Towards a psychology of solar energy: Analyzing the effects of the Big Five personality traits on household solar energy adoption in Germany

Abstract

This research paper investigated the effect of consumers’ Big Five personality traits on the adoption of residential photovoltaic systems in Germany. To account for different types or groups of households, a multigroup structural equation model with N = 9,281 individuals was analyzed using data from a nationwide, representative household panel. It could be shown that the ways in which personality traits are mediated through environmental concern and risk propensities change depending on whether there is a single household or if additional individuals are involved in the decision-making process. In the aggregated view, direct effects of extraversion could be found for households comprising only a couple. For other households with additional members, no direct effects were found. All five personality traits were mediated by risk preference while openness, agreeableness, and neuroticism were mediated by environmental concern. On the individual level, the examination revealed that the head of household’s neuroticism and the partner’s openness and extraversion showed significant effects on the purchase of a photovoltaic system – albeit with small effect sizes. The results provide important insights into how household decisions can be better understood in order to contribute to the energy-system transformation.

Full text

Energy Research & Social Science 77 (2021) 102087
2214-6296/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Towards a psychology of solar energy: Analyzing the effects of the Big Five
personality traits on household solar energy adoption in Germany
Stefan Poier
University of Gdansk, Faculty of Economics, Armii Krajowej 119/121, 81-824 Sopot, Poland
ARTICLE INFO
Keywords:
Big Five
Renewable energy
Structural equation modeling
Consumer behavior
Household decision-making
ABSTRACT
This research paper investigated the effect of consumers’ Big Five personality traits on the adoption of residential
photovoltaic systems in Germany. To account for different types or groups of households, a multigroup structural
equation model with N =9,281 individuals was analyzed using data from a nationwide, representative house-
hold panel. It could be shown that the ways in which personality traits are mediated through environmental
concern and risk propensities change depending on whether there is a single household or if additional in-
dividuals are involved in the decision-making process. In the aggregated view, direct effects of extraversion could
be found for households comprising only a couple. For other households with additional members, no direct
effects were found. All ve personality traits were mediated by risk preference while openness, agreeableness,
and neuroticism were mediated by environmental concern. On the individual level, the examination revealed
that the head of household’s neuroticism and the partner’s openness and extraversion showed signicant effects
on the purchase of a photovoltaic system – albeit with small effect sizes. The results provide important insights
into how household decisions can be better understood in order to contribute to the energy-system
transformation.
1. Introduction
In Germany, the decision to phase out coal-red power generation was
made in 2019 after the abandonment of nuclear power had been ensured
as a consequence of Japan’s Fukushima Daiichi nuclear disaster in 2011.
But the irregular distribution of locations with renewable electricity
generation in the north of Germany and high-energy demand in the south,
with the associated need for power lines across the Republic, has resulted
in an increasing number of citizens’ comments not only for but also
opposed to power lines and wind farms. While large quantities of coal in
central Germany continue to be converted into electricity, this blocks
existing transit lines from the north to the south. To prevent supply fail-
ures without importing energy sources, a country with few natural energy
resources must push for the expansion of renewable energies.
One option would be for all home builders to choose to generate their own
electricity – for example, with a photovoltaic system. However, the lowest
proportion of newly built houses, by far, include a rooftop PV plant.
Consequently, there is a need to understand additional factors that inuence
the public’s decision to adopt residential photovoltaic systems. For many
years, economic researchers used theories of utility maximization to model
consumer behavior [1,2]; however, these models were normative rather
than descriptive. In the past few decades, cognitive psychology has taken a
much more differentiated view of purchase behavior and revealed the
concept of the “homo oeconomicus” as an illusion [3].
The investment in a PV system can be considered from different
perspectives. Since the purchase of a PV system is not only an acquisition
of a consumer good, but also a large investment under uncertainty, the
willingness of the household members to take risks must be taken into
account in addition to the nancial resources of the household when
making the purchase decision. Furthermore, the willingness to protect
the environment also plays a role in the purchase, because the self-
generation of electricity requires less energy from exhaustible sources
and thus protects the environment. These concerns argue for the
consideration of risk preference and environmental concern as de-
terminants of the purchase decision. Informational searches and evalu-
ations are determined by one’s level of involvement and cognitive
abilities; evaluations also depend on personal preferences and cultural
inuences [4]. Even the perception of risks for the household’s health or
nancial situation is shaped by personality disposition [5]. In the last
few decades, many studies focused on the inuence of consumers’ in-
dividual dispositions on their decisions. An important psychological
construct for the description and differentiation of individuals are,
E-mail address: s.poier.125@studms.ug.edu.pl.
Contents lists available at ScienceDirect
Energy Research & Social Science
journal homepage: www.elsevier.com/locate/erss
https://doi.org/10.1016/j.erss.2021.102087
Received 13 January 2021; Received in revised form 12 April 2021; Accepted 19 April 2021

Energy Research & Social Science 77 (2021) 102087
2
besides e.g. the basic human values, the Big Five personality traits of
consumers. The Big Five (openness to experience, conscientiousness,
extraversion, agreeableness, and neuroticism) have established them-
selves as the best-known and most researched instruments for describing
patterns of human behavior.
At the household level, there are already studies that address the in-
uence of risk, environmental concern, and the Big Five on PV adoption.
In regard to family decisions, another component comes into play. In
2016, nearly 63% of the households in Germany were not single-person
households [6]. Although in many surveys a specic “head of house-
hold” who has the most information about the household’s affairs must be
identied, this does not necessarily mean that he or she is the sole
decision-maker. In numerous cases, decision-making is the result of a
bargaining process between the individuals involved. Whether they agree
to compromise or attempt to enforce their opinions in order to maximize
their own utility, the personalities of the individuals involved may shape
the resulting decision. While most studies treat the household as a single
entity, it is the aim of this article to close this research gap and to consider
the personalities of all the different decision-makers in a household. The
objective of the present study was to look at the inuence of personality
traits of all adult household members on the purchase of a PV system in a
differentiated manner. Thus, the bias of aggregating at the household
level should be eliminated and reveal who is ultimately responsible (based
on the Big Five) for the household’s decision.
The remainder of this article is organized such that a review of the
literature is provided in Chapter 2, followed by a detailed presentation
of the Big Five personality traits and decision making in the household,
as well as the hypothesis formulation. Chapter 3 presents the source of
the data and describes how the working sample is composed. After that,
the research procedure is explained. The results are presented in chapter
4 – rst on the household level and then separated by individual
household members per household type. This is followed by a discussion
of the results in Chapter 5. The article ends with the conclusions.
2. Background
The purchase of a photovoltaic system can be viewed from different
angles. Thus, as an energy efciency investment, it represents a way for a
household to do something to protect the environment. As a novel high-tech
good, it is at the same time an innovation. But it is also a consumer good
(albeit an unusual one). Wolske, Stern, and Dietz provide a framework
combining theories from these different elds to explain the adoption of PV
systems in the US [7]. According to their study, these are the diffusion of
innovations theory [8] in the case of innovations, the theory of planned
behavior (TPB) [9] in the case of consumer goods and the value belief-norm
theory (VBN) [10] in the case of pro-environmental decision making. The
authors emphasize that these theories should be understood as comple-
mentary rather than competitive. Since the large data base of the SOEP is to
be used in this study in order to be able to examine many and heterogeneous
households, only two of the three theories mentioned by Wolske et al. are
applied. As the SOEP panel contains questions on risk preference and envi-
ronmental attitudes but no measurement instrument for the diffusion of
innovation theory, the VBN theory and the TPB were used to explain the
causal chain from personality traits to the adoption of the PV system. Korcaj,
Hahnel, and Spada also used the TPB to determine intentions to adopt PV
systems [11]. They found that purchase intention was strongly determined
by peer expectations and behavior. Status, nancial gain, but also potential
costs and risk formed the attitude toward PV plants. Similarly, Sun et al.
investigated the drivers and barriers of PV adoption in Taiwan. They found
that, contrary to previous expectations, environmental concern did not have
a signicant inuence on the intention to buy, but the environmental life-
style of the prospective buyers did [12]. Palm et al. identied nancial
motives as the main driver and fear of cost as barriers to PV adoption [13]. In
order to be able to causally justify the effect of personality traits on purchase
decisions, this study takes the approach adopted by Wolske et al. as a basis.
However, the VBN theory refers to values as the starting point of the chain of
effects on the assumption that pro-environmental behavior is involved. In
order to account for the case that the purchase of a PV system is not pro-
environmental behavior for the user, the value-attitudes-behavior hierar-
chy [14] is added to the framework. Here, human values provide for the
formation of an attitude towards the PV system, which in turn leads to a
behavior. The missing link between values and traits can be derived from
personality research [15–17]. Here it is postulated that traits are created
before the development of values and remain more or less stable over the life
cycle, while values evolve (Fig. 1). Busic-Sontic et al. have conducted initial
studies addressing the inuence of personality traits on the adoption and
diffusion of energy efciency measures in Germany and the UK, also using
panel data [18–20].
The decision-making process is shaped by incentives and barriers, or
expressed in economic terms, it is the result of a utility-maximization
process where the household or consumer will opt for an investment
when the discounted savings exceed the discounted costs [21].
Fig. 1. Causal Chain from Big Five to Behavior.
S. Poier

