Soc. Sci. 2015, 4, 313–338; doi:10.

3390/socsci4020313
OPEN ACCESS

social sciences
ISSN 2076-0760
www.mdpi.com/journal/socsci
Article

Awareness of Humanities, Arts and Social Science (HASS)

Research Is Related to Patterns of Citizens’ Community and
Cultural Engagement
Murray A. Rudd

Environment Department, University of York, Heslington, York YO10 5DD, UK;

E-Mail: [email protected]; Tel.: +44-1904-324063; Fax: +44-1904-322998

Academic Editor: Martin J. Bull

Received: 24 November 2014 / Accepted: 22 April 2015 / Published: 30 April 2015

Abstract: Why should societies invest resources in humanities, arts, and social sciences
(HASS) research? While citizens’ quality of life should be affected by the type and level of
cultural amenities they have access to, the broader links between HASS research and its
impacts on quality of life attributes can be tenuous because of the research attribution
challenge, temporally and spatially linking specific HASS research and its ultimate impact
on well-being and society. From a survey of 1920 Canadians, here I report perceived
values, awareness of HASS research, threats to quality of life, and levels of community
and cultural engagement. The key finding of this exploratory study was that HASS
research awareness acted as a powerful predictor of threat perceptions, levels of community
activity, and cultural engagement at the local level. It was not, however, a significant
predictor of core values. From a theoretical perspective, this is in line with a priori
expectations that core values are a precursor to worldviews, threat perceptions, and
behaviors. There are very different policy prescriptions for increasing HASS research
awareness and, by extension, Canadian citizens’ propensity for cultural and physical
engagement, depending on how HASS research awareness affects their threat perceptions,
values, and behavior. They include alternatives that focus on experiential learning early in
life and adult-oriented awareness-building activities. The strong relationship between
HASS research awareness and citizen engagement implies that there are important roles for
education and awareness-building activities beyond simply encouraging future consumption
of cultural commodities among HASS-aware citizens.

Keywords: humanities; arts; social science; engagement; research impact; latent class analysis
Soc. Sci. 2015, 4 314

1. Introduction

Why should societies invest resources in humanities, arts, and social sciences (HASS) research?
That question is increasingly important for public sector policy makers and funding agencies given
competing demands for scarce financial resources. While citizens’ quality of life should intuitively be
affected by the cultural amenities they have access to, the links between HASS research and impacts
can be tenuous because of the research attribution challenge, temporally and spatially linking specific
research activities and their ultimate impacts on quality of life and society well-being (e.g., [1–3]).
Furthermore, there are multiple conceptions of the types of value with which HASS research supplies
society [3,4], and not all are amenable to current forms of impact evaluation [5]. Adding further
complexity, the need for engagement of HASS scholars in transdisciplinary environmental change
research has also been increasingly recognized [6–9], as has the potential role of HASS research in
value adjudication in complex environmental management challenges [10]. Cross-disciplinary
engagement on broad societal challenges beyond the traditional HASS domain is still largely lacking
(e.g., [11]) but as levels of cross-disciplinary integration increase to address pressing environmental
challenges, attribution will only become a more vexing issue.
It is already challenging to build a “business case” to fund and sustain HASS research, especially
for governments that are focused on “value for money” (i.e., looking for outputs that contribute to
technological development and economic growth) and academic institutions shaped by funding tied to
impact evaluation frameworks [5,12]. HASS research has been viewed as not being competitive with
other types of science, technology, engineering, and medical (STEM) research [13,14]. STEM research,
which may compete for the same pool of limited public funding, has a much more established track
record regarding its economic impact [15–17] and is also often viewed as central to national
innovation policy [18]. As the Australian Council for the Humanities, Arts and Social Sciences
emphasized [19], in the public eye, the perception is that economic and social benefits are derived
mostly from (natural and applied) scientific research and that, until that situation changes, HASS
research will be undervalued by society. One approach to demonstrating the value of HASS research
has been to quantify the economic impact of the arts and culture industries [1,20,21]. Given the public
good nature of many of the quality of life attributes to which HASS research contributes, it may also
be possible to quantify the nonmarket economic benefits of changes in specific attributes [22–25]. As
in environmental economics [26], nonmarket values for HASS impacts can be used in cost-benefit
analyses (CBA) and may potentially tip the balance in favor of HASS investment. As in environmental
economics, there are, however, risks with commodification and “playing the CBA game”. Because it is
based on narrow satisfaction of consumer-oriented preferences [27], “valued” and “valuable” attributes
(this goes back to Dewey, who argued that to be valuable something needs to be subject to critical
evaluation and reflection [28]) can get confused and may not be appropriate when attributes being
valued occur at different geographic and temporal scales [29], and can be co-opted by governments
with particular ideological agendas [30]. There have also been efforts to position HASS-based
creativity as a vital input in the commercial innovation process [31], thus making HASS research more
relevant for development of market goods and services. Still, viewing HASS research through an
innovation lens is not without its problems [12,32].
Soc. Sci. 2015, 4 315

