environmental attitude

Global environmental problems of shrinking natural resources, pollution and population growth challenge the ways people live. As with many other disciplines, psychology attempts to develop human societies less exploitive in their use of the earth’s natural resources cf. Stern, 1992a; Kruse, 1995.. Because psychologists refer to individual behaviour rather than to behaviour of whole societies they ask questions such as what determines an individual’s ecological behaviour wi.e. ‘actions which contribute towards environmental preservation andror conservation.’ Axelrod & Lehman, 1993: p. 153..x or how can behaviour be changed in a more ecological direction. In answering these questions, environmental attitude is considered one of the most promising concepts Newhouse, 1990.. In fact, almost two-thirds of all environmental psychological publications include environmental attitude in one way or another.2 Not surprisingly, the relationship between environmental attitude and ecological behaviour is well-explored.3 However, the relationship appears to be at best moderate across different studies Hines et al., 1986r87.. This lack of a stronger correlation occasionally results in rather pessimistic views of the usefulness of environmental attitude as a predictor of ecological behaviour4 Stern, 1978; Lloyd, 1980.. The present paper proposes three reasons, one theoretical and two methodological, that affect the predictive power of environmental attitude concepts. From a theoretical point of view, there are at least three main research traditions that use quite different attitude concepts. The differences confuse the comparison of research results in the ecological
domain. The two methodological flaws that affect any attitude]behaviour relationship also affect the
environmental attitude and ecological behaviour relationship. These two flaws are the lack of measurement correspondence and the lack of consideration of situational influences on a given behaviour. Measurement correspondence refers to measurement of
attitude and behaviour on the same level of specificity Ajzen & Fishbein, 1977.. Because of the multitude of situational influences, the level of specificity should be rather general. Situational influences refer to constraints and facilities on behaviour beyond people’s control Ajzen & Madden, 1986.. Inclusions of such behaviour influences are seen as particularly important in the ecological
domain Hines et al., 1986r87; Granzin & Olsen, 1991; Stern, 1992b; Vining& Ebreo, 1992; Foppa et al., 1995; Guagnano et al., 1995.. Such influences are usually considered either as moderator effects on the relationship between environmental attitude and ecological behaviour or as direct influences on ecological behaviour. Both approaches require a rather arbitrary selection of possible influences. The present paper promotes the theory of planned behaviour Ajzen, 1985. as an overall theoretical framework in the ecological domain. Moreover, both of the methodological shortcomings can be overcome by using a probabilistic measurement approach for the assessment of ecological behaviour.

Three environment attitude
In essence, two types of environmental attitude5
are used to predict ecological behaviour: 1. atti-
tudes toward the environment, and 2. attitudes
toward ecological behaviour Hines et al., 1986r87;
the same is proposed for energy conservation by
Olsen, 1981.. Either the object of one’s attitude is
the natural environment itself or some aspects of it
e.g. air quality. or the attitude object is ecological
behaviour e.g. recycling or political activism.. Environmental
attitude towards ecological behaviour
refers to the Fishbein and Ajzen tradition of attitude
research that will be described in more detail
later. Only a minority of the studies approximately
20%, according to Hines et al., 1986r87. that relate
environmental attitude with ecological behaviour
refer to the framework of the theory of reasoned
action Ajzen & Fishbein, 1980. and its developed
version, the theory of planned behaviour Ajzen,
1985.. In contrast, attitude towards the environment
commonly refers to environmental concern
Vining & Ebreo, 1992.. Environmental concern is
used either as a multiple or a single component
approach Fuhrer, 1995. and covers either environment
in general or some particular aspects of environment.