Energy Research & Social Science 77 (2021) 102087
3
However, this view is exclusively nancial. For an individual, good
feelings, positive expectations, and more-optimistic evaluations of
certain states also count as savings, and their opposites are perceived as
costs. Andreoni [22] described such feelings as a “warm glow,” referring
to individuals who derive satisfaction from certain behaviors as
“imperfect altruists” because they receive additional utility from that
behavior. Thus, this behavior is not purely altruistic but also egoistic.
Kahneman and Knetsch [23] were able to conrm through experiments
that individuals buy moral satisfaction by paying a higher price. Ad-
vantages of choosing a PV system might include increased independence
from the energy supplier and the development of electricity prices, the
conviction of having contributed to a sustainable energy transition, and
the positive feeling of having helped the environment. However, like
other investments, barriers can include large and specic costs in the
present and uncertainties in the future; for a PV system, these can
include developments in wholesale electricity prices, amount of self-
consumption, interest rates, opportunity costs, and the possibility of
technical damage to the system or even the loss of the home by re or
another disaster [24]. Moreover, uncertainty about the potentially
polluting production of solar panels or concerns that a re in the home
might not be extinguished because of the possibility of electric shock can
be perceived as risks or potential costs. Due to the lack of knowledge
about the future and the uncertain development of nancial-purchase
parameters, PV adoption is a decision under uncertainty. Thus, it will
be positively inuenced by the decision-maker’s risk propensities.
In summary, it can be assumed that the inuence of the Big Five
personality traits on investment in a photovoltaic power system is
mediated, on one hand, by attitudes toward the environment and, on the
other, by the consumer’s risk propensities. Thus, concerns about the
environment as well as higher risk preferences could represent moti-
vating factors for adopting a PV system [20].
2.1. The Big Five personality traits
Personality traits, and especially the Big Five, are constructs for
describing individuals. Assuming a certain stability, they can be useful for
describing or even predicting human behavior and, in particular, purchase
behavior. A number of denitions for personality traits are found in the
literature. For example, DeYoung describes them as “probabilistic de-
scriptions of relatively stable patterns of emotion, motivation, cognition, and
behavior, in response to classes of stimuli that have been present in human
cultures over evolutionary time” [25]. Following John, Naumann, and Soto
[26] and Valchev et al. [27], they are habitual patterns of behavior, thought,
and emotion that are stable over time and in comparable situations. What all
denitions have in common is “the emphasis on the relative consistency of
behavioral predispositions to behave in a particular manner across situa-
tions” [16]. Over the last few decades, researchers have developed several
frameworks to describe individuals’ personalities using descriptive terms for
patterns of behavior with different numbers of dimensions (See Table 5 in the
appendix for the ve factors, each one comprising six facets.). The most
frequently used and best-known models in contemporary research include
ve personality traits or factors and are known as ve-factor models (FFM) or
the Big Five [28–30].
In the process of making a decision, consumers attempt to balance all
known inuencing factors against one another reasonably to arrive at
the most-positive and benecial result possible. As they move through
these evaluation stages, individuals not only have to process information
rationally but also manage the non-cognitive preferences, beliefs, and
attitudes that could emerge based on their psychological characteristics.
Many researchers have found evidence for the inuence of individual
differences such as personality traits related to pro-environmental be-
haviors like switching off lights, recycling waste, or biking instead of
driving a car [31,32]; however, few studies have investigated the impact
of the Big Five personality traits on energy-efcient investments. Busic-
Sontic, Czap, and Fuerst [20] identied associations between the Big
Five and high-cost energy-efciency investments in the UK that were
mediated by perceived risk and environmental concern for households
that previously showed interest in those investments. Similarly, Brick
and Lewis suggested personality traits to predict pro-environmental
behavior [33]. In a study among 345 US American adults they found
that openness, extraversion, and conscientiousness could sufciently
predict emission-reducing pro-environmental behavior, mediated by
pro-environmental attitudes. The effect of openness is in line with other
literature. Personality traits also inuence household nances and
nancial decisions. Brown and Taylor found that only extraversion and
openness had an effect on the amount of the debts and assets held by the
household. Especially, openness positively affected high-risk in-
vestments [31]. In another study of the German PV market, Busic-Sontic
et al. again found indirect effects of the Big Five traits mediated by
environmental concern [19]. They also used data from the SOEP and
provided a rationale for why risk preference and environmental concern
should be considered as mediators. However, in contrast to the present
investigation, they considered the household as a single unit without
considering the different members and used path analysis instead of
ESEM. Jacksohn et al., on contrast, could not measure any signicant
effects of the Big Five on the purchase decision [34].
2.2. Household Decision-Making
In an attempt to develop a model of household decision-making,
households have usually been treated as single and homogeneous
decision-makers. Like individuals, they attempt to maximize their ex-
pected utility constrained by a given budget [35]. These traditional
models do not consider that family members, especially a spouse, can
have heterogeneous interests and that their resources are unequally
distributed, which may affect the nal decision [36]. Partners in a
household often disagree not only about decision-making [37] but may
also have different opinions about who the decision-maker is [38]. The
individuals within a household can achieve consensus through either a
cooperative or non-cooperative process. In a cooperative model, in-
dividuals attempt to reach a collective decision through negotiation as a
compromise of their preferences. In a non-cooperative model, family
members may either attempt to maximize their individual utility func-
tions or select a “benevolent patriarch” (or matriarch) who is believed to
have the most knowledge and makes decisions for the family [39].
Without spousal surveillance, the decisions of the head of household
widely reect his or her preferences [40]. Thus, non-cooperative models
are unlikely to achieve Pareto-efcient outcomes [41]. In the SOEP
study, the head of household is identied as the person who knows the
most about the household’s affairs; however, he or she is not necessarily
the sole decision-maker. Most participants declare that they make de-
cisions with their partner. Thus, it is clear that focusing on only one
person’s preferences and predispositions in a household will lead to
biased results. Instead, a household can be treated either as a unit
consisting of single individuals nested in it, or the household members
could be observed on the individual level.
It is unlikely that decisions result from rational bargaining, and even
if the “benevolent dictator” acts as responsibly as possible, it is also
unlikely that he or she will be able to process all the available infor-
mation and make an efcient evaluation including all weighted prefer-
ences. Yilmazer and Lyons revealed that spouses with more negotiating
power or higher cognitive abilities or nancial resources may inuence
the household’s decisions in favor of their own preferences [42]. An
intensive discussion about a topic is even more important when it does
not involve habitual, everyday decisions but rather cost-intensive, one-
time purchasing decisions such as buying a new car, investing in stocks,
or purchasing a photovoltaic system for the home. Because of the long-
term strategic importance, capital intensity, and problematic revers-
ibility of these decisions, they can be dened as investments rather than
consumption. Whether all households can be treated equally is ques-
tionable. If there is more than one household member, the interesting
question is whether it is possible to base the purchase decision on the Big
S. Poier

