
more proenvironmental attitudes aremore likely to demonstrate
proenvironmental behaviors. A hierarchal regres-sion was run,
controlling for gender (see Table 2). Findings showed
pro-environmental attitudes explained 22% of the variance, with an
R2 of .22 andan adjusted R2 of .22. There was a significant linear
relationship betweenproenvironmental attitudes and adolescents
environmental behaviors, F(2,836) = 116.68, p < .001.
The second hypothesis predicted adolescents who demonstrate
moreproenvironmental attitudes and high self-efficacy would have
more proenvi-ronmental behaviors than would adolescents who have
lower proenviron-
Meinhold, Malkus / ADOLESCENT ENVIRONMENTAL BEHAVIORS 521
TABLE 1Environmental Variables, Self-Efficacy, and Control
Variables: Correlations and
Descriptive Statistics
Variables 1 2 3 4 5 6 7
1. Self-efficacy .2. Environmental attitudes .23*** .3.
Environmental knowledge .16*** .18*** .4. Environmental behavior
.30*** .45*** .34*** .5. Gendera .13*** .11*** .08* .17*** .6.
Socioeconomic status (SES)b .06 .11** .23*** .08* .04 .7. Perceived
grade pointc .24*** .16*** .23*** .13*** .09** .25***M 3.6 . 16.6
11.3 0.55 3.7 3.4SD 0.62 . 4.7 3.6 0.50 0.90 0.69Range 1-5 . 0-32
0-20 0-1 1-5 1-5 .78 . .69 .78 . . .NOTE: N = 847.a. Gender is
coded 0 for male and 1 for female.b. SES is coded 1 for lowest SES
and 5 for highest SES.c. Perceived grade point is coded 1 for
mostly below Ds, 2 for mostly Ds, 3 for mostly Cs, 4 for mostlyBs,
and 5 for mostly As.*p < .05. **p < .01. ***p < .001.
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
mental attitudes and lower self-efficacy. A hierarchical
regression was run,controlling for gender, to evaluate the
moderating effect self-efficacy has onthe attitude-behavior
relationship (see Table 3). For model 1, the R2 was .03,and the
adjusted R2 was .03. Model 2 produced an R2 of .26 and an adjusted
R2of .26. The model itself was significant, F(2, 814) = 94.55, p
< .001. Model 3produced an R2 of .26 and an adjusted R2 of .26.
The model was also signifi-cant, F(1, 813) = 71.07, p < .001.
However, the interaction term added littleto the explained variance
of the model 2. Total self-efficacy was not a moder-ator of the
environmental attitudes-behavior relationship, indicating
thatadolescents who demonstrate high environmental attitudes and
greater
522 ENVIRONMENT AND BEHAVIOR / July 2005
TABLE 2Summary of Hierarchical Regression Analysis for Gender
and Environmental
Attitudes Variables Predicting Adolescents Environmental
Behaviors
Model 1 Model 2
Variable B SE B B SE B
Gender 1.25 0.24 0.17*** 0.91 0.22 0.13***Environmental
attitudes 1.56 0.11 0.44***R 2 0.03 0.22F for change in R 2 26.22
200.88NOTE: N = 839.*p < .05. **p < .01. ***p < .001.
TABLE 3Summary of Hierarchical Regression Analysis for Gender,
EnvironmentalAttitudes, Self-Efficacy, and Attitudes Self-Efficacy
Variables Predicting
Adolescents Environmental Behaviors
Model 1 Model 2 Model 3
Variable B SE B B SE B B SE B
Gender 1.29 0.25 0.18*** 0.79 0.22 0.11*** 0.79 0.22
0.11***Environmental
attitudes 1.40 0.11 0.39*** 0.94 0.56 0.27Self-efficacy 1.15
0.18 0.20*** 1.16 0.18 0.20***Attitudes
self-efficacy 0.13 0.16 0.13R 2 0.03 0.26 0.26F for change
in R 2 27.08 124.21 0.69NOTE: N = 818.*p < .05. **p < .01.
***p < .001.
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
amounts of self-efficacy will not always demonstrate greater
proenviron-mental behaviors.
Although the hypothesis was not supported, there is an apparent
relation-ship between adolescent environmental behaviors and
self-efficacy seen inthe model. A possible explanation is
self-efficacy is its own independent vari-able and not a moderating
variable. Correlations for environmental attitudesand behaviors by
sex and by high and low self-efficacy can be seen in Table 4.With
the exception of the female subsample, the correlations between
envi-ronmental attitudes and behaviors are similar among high and
low self-effi-cacy. These correlations indicate that although
self-efficacy does not appearto play a role in the environmental
attitudes-behavior relationship for males,it does appear to
moderate the relationship for females.
The third hypothesis predicted that adolescents who demonstrate
moreproenvironmental attitudes and high environmental knowledge
would reportmore proenvironmental behaviors than would adolescents
who demonstratelower proenvironmental attitudes and less
environmental knowledge. A hier-archical regression, controlling
for sex, was again conducted to evaluate themoderating effect
knowledge has on the attitude-behavior relationship (seeTable
5).
In model 2, the R2 was .29, and the adjusted R2 was .28,
explaining 28% ofthe variance. Model 2 tells us that
proenvironmental attitudes are signifi-cantly related to
proenvironmental behaviors, F(3, 808) = 107.91, p < .001.For
model 3, the R2 was .30, and the adjusted R2 was .29, explaining
29% ofthe variance. The third hypothesis was supported, F(4, 807) =
85.59, p < .001.Environmental knowledge is a moderator of the
environmental attitudes-behavior relationship, indicating that
adolescents who demonstrate moreproenvironmental attitudes and
greater environmental knowledge will reportgreater amounts of
proenvironmental behaviors.
Meinhold, Malkus / ADOLESCENT ENVIRONMENTAL BEHAVIORS 523
TABLE 4Correlations Among Environmental Attitudes and
Behaviors
for High and Low Self-Efficacy Groups, by Gender
Environmental Behaviors
Variable High Self-Efficacy Low Self-Efficacy
Environmental attitudesTotal sample .50*** .38***
Males .34* .38***Females .60*** .28*
*p < .05. ***p < .001.
at Universiti Teknologi Malaysia on April 8,
2015eab.sagepub.comDownloaded from
-
To illustrate the interaction and significance of the third
hypothesis, a nor-mal distribution was run on environmental
knowledge, and students wereplaced into high, medium, or low
knowledge groups. Using this normal dis-tribution, researchers
selected those adolescents who fell more than 1 stan-dard deviation
above the mean (> 17.12) as the high environmental knowl-edge
group, those falling less than 1 standard deviation below the mean
(