If attitude towards environment refers to a multiple
component approach, the distinction between
cognitive affective and intentional components of
attitude proposed by Rosenberg & Hovland, 1960.
is usually made. This research tradition of environmental
attitudes can be traced back to two studies
by Maloney and colleagues Maloney & Ward, 1973;
Maloney et al., 1975.. An affect scale measures the
affective component, factual knowledge about the
environmental measures the cognitive aspects, and
verbal commitment measures the behaviour inten-
tion component of environmental attitude Smythe
& Brook, 1980.. A fourth scale measures ecological
behaviour.6
Originally, all three environmental attitude com-
ponents}affect, knowledge i.e. cognition., and intention}
were used in parallel to predict ecological
behaviour. Recent versions of this approach vary:
some propose the affect component as the single
indicator of environmental attitude Langheine &
Lahmann, 1986; Newhouse, 1990., others abandon
ecological behaviour intention Dispoto, 1977., while
a third group uses the ecological behaviour intention
component as the single indicator of environ-
mental attitude Schahn & Holzer, 1990a, 1990b;
Auhagen & Neuberger, 1994..
Moreover, instead of using these environmental
attitude components in parallel, some approaches
use the concepts knowledge, affect and intention.
in a more sequential way to predict either environ-
mental attitude or ecological behaviour Geller,
1981; Diekmann & Preisend¨orfer, 1992; Grob, 1995..
Consequently, environmental attitude is, occasionally,
measured independently from its cognitive,
affective and intentional components. Hence, one’s
attitude towards the environment can become a
single component measure Arbuthnot, 1977; Van
der Pligt, 1985; Oskamp et al., 1991; Lansana,
1992; Derksen& Gartrell, 1993; Gamba& Oskamp,
1994.. If attitude towards the environment refers to
a single component approach, this attitude can be
predicted by knowledge, affect and intention as
already mentioned. Occasionally, however, environmental
attitude is measured by knowledge, affect
and intention items Sia et al., 1985r86; Berger &
Corbin, 1992; Axelrod & Lehman, 1993..
The new environmental paradigm NEP., which
is the third and most recently developed tradition
of environmental attitude research Dunlap & Van
Liere, 1978; Stern et al., 1993; Scott & Willits,
1994., is an alternative, single component measure
of environmental attitude. Some question its unidimensionality
and use it instead as a multiple component
measure consisting of dimensions such as
balance of nature, limits of growth and humans
over nature cf. Vining & Ebreo, 1992.. Because
proponents of this tradition regard one’s moral values
as the core concept of environmental attitude Stern et al., 1993. it may be argued that the NEP
represents a shift towards a more evaluative con-
ception of attitude Schahn & Holzer, 1990a; or cf.
Dunlap & Van Liere, 1978; such a shift can be seen
in other attitude concepts as well: Leonard-Barton,
1981; Van Liere & Dunlap, 1981; Verhallen & Van
Raaij, 1981; Axelrod & Lehman, 1993.. This interpretation
is additionally supported by the fact that
NEP findings barely match those regarding the
relationship between environmental attitude and
ecological behaviour. In short, the strength of the
relationship between the NEP and ecological be-
haviour ranges from nonexistent Smith et al., 1994.
to weak Dunlap & Van Liere, 1978; Scott & Willits,
1994.. Conversely, environmental attitude and ecological
behaviour appear to be at least moderately
related Hines et al., 1986r87..
Because the empirical findings of the field will be
presented in regard to a general theoretical framework,
the following section introduces the theory of
planned behaviour as this frame. The proposed
frame encompasses most aspects of the three formerly
mentioned attitude concepts. A general framework: the theory of
planned behaviour
In the theory of reasoned action Ajzen & Fishbein,
1980. and its developed version, the theory of
planned behaviour Ajzen, 1985., behaviour intention
to perform the behaviour in question is the
immediate antecedent of overt behaviour. Intention,
in turn, is seen as a function of one’s attitude
towards performing a particular act and one’s sub-
jective norms i.e. the perception of the expectations
of relevant others.. Because attitude includes not
just the evaluation of a certain outcome but also
the estimation of the likelihood of this outcome,
salient information or factual knowledge is a neces-
sary precondition for any attitude Stutzman &
Green, 1982..7 As subjective norms refer to the
strength of normative beliefs and the motivation to
comply with these beliefs, social and moral values
i.e. what one should do from a normative stance,
social expectations as well as moral principles, can
be considered as an approximation of one’s subjective
norms.