Energy Research & Social Science 77 (2021) 102087
4
Five personality traits of a particular dominating member.
The aim of this study is to examine the extent to which one-person
households, couple households, and other households with more than
one adult differ with respect to the effect of personality traits on the
decision to buy a PV system. Which of the household members con-
tributes the most dominant, signicant effect to the purchase decision, if
it is, in fact, the head of the household, and whether the following hy-
potheses are upheld will be investigated. For this reason, following
Busic-Sontic et al. [18–20], the rst decision was to test the signicance
of the results for the contribution of the Big Five on PV adoption. From
the literature review, the possibility emerged that risk preference and
environmental concern could play a mediating role – therefore, it was
subsequently decided to test the signicance of the direct and indirect
effects as well.
2.3. Hypotheses
According to the existing literature, hypotheses are developed based
on how the Big Five personality traits inuence the decision to invest in
a rooftop photovoltaic system and their impacts on the four mediators as
well (Table 1). The resulting exploratory structural equation model is
presented in Fig. 5.
•Openness to Experience: people who score high on openness tend to
be intellectually curious and open-minded toward new experiences
and pro-environmental activities [43]. These individuals have a
higher likelihood of taking risks [20,31]. Moreover, they tend to
adopt new technologies faster and more easily than others.
•Conscientiousness: Those who score high on conscientiousness tend
to have condence in their abilities and to achieve goals in controlled
situations. A lower score in conscientiousness is an indication of a
higher willingness to take on debt [31]; a higher score is negatively
associated with nancial risk tolerance [44,45].
•Extraversion: Individuals who score high on extraversion are inter-
ested in others and in their company; they strive for social ascen-
dancy and want to enjoy their lives. According to the existing
literature, there is no clear contribution to environmental concern
[43,46], while there are contradictory results as far as the total effect
on pro-environmental behavior is concerned [20,33].
•Agreeableness: People scoring high on agreeableness are trustful and
altruistic. They tend to be modest and self-effacing [30]. Thus, it is
hypothesized that A has weak to negative contributions on risk
propensities. Because those high in A care about the environment
and the welfare of others, it is expected that they are likely to
demonstrate environmental concerns [46].
•Neuroticism: Individuals with a high neuroticism score tend to be
anxious, shy, and nervous. They act emotionally and impulsively and
are less resistant to emotional stress. Consequently, it is possible that
neuroticism relates positively to environmental concerns because
these persons are anxious about actual developments and cannot
imagine the consequences [46]. People who score high in N are
unlikely to reconsider an issue over a longer period and likely to
avoid risk. Thus, neuroticism could contribute negatively to risk
preferences and photovoltaic-system adoption [45]. Conversely,
they might make risky decisions because of a lack of ability to
objectively evaluate situations.
3. Methodology
The SOEP is a representative, nationwide survey across more than
11,000 private households. For this wide-range longitudinal study, more
than 30,000 respondents are interviewed year to year. The survey began
in 1984; when this article was prepared, wave 33 from 2016 had the
most actual data [6,47]. In addition to those questions that are com-
ponents of every wave of the survey, special topics ow into the inves-
tigation. Among numerous other variables, the SOEP includes variables
about the Big Five personality traits, risk preferences, and concerns in
several domains and other psychological items. The data can be
retrieved from the German Institute for Economic Research (Deutsches
Institut für Wirtschaftsforschung, DIW) free of charge but are reserved
for academic use exclusively and for registered researchers. It is reported
how the sample size was determined, all data exclusions, all manipu-
lations, and all measures in the study. In the years 2005, 2009, and 2013,
a self-completion questionnaire for the Big Five personality traits was
part of the SOEP study [48]. A short version of the 44-item Big Five
Inventory (BFI) with 15 questions, called the Big Five Inventory Short
(BFI-S), was used. Each Big Five dimension was measured with three
items, using a seven-point Likert scale, where 1 meant “Does not apply”
and 7 meant “Applies fully”. In 2009 and 2013, a fourth item (“being
inquisitive”) was added to the Openness dimension [49]. Before the BFI-
S was added to the SOEP panel, its external validity was tested and
considered sufcient for capturing users’ personality traits [50]. How-
ever, with Cronbach’s alpha values of 0.653, 0.580, 0.668, 0.458, and
0.638 the validity was weak for openness, conscientiousness, extraver-
sion, agreeableness, and neuroticism, respectively (see Table 7 in the
appendix for correlations). However, while most studies concerning
personality traits investigate student samples, which result in a bias
toward young adults with higher levels of education, the great advan-
tage of nationwide studies is their representativeness.
Annually, the SOEP measures individuals’ worries in several do-
mains with a Likert-type scale ranging from 1 to 3. Regarding the pur-
chase of residential solar-power systems, “Worried about environment”
and “Worried about consequences of climate change” were used to
construct a variable “environmental concern” (EC, Cronbach’s
α
=
0.86). After calculation, EC was reverse coded because, in the ques-
tionnaire, the response 1 means “very concerned” and 3 means “not
concerned at all.” The question concerning risk propensity is asked
annually using a Likert-type scale ranging from 0 (risk averse) to 10
(fully prepared to take risks). Risk preference is moderately stable over
time [51]. For a brief, two-year period, mean-level stability is sufcient
to use as a determinant. Over the life span, risk preferences decrease,
meaning that older individuals are less willing to take risks than younger
people are [52]. Busic-Sontic, Czap, and Fuerst [20] also tested for
stability over a period of three to four years to use risk preference as a
mediating factor in green decision-making. Household income (a free
input eld) was divided by the number of household members to
generate the per capita income variable.
3.1. Construction of the working sample
In 2016, a total of 29,178 individuals ages 18 to 102 participated in
the survey. The working sample should include adults living in their own
dwellings in the survey year who reported being the head of household,
plus all other members aged 18 years and older of the household. In the
SOEP for 2015 and 2016, a special module for energy consumption with
questions about ownership of a photovoltaic system and its yield in the
previous year was included. Additionally, the respondent should own
the dwelling because renters are unlikely to be in a position to make a
decision about altering the dwelling without the owner’s consent.
Moreover, renters would not make such a substantial nancial
Table 1
Hypotheses for Effect Directions in Mediation Analysis.
Trait EC Risk PV
Openness to Experience + + +
Conscientiousness 0 – 0
Extraversion 0 0 0
Agreeableness +– +
Neuroticism +– –
Note: Table 1 presents causal directions of mediation effects: Environmental
Concern (EC), Risk Preference (Risk), effect on the purchase of a PV system (PV),
+ = positive effect, 0 =no effect, − = negative effect.
S. Poier

Energy Research & Social Science 77 (2021) 102087
5
investment in property they do not own. A dichotomous dependent
variable was created to identify owners of a PV system. A value of 1
means that the household has a solar-power system; 0 means it does not.
In the present study, no distinction is made between dwellings that had a
solar-power system installed before they were purchased and those
where solar panels were installed following the purchase. Greater will-
ingness to pay for dwellings with PV systems will be considered as a
decision similar to the purchase decision. A total of 6,300 individuals
reported being the head of household (HoH) in an owner-occupied
dwelling. Male respondents accounted for 60.5% (N =3,591),
exceeding the share of 48.3% in the German population. Clearly, male
household members are more often regarded as the person who knows
best about the household. Only 8.3% (N =526) reported that they had
the last word on nancial decisions; 4.6% (N =292) named their part-
ners; and the vast majority, 68.5% (N =4,317), stated that they made
nancial decisions with their spouse. Of the 6,300 HoH (each repre-
senting a household), information about a photovoltaic system was
provided by 5,931; 5,515 (93%) reported not owning a PV system; and
416 (7.0%) reported owning a PV system. In the next step, the corre-
sponding household members were added to their HoH from the dataset.
As the SOEP dataset contains only adult participants, children under 18
were not included in this investigation. Finally, all cases were deleted
where no information about PV ownership or at least one personality
item was given. The nal sample comprised 9,281 individuals (mean
age =55.74, 52.5% female): 5,392 heads of household; 3,374 partners;
and 515 other household members. Of the other household members,
487 (95%) were children or stepchildren of the heads of household and
their partners. The summary statistics (Table 6 in the appendix) show
that the average monthly household income of PV owners (4,134 EUR)
is remarkably higher than that of non-adopters (3,582 EUR), while the
number of persons in non-adopter households is slightly lower (mean =
2.63) than that in PV-owner households (mean =3.12). Non-adopters
also have fewer children (mean =1.97) than adopters of photovoltaic
systems (mean =2.12).
3.2. Statistical analysis
In many cases, the change in the dependent variable is explained by a
third variable, a mediator, which is affected by the independent variable.
To discover the indirect effects of further variables, a mediation analysis
can be performed through structural equation modeling. Here, the Big
Five personality traits were the antecedent or independent variables (X
i
).
Environmental concern (M1) and risk preference (M2) as measured by the
German SOEP study were the mediating variables, and the adoption of a
rooftop photovoltaic system was the dependent variable (PVI).
In a mediation analysis, the total effect c can be decomposed into the
direct effect c’ (holding all mediators and control variables constant) and
the indirect effects M
1
– M
2
(controlling for c’). Fig. 2 illustrates the total
effect. The indirect effects represent the number of Big Five personality
traits mediated by environmental concern and risk preference (Fig. 3). In
the present study, a multigroup structural equation analysis with latent
variables and parallel mediators was conducted using Mplus [53], where
the groups were represented by the different types of households. Mplus is
a statistical software to conduct not only path analyses with observed
variables but also structural equation modeling with latent variables and
factor analyses with a binary outcome – which was needed for the present
study. The software also provides the possibility to report total indirect,
direct and total effects, even with a binary outcome [54].
Before structural equation modeling was conducted, whether the
items measure the correct latent constructs had to be tested. Exploratory
factor analysis (EFA) revealed that, for ve traits, all items loaded on the
correct factor. Many studies in the eld of personality traits had prob-
lems with poor model t in CFA. This means the model could not reect
the theoretical substructure [55]. The reason was that the CFA – in
contrast to the EFA – assumed that the observed variables loaded only on
their respective factors and the factor loads to other factors are xed at
zero. This highly restrictive assumption is unrealistic not only in practice
but also in theory. A solution that has been used increasingly in recent
years is exploratory structural equation modeling (ESEM). Here, the
factor loadings of the specic items are estimated as before, but all other
items are not xed exactly at zero but estimated as close to zero as
possible [56]. This small variability makes the model much more real-
istic and increases the model t. In the present research the most widely
used indicators of model t are used, namely the comparative t index
(CFI), Tucker-Lewis index (TLI), root mean square residual (SRMR), and
root mean square error of approximation (RMSEA) [57]. According to
the literature, the following values reected a good t of the model to
the data: CFI and TLI greater than 0.95, RMSEA <0.05, and SRMR <
0.05 [55]. Fig. 4 shows the restricted factor loadings as dotted lines.
The measurement model consisted of the ve latent variables
openness, conscientiousness, extraversion, agreeableness, and neuroti-
cism, each measured by three items as reective indicators – only
openness was measured by four items. Per capita income as an exoge-
nous variable as well as risk preference, environmental concern, and PV
Fig. 3. Direct and Indirect Effects.
Fig. 2. Total Effect.
S. Poier