In Figure 1, the theory of reasoned action is
presented graphically. The theory of planned behaviour
extends the theory of reasoned action by its
inclusion of influences on behaviour beyond people’s
control. If these influences are measured by means
of the perception of one’s control, two assumptions
FIGURE 1. The theory of reasoned action.
have to be made: 1. the predicted behaviour must
be, at least partially, beyond volitional control, and
2. perception of control must reflect actual control
upon behaviour with some accuracy Ajzen &
Madden, 1986.. While the latter assumption has to
be seen as a possible flaw of the planned behaviour
approach, the former assumption is often claimed
in the ecological domain.
Ecological behaviour appears to be susceptible to
a wide range of influences beyond one’s control
Hines et al., 1986r87.. Outside temperature Ol-
sen, 1981. and home characteristics Verhallen &
Van Raaij, 1981., for instance, affect energy consumption;
cost of water affects water conservation
Moore et al., 1994., and the number of people in a
given household Gamba & Oskamp, 1994., house
ownership Lansana, 1992., s torage space Wil-
liams, 1991. and type of residence Oskamp et al.,
1991. affect recycling behaviour. Examples of community
or neighbourhood-related influences include
political measures that support public transportation
systems that provide an alternative to commuting
by automobiles, or political measures that
facilitate recycling or force people to pay for garbage
disposal, which further reduces waste generation
and promotes recycling. In short, socio-cultural constraints
determine, to some extent, which ecological
behaviour is easier to carry out and which is harder.
As a consequence, people appear to behave inconsistently,
since even someone who claims to be
ecologically oriented may behave ecologically in one
domain and unecologically in another cf. Oskamp
et al., 1991; Vining & Ebreo, 1992; Pickett et al.,
1993; Scott & Willits, 1994.. Thus, not surprisingly,
the theory of planned behaviour, which includes
behaviour constraints beyond volitional control, has
to be considered especially useful in predicting ecological
behaviour. Because the theory of reasoned
action does not include such constraints, previous
research may have been affected by neglecting
socio-cultural constraints cf. Stutzman & Green,
1982.. However, selection of possible socio-cultural constraints remains a challenging problem; possible
ways of meeting this challenge are discussed later.
Environmental attitude, factual
knowledge, values, intention and
ecological behaviour
To include all three attitude concepts i.e. attitudes
toward the environment, the new environmental
paradigm and attitudes toward ecological be-
haviour. in one general framework i.e. the theory
of planned behaviour., this framework has to consist
of at least three components: factual knowledge
about the environment, social and moral values
regarding environment, social and moral values regarding
environment and ecological behaviour intention.
The theory of reasoned action, as well as
the theory of planned behaviour, proposes that attitude
influences behaviour, mediated by intention
see Figure 1.. Factual knowledge can be seen as a
precondition of any attitude and, thus, the relationship
between factual knowledge and behaviour is
mediated by intention as well. Moveover, subjective
norms, or at least one’s values, are also mediated
by intention and therefore predict behaviour indirectly.
Given these interrelations, research findings
in the ecological domain fit together quite well.
Attitude effect
If environmental attitude is assessed by one single
measure regardless of the type of environmental
attitude, the usual findings reveal either a moderate
relationship between environmental attitude
and ecological behaviour Weigel et al., 1974;
Langeheine & Lehmann, 1986; Hines et al.,
1986r87; Axelrod & Lehman, 1993; Smith et
al., 1994. or a weak relationship McGuinness
et al., 1977; Sia et al., 1985r86; Williams, 1991;
Berger & Corbin, 1992; Diekmann & Preisend¨,orfer
1992; Barker et al., 1994; Moore et al., 1994; Grob,
1995.. However, at least five studies report no such
relationship at all Arbuthnot, 1977; Van der Pligt,
1985; Oskamp et al., 1991; Lansana, 1992; Gamba
& Oskamp; 1994. and one study yields a strong
association between environmental attitude and
ecological behaviour Lynne & Rola, 1988.. If environmental
attitude refers to components, for
instance, environmental knowledge, environmental
values and ecological behaviour intention, the
following findings are reported.