Energy Research & Social Science 77 (2021) 102087
6
adoption as endogenous variables were included as manifest variables in
the model. The complete SEM is presented in Fig. 5. The estimations
were conducted using weighted least squares means and variance esti-
mator (WLSMV) which can handle data that do not meet the condition of
normality. The ESEM used target rotation and the estimations were
carried out in two steps. In the rst, individuals were clustered by their
household number and grouped by the type of household. In the second
step, individuals were clustered by their household number and grouped
according to their position in the household (HoH, partner, other adult
member).
4. Results
Using ESEM, the model provided a good t of the data (CFI =0.964,
TLI =0.913, SRMR =0.018, RMSEA =0.047 with 90% CI =0.045/
0.050). A prior regression analysis showed a signicant negative impact
Fig. 5. Exploratory Structural Equation Model. Note: Shown is the exploratory structural equation model with the Big Five personality traits – measured by 16 items
– affecting photovoltaic adoption, mediated by risk preference (Risk) and environmental concern (EC) and with per capita household income as covariate. Straight
arrows represent regressions; curved arrows represent covariances. The items plh212-plh255 are indicators for the ve latent factors openness (O), conscientiousness
(C), extraversion (E), agreeableness (A), and neuroticism (N).
Fig. 4. Measurement Model using
Exploratory Structural Equation
Modeling (ESEM). Note: Shown is the
measurement model for the Big Five
Personality Traits using Exploratory
Structural Equation Modeling. Solid ar-
rows represent estimated factor load-
ings; dashed arrows represent factor
loadings that are approximated as close
to 0 as possible. The items plh212-
plh255 are indicators for the ve latent
factors openness (O), conscientiousness
(C), extraversion (E), agreeableness (A),
and neuroticism (N).
S. Poier

Energy Research & Social Science 77 (2021) 102087
7
of extraversion (−0.106, p =.018) on the ownership of a PV system only
for the group of couple households. The individuals in the working
sample were clustered by household number and distinguished into
three groups: single households (n =847), couples in two-person
households (n =3,518), and other household variants with more than
one adult person (n =4,459).
4.1. Group comparison on the household level
The rst group (n =888, mean age =68.05, 58.7% female, 2.3%
owners) addressed single households where decisions were made by
only one person without interactions and where only the head of
household’s Big Five personality traits were regarded. Again, the model
had an acceptable t (CFI =0.939, TLI =0.919, SRMR =0.044, RMSEA
=0.033 with 90% CI =0.031/0.034). No signicant direct or indirect
effect could be detected with a non-measurable explanation of the
variance for PV adoption (R
2
=0.116, p =.248) and an R
2
of 0.228 (p <
.001) for risk preference and R
2
=0.067 (p =.003) for environmental
concern. While there were some signicant regressions from risk pref-
erence on openness (0.732, p <.001), agreeableness (−0.565, p <.001),
and neuroticism (−0.727, p <.001), and from environmental concern
on openness (0.131, p <.001), agreeableness (−0.075, p =.016),
neuroticism (0.069, p =.006), and per capita income (0.030, p <.001),
no signicant effects of mediators on the ownership of a PV system were
found (Table 2). From this, it can be interpreted that neuroticism
reduced the willingness to take risks, while it had a positive effect on
concern for the environment. Openness to experience, by contrast,
increased both willingness to take risks and sensitivity to environmental
concerns. Singles scoring higher on agreeableness were likely to take
fewer risks and to show less environmental concern.
Instead of single households, households that consisted exclusively
of couples without any other members (n =3.698, mean age =68.40,
49.8% female, 7% owners) were considered in the next step. This group
was slightly younger with fewer females and a three times higher share
of PV system owners. Now, the direct (−0.109, p =.015) and total effect
(−0.108, p =.017) of extraversion on the ownership of PV systems
became signicant with a non-signicant explained variance for PV
adoption of 0.024 (p =.092). Again, strong direct effects of openness
(0.476, p <.001), extraversion (0.203, p <.001), agreeableness
(−0.360, p <.001), and neuroticism (−0.537, p <.001) on risk pro-
pensities could be demonstrated. Rather weak direct effects of openness
(0.093, p <.001), conscientiousness (−0.032, p =.029), agreeableness
(0.054, p <.001), and neuroticism (0.069, p <.001) on environmental
concern were found. Per capita household income had a positive inu-
ence on risk preference and a negative one on environmental concern.
Neither of the two mediators had a signicant impact on the outcome
with an explained variance for risk preference of 0.141 (p <.001) and
for environmental concern of 0.050 (p <.001). It was striking that the
effect of agreeableness on environmental concern reversed its direction
and was now weakly positive. In addition to the negative effect of ex-
traversion on the outcome, it had a strengthening effect on risk
preference.
A more differentiated picture emerged for households with more
than two adults or single parents with children where interaction was
possible not only between partners but also with other family members
(n =4.692, mean age =45.82, 53.5% female, 9.6% owners). Members
of this group of households were about 20 years younger on average
than the rst groups. Although the explained variance for PV adoption
became signicant for the rst time (R
2
=0.016, p =.047), no direct
effect of a personality trait on the ownership of a PV system could be
determined. All ve traits had a signicant effect on risk-taking (R
2
=
0.138, p <.001). The strongest effects came from openness (0.463, p <
.001), agreeableness (−0.422, p <.001), and neuroticism (−0.544, p <
.001). Conscientiousness (−0.157, p =.001) and extraversion (0.245, p
<.001) had a smaller impact. Regarding environmental concern, only
openness (0.059, p <.001), agreeableness (0.062, p <.001), and
neuroticism (0.043, p <.001) had a signicant effect with an explained
variance of R
2
=0.033 (p <.001). Per capita household income had a
very small albeit signicant effect exclusively on environmental concern
(0.005, p =.003). The effects of the traits from the pairwise view had
thus retained their directions. In contrast to the study of single and
couple households, the effects of risk-taking and environmental concern
were now signicant for families. Therefore, the effects of personality
traits on the ownership of PV systems could now be mediated by envi-
ronmental concern and risk-taking. In fact, all ve characteristics were
mediated by risk-taking. The strongest indirect effects here were found
for openness to experience (0.020, p =.006), agreeableness (−0.018, p
=.005), and neuroticism (−0.023, p =.004). Weak but signicant in-
direct effects were found for conscientiousness (−0.007, p =.028) and
extraversion (0.010, p =.011). As the regressions for EC suggested, the
indirect effects here were signicantly smaller than for risk-taking.
Signicant indirect effects were registered for openness (0.007, p =
.039), agreeableness (0.008, p =.026), and neuroticism (0.005, p =
.035). Consequently, the direct effect of extraversion on the outcome of
the pair study was completely mediated by risk preference albeit with
the opposite sign.
4.2. Group comparison on the individual level
Since there were latent variables for all ve personality traits for
each individual, the individual households could be broken down into
their members. To shed light on how the structures within the house-
holds were designed, the model was recalculated on the level of the
individual. Household members were categorized as follows: a) partic-
ipants in single households, b) head of household in a couple household,
c) partner in a couple household, d) head of household in a household
with at least one additional member, e) partner in a household with head
Table 2
Direct and Indirect Effects of Personality Traits on the Household Level.
EC Risk Sum Direct Total
Trait Single Household Clusters (n =847)
Openness to
Experience
. 010
[0.607]
−0.021
[0.664]
−0.011
[0.825]
0.046
[0.808]
0.035
[0.836]
Conscientiousness 0.000
[0.896]
−0.002
[0.760]
−0.001
[0.822]
0.084
[0.142]
0.082
[0.667]
Extraversion −0.002
[0.681]
0.003
[0.702]
0.001
[0.870]
−0.148
[0.447]
−0.147
[0.447]
Agreeableness −0.006
[0.609]
0.016
[0.665]
0.010
[0.781]
−0.315
[0.111]
−0.304
[0.106]
Neuroticism 0.005
[0.610]
0.020
[0.664]
0.026
[0.602]
0.048
[0.780]
0.074
[0.608]
Couple Households (1,931 Clusters, n =3,518)
Openness to
Experience
0.006
[0.301]
0.001
[0.896]
0.008
[0.446]
0.086
[0.104]
0.094
[0.064]
Conscientiousness −0.002
[0.342]
0.000
[0.896]
−0.002
[0.366]
−0.069
[0.142]
−0.072
[0.126]
Extraversion 0.001
[0.503]
0.000
[0.896]
0.001
[0.722]
−0.109
[0.015]
−0.108
[0.017]
Agreeableness 0.004
[0.310]
−0.001
[0.896]
0.003
[0.671]
−0.017
[0.711]
−0.014
[0.754]
Neuroticism 0.005
[0.303]
−0.001
[0.896]
0.004
[0.718]
−0.071
[0.091]
−0.067
[0.087]
Households with other members (2,360 Clusters, n =4,459)
Openness to
Experience
0.007
[0.039]
0.020
[0.006]
0.027
[0.001]
−0.060
[0.149]
−0.033
[0.418]
Conscientiousness 0.003
[0.141]
−0.007
[0.028]
−0.004
[0.254]
−0.037
[0.341]
−0.041
[0.291]
Extraversion −0.003
[0.145]
0.010
[0.011]
0.007
[0.099]
0.043
[0.284]
0.050
[0.211]
Agreeableness 0.008
[0.026]
−0.018
[0.005]
−0.010
[0.164]
0.014
[0.720]
0.004
[0.915]
Neuroticism 0.005
[0.035]
−0.023
[0.004]
−0.018
[0.036]
0.024
[0.479]
0.006
[0.848]
Note: Table 2 presents standardized mediation effects: Environmental Concern
(EC), Risk Preference (Risk), the sum of the indirect effects (Sum), direct effect,
and total effect; p−values in brackets, 5,138 household clusters, n =8,824.
S. Poier

Energy Research & Social Science 77 (2021) 102087
8
of household and at least one additional member, f) other member in a
household with head of household and at least one additional member.
The rst group had been investigated in the previous subchapter. The
model t decreased slightly (CFI =0.933, TLI =0.917, SRMR =0.050,
RMSEA =0.034 with 90% CI =0.033/0.036).
As in the aggregated analysis, there was no signicant inuence of
environmental concern and risk propensity on the ownership of a PV
system. Therefore, no signicant mediation was possible for both, head
of household and partner. The explained variance of PV adoption was
not measurable for either the head of household (R
2
=0.036, p =.076)
or the partner (R
2
=0.043, p =.096). All ve personality traits of the
head of household affected risk signicantly with an R
2
of 0.128 (p <
.001). The strongest inuence was exerted by neuroticism (−0.500, p <
.001) and openness (0.418, p <.001), followed by agreeableness
(−0.260, p <.001), extraversion (0.234, p <.001), and conscientious-
ness (−0.154, p =.019). Environmental concern was only signicantly
affected by the head of household’s openness (0.078, p <.001), agree-
ableness (0.046, p =.015), and neuroticism (0.076, p <.001) with an R
2
of 0.045 (p <.001).
In regard to the partner’s traits in a couple household, only openness
(0.121, p <.001), conscientiousness (−0.069, p =.002), agreeableness
(0.067, p<=.003), and neuroticism (0.052, p =.005) had a signicant
inuence on environmental concern. The partner’s explained variance
of environmental concern was not measurable (R
2
=0.066 (p =.076).
This also corresponded to the pattern from the aggregated approach. As
in the aggregated model, openness (0.507, p <.001), extraversion
(0.202, p =.012), agreeableness (−0.430, p <.001), and neuroticism
(−0.495, p <.001) affected risk propensities signicantly (R
2
of 0.140,
p <.001).
In contrast to the aggregated study, the direct and total effect of
neuroticism of the head of household (−0.157, p =.010) on the outcome
became signicant, as did a direct effect of the partner for openness
(0.148, p =.044). Furthermore, it could be shown that the direct effect
for extraversion from the aggregated view was only signicant for the
partner in the split view (−0.181, p =.018). This could indicate that the
decision a household makes about a major investment in renewable
energy – such as a PV system – was signicantly inuenced by the HoH’s
partner rather than the head of the household (Table 3).
In households where there was at least one other family member in
addition to the head of the household, the picture for the HoH was
similar to that in couple households. Since there was no signicant effect
of environmental concern and risk-taking on the outcome, no mediation
took place. In addition to not signicant effects, the explained variance
of the outcome was R
2
=0.013 (p =.173). In the same way, openness
(0.065, p <.001), agreeableness (0.069, p <.001), and neuroticism
(0.050, p =.001) had a signicant inuence on environmental concern.
Risk preference was affected by openness (0.443, p <.001), extraversion
(0.208, p =.001), agreeableness (−0.509, p <.001), and neuroticism
(−0.509, p <.001). However, this was not consistent with the results of
the aggregated survey for this household type. As described above,
signicant mediator effects were found here for all traits and for both
mediators. The variance explained by the model for environmental
concern and risk preference was R
2
=0.033 (p <.001) and R
2
=0.132
(p <.001), respectively.
A relationship became apparent when one looked at the results for
the partners. As with the head of the household, there were signicant
regressions of risk propensity on openness (0.447, p <.001), extraver-
sion (0.243, p =.001), agreeableness (−0.376, p <.001), and neuroti-
cism (−0.541, p <.001) with an R
2
of 0.139 (p <.001). Instead, there
were signicant effects on environmental concern only from extraver-
sion (−0.043, p =.035) and agreeableness (0.069, p =.001) with a
variance explained of R
2
=0.034 (p =.001). Furthermore, the effects of
environmental concern (0.169, p =.018) and risk propensity (0.063, p
=.002) on the ownership of a PV system became signicant and the
model explained R
2
=4.2% of PV adoption’s variance (p =.030).
In 94.6% of the cases, the other person in a household is a child or
stepchild. Here openness (0.488, p =.002), conscientiousness (−0.603,
p <.001), extraversion (0.471, p <.001), and neuroticism (−0.627, p <
.001) had a signicant inuence on risk preference (R
2
=0.208, p <
.001). Only conscientiousness (0.101, p =.007) and per capita house-
hold income (0.011, p =.007) had an impact on environmental concern
Table 4
Direct and Indirect Effects in Households with Other Persons.
EC Risk Sum Direct Total
Trait Head of Household (n =2,347)
Openness to
Experience
0.007
[0.099]
0.011
[0.202]
0.019
[0.059]
−0.019
[0.739]
0.000
[0.996]
Conscientiousness −0.002
[0.351]
−0.001
[0.698]
−0.003
[0.302]
−0.019
[0.697]
−0.022
[0.658]
Extraversion −0.001
[0.572]
0.005
[0.222]
0.004
[0.373]
0.041
[0.451]
0.045
[0.406]
Agreeableness 0.008
[0.087]
−0.013
[0.196]
−0.005
[0.631]
0.038
[0.477]
0.033
[0.518]
Neuroticism 0.006
[0.102]
−0.013
[0.192]
−0.007
[0.485]
−0.042
[0.408]
−0.049
[0.300]
Partner (n =1,631)
Openness to
Experience
0.008
[0.151]
0.028
[0.008]
0.036
[0.002]
−0.141
[0.036]
−0.105
[0.105]
Conscientiousness 0.005
[0.162]
−0.006
[0.187]
−0.001
[0.895]
−0.044
[0.434]
−0.045
[0.424]
Extraversion −0.007
[0.132]
0.015
[0.020]
0.008
[0.324]
0.052
[0.374]
0.060
[0.307]
Agreeableness 0.012
[0.053]
−0.024
[0.012]
−0.012
[0.278]
0.015
[0.820]
0.003
[0.966]
Neuroticism 0.006
[0.130]
−0.034
[0.004]
−0.028
[0.021]
0.102
[0.061]
0.073
[0.168]
Other Person (n =402)
Openness to
Experience
0.008
[0.453]
0.036
[0.080]
0.044
[0.069]
0.071
[0.620]
0.115
[0.407]
Conscientiousness 0.010
[0.449]
−0.044
[0.069]
−0.035
[0.170]
−0.136
[0.291]
−0.171
[0.158]
Extraversion −0.002
[0.667]
0.035
[0.089]
0.033
[0.109]
−0.025
[0.857]
0.008
[0.956]
Agreeableness 0.004
[0.510]
−0.010
[0.371]
−0.006
[0.650]
−0.186
[0.099]
−0.191
[0.087]
Neuroticism 0.002
[0.616]
−0.046
[0.073]
−0.044
[0.088]
0.041
[0.718]
−0.002
[0.983]
Note: Table 4 presents standardized mediation effects: Environmental Concern
(EC), Risk Preference (Risk), the sum of the indirect effects (Sum), direct effect,
and total effect; p-values in brackets, n =4,380.
Table 3
Direct and Indirect Effects of Personality Traits in Couple Households.
EC Risk Sum Direct Total
Trait Head of Household (n =1,922)
Openness to
Experience
0.006
[0.345]
−0.002
[0.846]
0.004
[0.713]
0.051
[0.461]
0.055
[0.398]
Conscientiousness −0.001
[0.715]
0.001
[0.847]
0.000
[0.975]
−0.068
[0.247]
−0.068
[0.240]
Extraversion 0.000
[0.836]
−0.001
[0.846]
−0.001
[0.896]
−0.060
[0.331]
−0.060
[0.325]
Agreeableness 0.003
[0.371]
0.001
[0.847]
0.005
[0.517]
−0.081
[0.212]
−0.077
[0.233]
Neuroticism 0.006
[0.345]
0.002
[0.847]
0.008
[0.538]
−0.157
[0.010]
−0.149
[0.011]
Partner (n =1,596)
Openness to
Experience
0.007
[0.542]
0.005
[0.636]
0.012
[0.436]
0.148
[0.044]
0.160
[0.022]
Conscientiousness −0.004
[0.546]
0.000
[0.748]
−0.003
[0.602]
−0.081
[0.256]
−0.085
[0.233]
Extraversion 0.001
[0.692]
0.002
[0.643]
0.003
[0.567]
−0.181
[0.018]
−0.178
[0.020]
Agreeableness 0.004
[0.548]
−0.005
[0.637]
−0.001
[0.931]
0.029
[0.707]
0.028
[0.703]
Neuroticism 0.003
[0.553]
−0.005
[0.636]
−0.003
[0.840]
−0.004
[0.947]
−0.007
[0.912]
Note: Table 3 presents standardized mediation effects: Environmental Concern
(EC), Risk Preference (Risk), the sum of the indirect effects (Sum), direct effect,
and total effect; p-values in brackets, n =3,518.
S. Poier