Knowledge effect
Given that factual knowledge about the environment
is a precondition of one’s environmental attitude,
8 this knowledge should not be related with
ecological behaviour strongly because its influence
is attenuated both by environmental attitude and
ecological behaviour intention. Hence, it is not surprising
that several studies found either no relationship
between factual environmental knowledge
and ecological behaviour Maloney & Ward, 1973;
Maloney et al., 1975; Amelang et al., 1977; Schahn
& Holzer, 1990a, 1990b. or at best a moderate
relationship Arbuthnot, 1977; Dispoto, 1977;
Smythe & Brook, 1980; Stutzman & Green, 1982;
Hines et al., 1986r87; Oskamp et al., 1991.. When
this relationship appears to be stronger, it is knowl-
edge about an ecological behaviour i.e. knowledge
about what and how something can be done. rather
than factual knowledge about the environment that
is related to ecological behaviour e.g. Levenson,
1974; Sia et al., 1985r86; Smith-Sebasto& Fortner,
1994..
Value effect
As proposed by the theory of planned behaviour,
one’s subjective norms Olsen, 1981; Kantola et al.,
1983; Midden & Ritsema, 1983. and normative be-
liefs regarding environment McGuinness et al.,
1977; Stutzman & Green, 1982. affect his or her
intention to behave ecologically. However, this ef-
fect ranges from rather weak Midden & Ritsema,
1983. to fairly large McGuinn ess et al., 1977..
Furthermore, this relationship decreases if ecological
behaviour instead of ecological behaviour inten-
tion is considered McGuinness et al., 1977; Vining
& Ebreo, 1992., presumably indicating the mediating
effect of ecological behaviour intention. One’s
environmental values parallel these findings:
environmental values are related to ecological
behaviour intention Dunlap & Van Liere, 1978;
Van Liere & Dunlap, 1981; Axelrod, 1994.; and if
environmental values are related to ecological be-
haviour Dunlap & Van Liere, 1978; Van Liere &
Dunlap, 1981; Verhallen & Van Raaij, 1981; Stern
et al., 1993; Grob, 1995. they are presumably mediated
by a third variable. According to the theory of
planned behaviour, this mediator is ecological behaviour
intention.
Intention effect
The most striking effect usually found is between
ecological behaviour intention and ecological behaviour. Ecological behaviour intention is strongly
related to ecological behaviour Maloney & Ward,
1973; Maloney et al., 1975; Schahn&Holzer, 1990a,
1990b; Lansana, 1992; Auhagen & Neuberger,
1994. or at worst moderately related Smythe &
Brook, 1980; Stutzman & Green, 1982; Hines et al.,
1986r87; Moore et al., 1994; Diekmann & Franzen,
1995.. Unfortunately, there are some types of ecological
behaviour with which no such relationship
is found Auhagen & Neuberger, 1994; Fuhrer &
W¨olfing, 1997. and at least two studies in which the
relationship between ecological behaviour intention
and ecological behaviour appears to be small
McGuinness et al., 1977; Van Liere & Dunlap,
1981.. Note that it is not uncommon in the ecological
domain that one type of ecological behaviour is
affected by either environmental attitude, environmental
knowledge, environmental values or ecologi-
cal behaviour intention while others are not Lan-
geheine & Lehmann, 1986; Berger & Corbin, 1992..
One recommendation for dealing with this sort of
finding refers to measurement correspondence,
which means measuring related concepts on the
same level of specificity. Specific environmental attitude
measures are better predictors of specific
rather than general ecological behaviour measures
Weigel et al., 1974; McGuinness et al., 1977; Van
der Pligt, 1985; Smith et al., 1994.. However, specific
measures appear to be more strongly affected
by situational influences than general ones, which,
in turn, makes findings from different domains
hardly comparable. This has some important
methodological implications and consequences for
the ecological domain.