Energy Research & Social Science 77 (2021) 102087
9
(R
2
=0.078, p =.007). However, there is a signicant effect of risk but
no effect of the Big Five on the purchasing decision of the household.
The explained variance of the outcome was not signicant (R
2
=0.098,
p =.065). Nevertheless, no mediation can be found when estimating the
direct and indirect effects (Table 4).
5. Discussion
The results show that effects of the Big Five personality traits on PV
adoption could be detected. The important nding here is not that many
inuences are signicant but that there is a trend that provides more
measurable results the more nely the households are differentiated.
The rst model with single households revealed no signicant inuence
of the Big Five on the decision to opt for or against a residential rooftop
photovoltaic system. The model concerning couple households showed
only a negative direct effect of extraversion on the purchase decision in
the aggregated view, which seems unusual, as extraversion has little to
do with the theoretical basis for the purchase of a high-priced, envi-
ronmentally friendly technology. If one takes into consideration the
individual characteristics within the household, it becomes apparent
that the partner’s personality traits inuence the decision. Although the
partner’s openness increased the likeliness of a purchase and the head of
the household’s emotional instability reduced it, the partner’s extra-
version became so dominant that it was the only effect left on the
household level. A reason could be that when in couple households there
is an increased desire on the part of the partner for social contact and for
the company of others, there is a greater tendency to invest in hedonistic
projects than in projects like PV systems.
The partner’s importance to the purchase decision is further under-
lined in the third group of households. In the aggregated view, house-
holds with at least one other person showed signicant mediator effects
for all ve traits through risk preference and for openness, agreeable-
ness, and neuroticism through environmental concern while no direct
effect could be detected. In the differentiated analysis, neither the heads
of household nor the children (other members) showed signicant total,
direct, and indirect effects. Only among the partners could signicant
indirect effects be found for openness, extraversion, agreeableness, and
neuroticism – all of which were mediated through risk preference – with
the same direction as in the aggregated analysis. As indirect effects are
the product of two direct effects which were all smaller than 1, these
indirect effects are usually very small. This is in line with the results of
Busic-Sontic et al. [19,20]. However, despite the similarity of the topic,
the results cannot be directly compared to the ndings of Busic-Sontic
et al. [19] for several reasons. Although they used the same data
source, their dependent variable was not only PV systems but also any
other alternative energy source. Moreover, they examined at the
household level and did not distinguish different types of households or
their individual members. Even though signicant effects were only
found for certain groups and the effect size was only small - which is due
to the fact that there are many other inuencing variables that have
already been researched - an effect of the Big Five could still be
conrmed. The inuence of individual personal dispositions could,
however, be a decisive criterion if the target group becomes very narrow
and other differentiation criteria are therefore no longer applicable. This
would apply to electricity storage, for example.
6. Limitations
Because the direct effects are explained by the mediator variables
within couple households only to a small degree, it can be concluded
that there are still omitted inuence factors in regard to the purchase
decision that were not considered in this analysis [58]. Other psycho-
logical constructs that could inuence consumer behavior, for example,
locus of control or risk perception, should be investigated.
Because of the length of the questionnaire and the large number of
topics investigated by the SOEP study, only brief question batteries
could be used for each topic. Although the reliability of the BFI-S has
been tested and veried, the Cronbach’s alphas are very low and a
personality measure with 16 items instead of the more detailed original
versions is, of course, less exact. In addition, the single-item questions
for risk preference and environmental concern could be replaced by
batteries comprising several questions.
Although the effects in this study were rather weak, because many
other possible inuencing factors also play a role in the decision-making
process, it provides an interesting research method on high-priced en-
ergy efciency investments such as solar batteries, whose target group is
very narrow. Here, the other inuencing factors (e.g. type of house,
income, education level, electricity consumption) play only a minor role
or even no role at all, since they are uniform for the entire target group.
Moreover, this research revealed that in the negotiation process of PV
system purchase within a household, the head of household is not the
dominant component in terms of personality traits. Therefore, more
research is needed that addresses the nuanced study of individual con-
sumer dispositions in the area of energy efciency investments.
7. Conclusions
The aim of this article was not only to identify the direct effects and
mediating paths through which the consumers’ Big Five personality
traits inuence the outcome but also to highlight differences in the re-
sults between single households, two-person couple households, and
households with at least one other person. These models were compared
using data from a representative sample across Germany. The present
study revealed three important ndings.
First, effects of the Big Five personality traits on PV adoption were
demonstrated. Second, it matters whether the household is a single house-
hold, a couple household, or a household with other persons. Although the
same working sample was investigated, there were remarkable differences in
the results of the mediations. It could be shown that the way in which Big Five
personality traits affect the adoption of residential photovoltaic systems
changes depending not only on household member but also on the type of
household. In single households, there is no signicant effect of the con-
sumer’s personality traits on the purchase decision. However, as soon as
other adults join the household, the partner of the head of the household in
particular makes a signicant contribution to the decision. Third, the Big
Five of the different household members have different effects on the pur-
chase decision. In couple households, a stronger extraversion of the partner
has a direct negative effect on the result, which is also signicant in the
aggregated household. Also in the remaining households with at least one
other member, only the personality traits of the head of household’s partner
make a signicantly measurable contribution to the purchase decision. Here,
it is striking that openness shows a weakly positive indirect effect via risk
preference, which was expected, but has a strong negative direct effect. Here,
therefore, other concerns seem to weigh more heavily for the individual than
the risk issue. As expected, negative indirect effects of agreeableness and
neuroticism were found via risk-taking. In summary, mediator effects could
be observed only in households where there is at least one other member who
is not the partner besides the head of the household. However, the decisive
source of these effects is not the head of the household – it is the spouse.
This research paper promotes the understanding of how the Big Five
personality traits of household members affect the decision to adopt
residential photovoltaic systems. The results suggest that the inuence
of the spouse’s personality traits on family decisions is greater than
expected. Building on this knowledge, whether there are interactions
between the individual characteristics of the family members should be
further investigated.
Declaration of Competing Interest
The author declares that he has no known competing nancial in-
terests or personal relationships that could have appeared to inuence
the work reported in this paper.
S. Poier

Energy Research & Social Science 77 (2021) 102087
10
Appendix A
Table 5
The Big Five Personality Traits in Detail.
Personality traits Personality trait facets
Openness to experience:
the active seeking and appreciation of experiences for their own sake
Fantasy: receptivity to the inner world of imagination
Aesthetics: appreciation of art and beauty
Feelings: openness to inner feelings and emotions
Actions: openness to new experiences on a practical level
Ideas: intellectual curiosity
Values: readiness to re-examine one’s own values and those of authority gures
Conscientiousness:
degree of organization, persistence, control, and motivation in goal-directed
behavior
Competence: belief in one’s self-efcacy
Order: personal organization
Dutifulness: emphasis is placed on the importance of fullling moral obligations
Achievement Striving: need for personal achievement and sense of direction
Self-Discipline: capacity to begin tasks and follow through to completion despite boredom or
distractions
Deliberation: tendency to think things through before acting or speaking
Extraversion:
quantity and intensity of energy directed outwards into the social world
Warmth: interest in and friendliness toward others
Gregariousness: preference for the company of others
Assertiveness: social ascendancy and forcefulness of expression
Activity: pace of living
Excitement Seeking: need for environmental stimulation
Positive Emotions: tendency to experience positive emotions
Agreeableness:
the kinds of interactions an individual prefers from compassion to tough-
mindedness
Trust: belief in the sincerity and good intentions of others
Straightforwardness: frankness in expression
Altruism: active concern for the welfare of others
Compliance: response to interpersonal conict
Modesty: tendency to play down one’s achievements and be humble
Tender-Mindedness: attitude of sympathy for others
Neuroticism:
identies individuals who are prone to psychological distress
Anxiety: level of free-oating anxiety
Angry Hostility: tendency to experience anger and related states such as frustration and bitterness
Depression: tendency to experience feelings of guilt, sadness, despondency, and loneliness
Self-Consciousness: shyness or social anxiety
Impulsiveness: tendency to act on cravings and urges rather than reining them in and delaying
gratication
Vulnerability: general susceptibility to stress
Source: McCrae and Costa (1999).
Table 6
Summary Statistics.
Non-adopters Adopters Total
Variable N Mean/
%
SD
a
Min Max N Mean/
%
SD
a
Min Max N Mean/
%
SD
a
Min Max
Age 8,552 56.00 15.31 21 102 729 52.79 13.64 21 95 9,281 55.74 15.21 21 102
Gender
1 =male
2 =female
8,552
4,044
4,508
100
47.29
52.71
729
364
365
100
49.93
50.07
9,281
4,408
4,873
100
47.49
52.51
Marital Status
1 =married
2 =married but separated
3 =single
4 =divorced
5 =widowed
6 =husband/wife abroad
7 =registered same-sex
partnership, living together
8 =registered same-sex
partnership, living apart
8,538
6,407
128
902
511
574
1
12
3
100
74.9
1.5
10.5
6.0
6.7
0.0
0.1
0.0
728
618
10
57
24
18
0
0
1
100
84.8
1.4
7.8
3.3
2.5
0.0
0.0
0.1
9,266
7,025
138
959
535
592
1
12
4
100
75.7
1.5
10.3
5.8
6.4
0.0
0.1
0.0
Years of Education 8,552 12.90 2.81 7 18 729 13.25 2.83 7 18 9,281 12.93 2.81 7 18
Household Statistics
Monthly Household Income 5,003 3,582 2,212 0 40,000 389 4,134 1,802 500 12,000 5,392 3,622 2,189 0 40,000
Persons in Household 5,003 2.63 1.321 1 10 389 3.12 1.32 1 8 5,392 2.66 1.327 1 10
Children in Household 2,009 1.97 0.94 1 8 214 2.12 0.93 1 6 2,223 1.99 0.937 1 8
Note: Reported are the demographic statistics for the participants (heads of household) in the working sample (N =9,281);
a
Standard Deviation.
S. Poier