Methodological considerations
Two things have to be considered when dealing
with the relationship between environmental attitude
and ecological behaviour: measurement correspondence
and behaviour influences beyond people’s
control.
Measurement correspondence: general attitude and
general behaviour
The possible lack of measurement correspondence
between environmental attitude and ecological be-
haviour is well recognized Weigel et al., 1974;
Newhouse, 1990; Stern, 1992b; Vining & Ebreo,
1992; Axelrod & Lehman, 1993. and does not need
much further explanation. It can be summarized as
follows: if one’s environmental attitude is assessed
generally, ‘the behavioural criterion should be
equally general or comprehensive’ Weigel et al.,
1974: p. 728.. Note, however, that highly specific
rather than general measures of ecological
behaviour, even though corresponding with
environmental attitude, are occasionally refused as
a solution because they are highly susceptible to
situational influences beyond people’s control
Granzin & Olsen, 1991; Pickett et al., 1993.. As
specific measures appear to be affected more easily
than general measures, general environmental attitude
measures are proposed as better predictors of
comprehensive ecological behaviour criteria
Newhouse, 1990.. Even though some data
apparently confirm this notion by strong relationships
between general environmental attitude and
general ecological behaviour measures Lynne &
Rola, 1988; Axelrod & Lehman, 1993. others do not
McGuinness et al., 1977; Van Liere & Dunlap,
1981; Berger & Corbin, 1992; Diekmann &
Preisend¨orfer, 1992; Scott & Willits, 1994; Smith et
al., 1994; Grob, 1995.. These puzzling findings concerning
the relationship between general environmental
attitude and general ecological behaviour
may also be due to measurement problems related
to general ecological behaviour measures for a dis-
cussion see Kaiser, 1998.. Occasionally, such a gen-
eral measure is questioned in principle Lloyd, 1980;
Oskamp et al., 1991.. However, there is at least one
general measurement approach that includes a
broad range of different behaviours, which rules
out situational influences beyond people’s control
see Kaiser, 1998..
Consideration of influences beyond people’s control
As previously stated, the relationship between environmental
attitude and ecological behaviour may
be affected by influences beyond people’s volitional
control. Thus, situational factors ‘ . . . such as economic
constraints, social pressures and opportuni-
ties to choose different actions . . . ’ Hines et al.,
1986r87: p. 7. may interfere with one’s attitude.
For instance, recycling opportunities affect the
amount of recycling behaviour Williams, 1991;
Vining & Ebreo, 1992; Guagnano et al., 1995.. Such
situational influences can be considered in at least
three different ways. First, perceived control, pro-
posed by Ajzen and Madden 1986. as an indicator
of actual control, can be used as a predictor of
ecological behaviour. Second, moderators of the
relationship between environmental attitude and
ecological behaviour may be scrutinized. Because
moderators represent conditional aspects of a given
relationship, nonvolitional behaviour constraints that affect such a relationship can be chosen as
moderators e.g. residential area or season.. Third,
an ecological behaviour measure, established as a
Rasch-scale, that quantifies ecological behaviour
difficulties can be used as the outcome measure.
Perceived control. In the ecological domain, differ-
ent concepts of perceived control Levenson, 1974;
Berger & Corbin, 1992; Auhagen & Neuberger,
1994; Kals & Montada, 1994; Grob, 1995. are used;
for instance, internal locus of control Arbuthnot,
1977; Huebner & Lipsey, 1981; Sherman et al.,
1981; Sia et al., 1985r86; Hines et al., 1986r87;
Oskamp et al., 1991; Gamba & Oskamp, 1994;
Smith-Sebasto & Fortner, 1994., self efficacy
Kantola et al., 1983; Axelrod & Lehman, 1993. and
feelings of powerlessness Busch-Rossnagel &
Weigel, 1984.. None of them, however, indicate
people’s actual control as proposed by Ajzen &
Madden, 1986.. Rather, they represent different
predictors of either ecological behaviour Levenson,
1974; Arbuthnot, 1977; Busch-Rossnagel & Weigel,
1984; Sia et al., 1985r86; Hines et al., 1986r87;
Oskamp et al., 1991; Axelrod & Lehman, 1993;
Auhagen & Neuberger, 1994; Gamba & Oskamp,
1994; Smith-Sebasto & Fortner, 1994; Grob, 1995.
or ecological behaviour intention Huebner &
Lipsey, 1981; Sherman et al., 1981; Kantola et al.,
1983; Kals & Montada, 1994.. Unfortunately, the
relationship between perceived control and ecological
behaviour is inconsistent and ranges from
slightly negative Grob, 1995. to nonexistent
Oskamp et al., 1991; Gamba & Oskamp, 1994. to
very positive Auhagen & Neuberger, 1994..
Moderators. A selection of moderators used that
affect the environmental attitude]ecological be-
haviour relationship includes gender Schahn &
Holzer, 1990a, 1990b., socio-economic status Mid-
den & Ritsema, 1983., mode of behaviour assess-
ment Hines et al., 1986r87., group membership
environmentalists vs nonenvironmentalists: Hines
et al., 1986r87., income Lynne & Rola, 1988., ac-
cess to recycling programmes Derksen & Gartrell,
1993., season Becker et al., 1981. and nationality
Meseke, 1994.. All these moderators represent different
sorts of nonvolitional, socio-cultural behaviour
constraints. Usually, questions concerning
their scope remain unanswered: do they affect all or
just a few ecological behaviours?
Because moderator effects indicating situational,
socio-cultural influences. are either difficult to explain
or they demand further theoretical clarifica-
tion cf. Baron & Kenny, 1986. they raise usually
more questions than they answer and, hence, remain
somewhat problematic. Moreover, the arbitrariness
of their selection presumably affects
empirical findings as well. For instance, each
study uses some moderators while others are
neglected.
A general ecological behaviour measure
As the probability of a behaviour considers influences
beyond people’s actual control, an accurate
measure of ecological behaviour is actually a probability
that one carries out the specific behaviour
rather than anything else. Whether someone commutes
on a given morning or not may depend on
several factors beyond his or her control, for instance,
weather, traffic and availability of an automobile.
Moreover, all sorts of influences beyond
people’s control affect different behaviours in a way
that make them varyingly difficult to carry out.
Hence, some behaviours seem to be easier to carry
out than others. For example, recycling is easy to
carry out when recycling bins are readily accessible.
Therefore, influences beyond people’s control have
to be considered in two different ways by estimat-
ing the probability i.e. one’s tendency. of behaving
ecologically, as well as the probability of anyone
carrying out a certain behaviour i.e. behaviour
difficulty..
The General Ecological Behaviour GEB. scale
assesses general ecological behaviour by considering
different ecological and prosocial behaviours
see Appendix 1.. Each of these behaviours has a
given difficulty to be carried out, which, in turn,
represents an estimate of all the constraints beyond
people’s control. The easier a behaviour is to carry
out, the less constraints have to be assumed. This
behaviour difficulty is estimated for each behaviour
by considering the number of people who behave
correspondingly i.e. the probability that anyone
will behave that way regardless of his or her ten-
dency to behave ecologically.. One’s tendency to
behave ecologically is estimated by considering the
number of ecological behaviours he or she has car-
ried out i.e. the probability that somebody will
behave ecologically given that behaviours differ in
difficulties..
Because a measure of one’s ecological behaviour
considers the tendency to behave ecologically as
well as behaviour difficulties, people are free, to a
certain extent, to behave inconsistently across different
ecological behaviours. Someone, for instance,
who tends to behave ecologically on a very high newspapers, even though this behaviour is easy to
carry out. In contrast, someone who usually behaves
very unecologically may, for whatever reason,
not drive an automobile, a behaviour that is
commonly difficult not to carry out. And as such
inconsistencies result from, individually, different
socio-cultural constraints beyond people’s actual
control, situational influences are represented in
this behaviour measure in two different ways for
more details see Kaiser, 1998..