Energy Research & Social Science 77 (2021) 102087
11
Table 7
Correlation table of scales and constructs (N =9,281).
PLH215 PLH220 PLH225 PLH255 PLH212 PLH218 PLH222 PLH213 PLH219 PLH223 PLH214 PLH217 PLH224 PLH216 PLH221 PLH226 PLH0204
PLH215 Am
original
–
PLH220 Value
artistic
experiences
0.278
**
–
PLH225 Have
lively
imagination
0.426
**
0.329
**
–
PLH255 inquisitive 0.347
**
0.302
**
0.324
**
–
PLH212 Thorough
worker
0.171
**
0.060
**
0.074
**
0.207
**
–
PLH218 Tend to be
lazy
−0.045
**
−0.010 0.054
**
−0.047
**
−0.319
**
–
PLH222 Carry out
tasks efciently
0.233
**
0.085
**
0.133
**
0.287
**
0.496
**
−0.264
**
–
PLH213 Am
communicative
0.335
**
0.191
**
0.287
**
0.265
**
0.252
**
−0.134
**
0.242
**
–
PLH219 Am
sociable
0.331
**
0.206
**
0.299
**
0.234
**
0.144
**
−0.067
**
0.233
**
0.582
**
–
PLH223 Reserved −0.160
**
−0.015 −0.098
**
−0.039
**
0.038
**
0.024* 0.024* −0.335
**
−0.328
**
–
PLH214 Am
sometimes too
coarse with
others
0.142
**
−0.089
**
0.042
**
−0.025* −0.064
**
0.206
**
−0.074
**
0.020 0.012 −0.115
**
–
PLH217 Able to
forgive
0.102
**
0.112
**
0.100
**
0.139
**
0.111
**
−0.082
**
0.149
**
0.180
**
0.194
**
0.034
**
−0.132
**
–
PLH224 Friendly
with others
0.113
**
0.154
**
0.184
**
0.209
**
0.238
**
−0.155
**
0.295
**
0.243
**
0.221
**
0.112
**
−0.318
**
0.287
**
–
PLH216 Worry a
lot
0.028
**
0.042
**
0.039
**
−0.055
**
0.059
**
−0.028
**
0.004 0.018 −0.017 0.108
**
0.107
**
0.027
**
0.048
**
–
PLH221 Somewhat
nervous
−0.063
**
0.074
**
0.049
**
−0.087
**
−0.067
**
0.101
**
−0.097
**
−0.051
**
−0.049
**
0.126
**
0.130
**
−0.047
**
−0.026* 0.382
**
–
PLH226 Deal well
with stress
0.186
**
0.040
**
0.140
**
0.234
**
0.124
**
−0.058
**
0.208
**
0.143
**
0.164
**
−0.019 −0.049
**
0.146
**
0.146
**
−0.299
**
−0.436
**
–
PLH0204 Risk
preference
0.224
**
0.075
**
0.146
**
0.123
**
0.002 0.033
**
0.050
**
0.108
**
0.141
**
−0.155
**
0.106
**
0.021* −0.021* −0.157
**
−0.125
**
0.162
**
–
ENV_CON
Environmental
Concern
0.056
**
0.127
**
0.073
**
0.069
**
0.046
**
−0.031
**
0.013 0.042
**
0.045
**
0.027* −0.027* 0.042
**
0.069
**
0.088
**
0.048
**
−0.011 −0.040
**
*p <.05; **p >.01; items PLH215 to PLH226 measured on a Likert scale from 1 (“Does not apply”) to 7 (“Applies fully”), item PLH0204 measured on a Likert scale from 0 (risk averse) to 10 (fully prepared to take risks),
ENV_CON is an average score from “Worried about environment” and “Worried about consequences of climate change” using a Likert scale from 1 (“not concerned at all”) to 3 (“very concerned”).
S. Poier