Hypotheses
As explained in the previous sections, three
shortcomings affect the predictive power of environmental
attitude concepts regarding ecological be-
haviour: 1. the lack of a unified attitude concept,
2. the lack of measurement correspondence between
attitude and behaviour on a general level,
and 3. the lack of considerations of situational
behaviour constraints beyond people’s control. The
present paper proposes to use an abbreviated version
of the theory of planned behaviour by Ajzen as
a unifying frame for environmental attitude concepts;
it also uses a probabilistic measurement approach
that may overcome the methodological
shortcomings.
Environmental knowledge, environmental values
and ecological behaviour intention are suggested as
the conceptual skeleton of the theory of planned
behaviour that encompasses the three most commonly
used attitude approaches in environmental
psychology: attitude towards the environment, attitude
towards ecological behaviour and the NEP. In
addition, the relationship between environment attitude
and ecological behaviour should be considerably
strengthened by adopting a general ecological
behaviour measure that assesses behaviour by
means of behaviour difficulties and behaviour tendencies.
This relationship between general attitude
and general behaviour should remain consistent
even across very ideologically distinct groups of
people. To test the latter, two known groups with
different ecologically relevant ideologies are scrutinized.
The present paper explores the following
predictions in further detail: 1. environmental
knowledge and environmental values are significant
preconditions of ecological behaviour intention
see Figure 1.; 2. ecological behaviour intention
affects ecological behaviour strongly if both of them
are assessed rather generally and if the ecological
behaviour measure considers situational behaviour
constraints; and 3. all relationships between the
three environmental attitude concepts i.e. environmental
knowledge, environmental values and eco-
logical behaviour intention. and between ecological
behaviour intention and ecological behaviour are
not moderated by ideology; rather they can be generalized
across ideologically heterogeneous groups.
Method
Participants and procedures
The present sample was constituted from an initial
pool of 3000 members fromeach of two Swiss transportation
associations. The associations can be differentiated
ideologically. One aims to promote a
transportation system that has as little negative
impact on humans and nature as possible, and the
other represents primarily automobile drivers’ interests.
To include as much diversity as possible,
the two associations were further stratified by pri-
mary language French, Italian, German. and type
of residential area city, suburb, village.. Of all
members of both associations, 27.4 per cent 1643.
were willing to participate. This presumably, even
for the two associations, unrepresentative pool was
asked to complete three questionnaires: the first
was sent out during December 1993, and 1371 peo-
ple 83.5%. compl eted it Seiler, 1994; Fuhrer et al.,
1995.. The targeted participants of the second questionnaire
were those who had completed the first
questionnaire. The second questionnaire was mailed
in May 1994, and 1189 86.7%. of those who completed
the first questionnaire participated in this
second study. The present, third study, was undertaken
during November 1994, and targeted only
the German speaking subgroup fromthe first study.
Note that the German speaking subgroup of the
first and second studies numbers 579 42.2% of the
total sample. and 438 36.8%., respectively. After
36 people declined further participation, 543
93.8%. of the German speaking participants in the
first study remained to be surveyed in the third
study. Of these 445 82.0%. returned completed
questionnaires. Participants’ 62.5% male. median
age was 45.5 years, Ms46.6, ranges20]82.
The high participation rate within the pool can
be seen as a result of a self-selection process of
more ecologically-concerned participants. Members
of the automobile drivers’ association were less
well-represented in the sample 25.8%. in contrast
to members of the association promoting a more
ecological transportation system 74.2%.. Hence, the
sample seems to be biased towards more ecologically
concerned participants. For the purpose of the present study, it is sufficient that the participants
reflect a wide range of diversity, as for instance, in
ecological concern. Since relationships and not
means are tested, any sample bias is of minor
importance. Additionally, the generalizability of the
proposed relationships will be scrutinized by statistical
means, which is basically what fit statistics
are all about.
Measures
The questionnaires consisted of a Social Desirability
scale, a General Ecological Behaviour measure,
and three scales that represent the environmental
attitude related concepts.
The Social Desirability SD. scale presented by
Amelang and Bartussek 1970. consists of 32 items.9
Fourteen items had to elicite a ‘yes’ response e.g. ‘I
never claim to know more than I actually do’. and
18 items a ‘no’ response e.g. ‘I have taken advan-
tage of people in the past’. to contribute to the SD
sum score. To be consistent with the response options
for the ecological behaviour items, the original
truerfalse format was changed to a yesrno format.
Missing values ns109; 0.8%. were treated as if
participants answered in a nonsocially desired way.
The General Ecological Behaviour GEB. measure
consists of 38 items representing different
types of ecological behaviour and some nonenviron-
mental, prosocial behaviours as well see Appendix
1.. A yesrno response format for these items was
used. Negatively formulated items were reversed in
coding. Missing values ns80; 0.5%. were handled
as ‘no’ responses in general assuming missing values
represented participants’ doubt, an indicator of
not behaving alike in general.. The GEB measure
has been calibrated as an unidimensional Rasch-
scale Kaiser, 1998. based on item response theory
Wright & Masters, 1982.. Additionally, all
attempts to validate the GEB measure with
criterion-related self-reported data and with ob-
served behavioural data were promising see Kaiser,
1998.. The GEB items and the 32 SD items were
distributed randomly throughout the questionnaire.
Twenty-eight items, which were used to establish
the three environmental attitude related scales
during the first study Seiler, 1994; Fuhrer et al.,
1995; Fuhrer & W¨olfing, 1997., were re-evaluated.
These scales are Environmental Knowledge EK.,
Environmental Values EV. and Ecological Be-
haviour Intention EBI.. A 5-point Likert scale that
ranged from agree totally 1. t o disagree 5. was the
response format used. It is worth noting that only
three out of 28 items are negatively turned, which
puts these scales at a certain risk to be biased i.e.
acquiescence response set.. All 10 knowledge items,
which represent the EK scale, were distributed
randomly throughout a total of 24 knowledge items.
All seven value items, which represent the EV
scale, were distributed randomly throughout a total
of 13 value items. All 11 intention items were
grouped together in the questionnaire. Negatively
formulated items were reversed in coding. A
principal-factor analysis PFA. was performed to
confirm the three-factor structure of the first study
see Table 1.. Communality estimates were iteratively
derived using the highest correlation of each
variable with any other variable as a starting value.
The final solution was varimax rotated. Three hundred
and ninety-one participants remained in the
analysis; fifty-four people were excluded because of
missing values. Twenty-eight items with a total of
49.1 per cent explainable variance remained in the
analysis. The final three-factor solution accounted
for 74.9 per cent of this remaining variance. Factor
loadings of the varimax rotated final solution can
be seen in Table 1.
After rotation, the explained variance was attributable
to each of the three factors as follows:
EKs31.9 per cent, EVs24.4 per cent and EBIs
43.7 per cent. The three factors either correlated
nonsignificantly p ) 0.05. or correlated only
marginally R2s1.2%., though si gnificantly p-
0.05.: r EK] EV. s 0 .11, r EK] EBI. s 0.08,
r EV]EBI.s0.09. Note that these rather weak or
nonexistent relationships between EK, EV and EBI
derive from applying a varimax rotation. They confirm
that the three scales measure statistically
independent constructs which is required to reason-
ably check their empirical interrelations see Re-
sults.. Generally, the content of all 28 items is
related to the topic of pollution see Table 1.. The
internal consistencies of the three factors in the
solution were estimated by using Cronbach’s alpha:
a EK.s0.84 ns418., a EV.s0.73 ns425. and
a EBI.s0. 85 ns 423.. For subsequent analyses,
scores for EK, EV and EBI were obtained by taking
the mean of the constituent items. Mean values
were calculated only if participants had answered
at least half of the items for each factor. The correlations
between factor scores and mean values of
factors indicate the latter are useful approxima-
tions of the former: r EK.s0 .89, r EV.s0.90,
r EBI.s0.95, ns391. By using mean values
w n EK.s4 41, n EV.s4 40, n EBI.s4 42x instead
of factor scores ns391., data for additional participants
could be included in further analyses.