Energy Research & Social Science 77 (2021) 102087
12
Appendix B. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.erss.2021.102087.
References
[1] P. Bonger, Grundzüge der Volkswirtschaftslehre: Eine Einführung in die
Wissenschaft von M¨
arkten, 4th ed., Pearson, Hallbergmoos, 2015.
[2] H. Beck (Ed.), Behavioral Economics, Springer Fachmedien Wiesbaden, Wiesbaden,
2014.
[3] D. Kahneman, A. Tversky, Prospect Theory: An Analysis of Decision under Risk,
Econometrica 47 (2) (1979) 263–292, https://doi.org/10.2307/1914185.
[4] A. Ramos, A. Gago, X. Labandeira, P. Linares, The role of information for energy
efciency in the residential sector, Energy Econ. 52 (2015) 17–29, https://doi.org/
10.1016/j.eneco.2015.08.022.
[5] B. Chauvin, D. Hermand, E. Mullet, Risk perception and personality facets, Risk
Anal. 27 (1) (2007) 171–185, https://doi.org/10.1111/j.1539-6924.2006.00867.x.
[6] J. Schupp, J. Goebel, M. Kroh, C. Schr¨
oder, C. Bartels, K. Erhardt, et al., Sozio-
oekonomisches Panel (SOEP), data for years 1984–2016, SOEP Socio-Economic
Panel Study, Berlin, 2016.
[7] K.S. Wolske, P.C. Stern, T. Dietz, Explaining interest in adopting residential solar
photovoltaic systems in the United States: Toward an integration of behavioral
theories, Energy Res. Social Sci. 25 (2017) 134–151, https://doi.org/10.1016/j.
erss.2016.12.023.
[8] E.M. Rogers, Diffusion of innovations, Free Press, New York, London, Toronto,
Sydney, 2003.
[9] I. Ajzen, The theory of planned behavior, Organ. Behav. Hum. Decis. Process. 50
(2) (1991) 179–211, https://doi.org/10.1016/0749-5978(91)90020-T.
[10] P.C. Stern, T. Dietz, T. Abel, G.A. Guagnano, L. Kalof, A value-belief-norm theory of
support for social movements: the case of environmentalism, Res. Human Ecol. 6
(2) (1999) 81–97.
[11] L. Korcaj, U.J.J. Hahnel, H. Spada, Intentions to adopt photovoltaic systems
depend on homeowners’ expected personal gains and behavior of peers, Renewable
Energy 75 (2015) 407–415, https://doi.org/10.1016/j.renene.2014.10.007.
[12] P.-C. Sun, H.-M. Wang, H.-L. Huang, C.-W. Ho, Consumer attitude and purchase
intention toward rooftop photovoltaic installation: the roles of personal trait,
psychological benet, and government incentives, Energy Environ. 31 (1) (2020)
21–39, https://doi.org/10.1177/0958305X17754278.
[13] J. Palm, Household installation of solar panels – motives and barriers in a 10-year
perspective, Energy Policy 113 (2018) 1–8, https://doi.org/10.1016/j.
enpol.2017.10.047.
[14] P.M. Homer, L.R. Kahle, A structural equation test of the value-attitude-behavior
hierarchy, J. Pers. Soc. Psychol. 54 (4) (1988) 638–646, https://doi.org/10.1037/
0022-3514.54.4.638.
[15] G. Caprara, M. Vecchione, S.H. Schwartz, Mediational role of values in linking
personality traits to political orientation, Asian J. Soc. Psychol. 12 (2) (2009)
82–94, https://doi.org/10.1111/j.1467-839X.2009.01274.x.
[16] R. Fischer, Personality, values, culture: An evolutionary approach, Cambridge
University Press, Cambridge, 2018.
[17] L. Parks-Leduc, G. Feldman, A. Bardi, Personality Traits and Personal Values: A
Meta-Analysis, Personality & Social Psychology Review 19 (1) (2015) 3–29,
https://doi.org/10.1177/1088868314538548.
[18] A. Busic-Sontic, F. Fuerst, Does your personality shape your reaction to your
neighbours’ behaviour? A spatial study of the diffusion of solar panels, Energy
Build. 158 (2018) 1275–1285, https://doi.org/10.1016/j.enbuild.2017.11.009.
[19] A. Busic-Sontic, C. Brick, Personality Trait Effects on Green Household
Installations. Collabra, Psychology (2018), https://doi.org/10.1525/collabra.120.
[20] A. Busic-Sontic, N.V. Czap, F. Fuerst, The role of personality traits in green
decision-making, J. Econ. Psychol. 62 (2017) 313–328, https://doi.org/10.1016/j.
joep.2017.06.012.
[21] H. Allcott, M. Greenstone, Is there an energy efciency gap? J. Econ. Perspect. 26
(1) (2012) 3–28, https://doi.org/10.1257/jep.26.1.3.
[22] J. Andreoni, Impure altruism and donations to public goods: a theory of warm-
glow giving, Econ. J. 100 (401) (1990) 464–477, https://doi.org/10.2307/
2234133.
[23] D. Kahneman, J.L. Knetsch, Valuing public goods: the purchase of moral
satisfaction, J. Environ. Econ. Manage. 22 (1) (1992) 57–70, https://doi.org/
10.1016/0095-0696(92)90019-S.
[24] D. Stevens, F. Fuerst, H. Adan, D. Brounen, D. Kavarnou, R. Singh, Risks and
uncertainties associated with residential energy efciency investments, SSRN
Electron. J. (2018), https://doi.org/10.2139/ssrn.3254854.
[25] C.G. DeYoung, Cybernetic Big Five Theory, J. Res. Pers. 56 (2015) 33–58, https://
doi.org/10.1016/j.jrp.2014.07.004.
[26] O.P. John, L.P. Naumann, C.J. Soto, Paradigm shift to the integrative Big Five
taxonomy: History, measurement, and conceptual issues, in: O.P. John, R.
W. Robins, L.A. Pervin (Eds.), Handbook of personality: Theory and research,
Guilford Press, New York, NY, 2010, pp. 114–158.
[27] V.H. Valchev, F.J.R. van de Vijver, J.A. Nel, S. Rothmann, D. Meiring, The use of
traits and contextual information in free personality descriptions across
ethnocultural groups in South Africa, J Pers Soc Psychol 104 (6) (2013)
1077–1091, https://doi.org/10.1037/a0032276.
[28] L.R. Goldberg, J.A. Johnson, H.W. Eber, R. Hogan, M.C. Ashton, C.R. Cloninger, et
al., The international personality item pool and the future of public-domain
personality measures, J. Res. Pers. 40 (1) (2006) 84–97, https://doi.org/10.1016/j.
jrp.2005.08.007.
[29] R.R. McCrae, O.P. John, An Introduction to the Five-Factor Model and Its
Applications, J. Pers. 60 (2) (1992) 175–215, https://doi.org/10.1111/j.1467-
6494.1992.tb00970.x.
[30] R.R. McCrae, P.T. Costa, A ve-factor theory of personality, in: L.A. Pervin, O.
P. John (Eds.), Handbook of personality: Theory and research, 2nd ed., Guilford
Press, New York, NY, 1999, pp. 139–153.
[31] S. Brown, K. Taylor, Household Finances and the ’Big Five’ Personality Traits,
Forschungsinstitut zur Zukunft der Arbeit GmbH, Bonn, 2011.
[32] P.C. Stern, New Environmental Theories: Toward a Coherent Theory of
Environmentally Signicant Behavior, Journal of Social Issues 56 (3) (2000)
407–424, https://doi.org/10.1111/0022-4537.00175.
[33] C. Brick, G.J. Lewis, Unearthing the “Green” Personality: Core Traits Predict
Environmentally Friendly Behavior, Environment and Behavior 48 (5) (2016)
635–658, https://doi.org/10.1177/0013916514554695.
[34] A. Jacksohn, P. Gr¨
osche, K. Rehdanz, C. Schr¨
oder, Drivers of renewable technology
adoption in the household sector, Energy Econ. 81 (2019) 216–226, https://doi.
org/10.1016/j.eneco.2019.04.001.
[35] G.S. Becker, A Theory of Social Interactions, Journal of Political Economy 82 (6)
(1974) 1063–1093, https://doi.org/10.3386/w0042.
[36] A. Schneebaum, K. Mader, The gendered nature of intra-household decision
making in and across Europe, Vienna (2013).
[37] S. Dobbelsteen, P. Kooreman, Financial Management, Bargaining and Efciency
within the Household; an Empirical Analysis, De Economist 145 (3) (1997)
345–366, https://doi.org/10.1023/A:1003008229588.
[38] H.W. Elder, P.M. Rudolph, Who makes the nancial decisions in the households of
older Americans? Financial Services Review 12 (4) (2003) 293–308.
[39] G.S. Becker, A treatise on the family, Harvard University Press, Cambridge, Mass,
2010.
[40] N. Ashraf, Spousal Control and Intra-Household Decision Making: An Experimental
Study in the Philippines, Am. Econ. Rev. 99 (4) (2009) 1245–1277.
[41] M. Browning, P.-A. Chiappori, Y. Weiss, Economics of the family, Cambridge
University Press, Cambridge, 2014.
[42] T. Yilmazer, A.C. Lyons, Marriage and the Allocation of Assets in Women’s Dened
Contribution Plans, J. Fam. Econ. Issues 31 (2) (2010) 121–137, https://doi.org/
10.1007/s10834-010-9191-6.
[43] E.M. Markowitz, L.R. Goldberg, M.C. Ashton, K. Lee, Proling the “pro-
environmental individual”: a personality perspective, J. Pers. 80 (1) (2012)
81–111, https://doi.org/10.1111/j.1467-6494.2011.00721.x.
[44] T. Pinjisakikool, The Inuence of Personality Traits on Households’ Financial Risk
Tolerance and Financial Behaviour, Journal of Interdisciplinary Economics 30 (1)
(2018) 32–54, https://doi.org/10.1177/0260107917731034.
[45] A. Oehler, F. Wedlich, The relationship of extraversion and neuroticism with risk
attitude, risk perception, and return expectations, Journal of Neuroscience,
Psychology, and Economics 11 (2) (2018) 63–92, https://doi.org/10.1037/
npe0000088.
[46] J.B. Hirsh, Personality and environmental concern, Journal of Environmental
Psychology 30 (2) (2010) 245–248, https://doi.org/10.1016/j.jenvp.2010.01.004.
[47] J. Goebel, M.M. Grabka, S. Liebig, M. Kroh, D. Richter, C. Schr¨
oder, et al., The
German Socio-Economic Panel (SOEP), Jahrbücher für National¨
okonomie und
Statistik 239 (2) (2019) 345–360, https://doi.org/10.1515/jbnst-2018-0022.
[48] DIW Berlin. DIW Berlin: Über uns. [May 13, 2018]; Available from: http://www.
diw.de/de/diw_02.c.221178.de/ueber_uns.html#299767.
[49] D. Richter J. Rohrer M. Metzing W. Nestler M. Weinhardt J. Schupp SOEP Scales
Manual (updated for SOEP-Core v32.1) 2017 Berlin.
[50] M. Dehne, J. Schupp, Pers¨
onlichkeitsmerkmale im Sozio-oekonomischen Panel
(SOEP) - Konzept, Umsetzung und empirische Eigenschaften, Berlin, 2007.
[51] R. Mata, R. Frey, D. Richter, J. Schupp, R. Hertwig, Risk Preference: A View from
Psychology, The Journal of Economic Perspectives 32 (2) (2018) 155–172, https://
doi.org/10.1257/jep.32.2.155.
[52] H. Schildberg-H¨
orisch, Are Risk Preferences Stable? The Journal of Economic
Perspectives 32 (2) (2018) 135–154, https://doi.org/10.1257/jep.32.2.135.
[53] L.K. Muth´
en, B. Muth´
en, Mplus User’s Guide: Version 8, 8th ed., Muth´
en &
Muth´
en, Los Angeles, CA, 1998–2020..
[54] B.O. Muth´
en, L.K. Muth´
en, T. Asparouhov, Regression and mediation analysis
using Mplus, Muth´
en & Muth´
en, Los Angeles, CA, 2016.
[55] H.W. Marsh, O. Lüdtke, B. Muth´
en, T. Asparouhov, A.J.S. Morin, U. Trautwein, et
al., A new look at the big ve factor structure through exploratory structural
equation modeling, Psychol Assess 22 (3) (2010) 471–491, https://doi.org/
10.1037/a0019227.
[56] T. Asparouhov, B. Muth´
en, Exploratory Structural Equation Modeling, Structural
Equation Modeling: A Multidisciplinary Journal 16 (3) (2009) 397–438, https://
doi.org/10.1080/10705510903008204.
[57] J. Aichholzer, Einführung in lineare Strukturgleichungsmodelle mit Stata, Springer
VS, Wiesbaden, 2017.
[58] X. Zhao, J.G. Lynch, Q. Chen, Reconsidering Baron and Kenny: Myths and Truths
about Mediation Analysis, Journal of Consumer Research 37 (2) (2010) 197–206,
https://doi.org/10.1086/651257.
S. Poier