HASS research is often oriented towards criticism, broadening thinking, maintaining a diversity of
ideas in society, and encouraging deliberation about what constitutes value and knowledge [32–34].
It is possible that valuable spin-off benefits may also arise as a result of citizens’ direct engagement in
artistic and cultural activities. Benefits from increased levels of citizen engagement could include
factors such as increased personal life satisfaction and happiness, improved physical and mental
health, expanded social networks, and more resilient and vibrant communities and societies [6,35–40].
As Galloway ([41], p. 339) noted, however, “this is an area where policy rhetoric outstrips research
evidence”. While HASS research might build society’s capacity for undertaking action [3] and decrease
vulnerability (i.e., by increasing adaptive capacity within households and societies), the assessment of
the economic benefits arising from capacity-building is currently beyond accounting-oriented logic
used in most government evaluation frameworks.
In order to adequately assess the influence of HASS research across a wide spectrum of social and
economic impacts thus requires an improved understanding of the contributions HASS research makes
to societal well-being—the attribution problem—as well as a diversity of methodological approaches
that capture the range of those benefits [42]. In this paper I report results from an exploratory study
(n = 1920) on the relationships between Canadians’ values, awareness of HASS research, perceived
threats to quality of life, and levels of community and cultural engagement. This topic should be of
broad interest in other regions because increasing our understanding of the antecedents of cultural
participation and consumption are rather poorly understood and under-researched internationally [43],
yet have important policy implications. The results of this study show that HASS research awareness is
strongly related to threat perceptions, participation in cultural activities, and levels of community
engagement. Furthermore, it is a stronger predictor of those factors than educational attainment or
other standard demographic covariates. This suggests that the impact of HASS research amongst the
general public may be much broader than typically thought and that research awareness may be a
measurable proxy for a deeper variable that broadly affects active citizenship and engagement.

2. Data Generation and Analysis

2.1. Survey Design and Delivery

The focus of this paper is on four specific questions that were part of a national survey that explored
willingness to pay for improved quality of life attributes for Canadians [24] (additional details on the
rationale for survey design and its implementation are available in that article). In the four questions
that are the current focus, survey respondents were asked to: (1) rate their level of agreement with a
series of 15 value statements; (2) rate 12 potential threats to quality of life in Canada; (3) indicate
whether they were members of any of seven types of community-based clubs or organizations; and (4)
indicate whether they participated in any of six particular types of social or cultural activities over the
past 12 months.
I used a slight modification of value-belief-norm (VBN) theory [44] to inform the survey design
(Figure 1). VBN theory hypothesizes that actual behavior depends on a number of different factors,
including: core values (enduring perspectives that last much of a person’s life and are heavily
influenced by culture and upbringing); worldviews (relatively more malleable perspectives on how the
Soc. Sci. 2015, 4 316

world works); threat perceptions; behavioral intent; and realized behavior. Threat salience can be
viewed as a function of worldview, a person’s education and awareness, and access to information.
Behavioral intentions are not always realized because various material and social factors can constrain
intended behavior. While a worldview-oriented scale based loosely on the New Ecological Paradigm
(NEP) [45] was developed in Australia [43], I chose to use only the Schwartz-based items [44,46] for
reasons of theory (it captures stable, core values rather than more fluid worldviews) and practicality
(negative reaction from focus groups regarding how seriously they took NEP-style questions in this
study and in prior environmental valuation surveys).

Research Access to
Age, gender, awareness information
income,
education

Demographic Knowledge / Behavioral

variables expertise constraints

Cultural Experience Beliefs / Threat Behavioral Actual

experiences variables worldview salience intent behavior

Values and
ethics

Perceived Activities,
Core values quality of life memberships,
threats actions

Figure 1. Causal linkages between threat perceptions and their precursors, and intended
and realized behavior in a modified value-belief-norm theoretical framework.

Respondents were asked to rate how important each of 15 value statements was as a guiding
principle in their lives. Of the 15 statements, twelve were taken directly from the Schwartz “brief
scale” [46], a subset of the original 56-item instrument [47,48]. I modified three environment-specific
statements [46] to form culturally oriented survey rating tasks. I first asked respondents to rate each
statement on a scale of 1 to 7, whereby 1 represented not at all important and 7 was extremely
important. If a respondent rated an item as not at all important, a follow-up question asked if the
statement was simply not relevant or whether it was actually in opposition to their values.
The 15 statements used were: (1) protecting and preserving culture and heritage (hereafter referred
to as Culture); (2) a creative world, rich in arts and cultural activity (Arts); (3) respecting people of the
world, harmony with other cultures (Harmony); (4) a world at peace, free of war and conflict (Peace);
(5) social justice, correcting injustice, care for the weak (Justice); (6) equality, equal opportunity for all
(Equality); (7) honoring parents and elders, showing respect (Heritage); (8) family security, safety for
loved ones (Family security); (9) self-discipline, self-restraint, resistance to temptation (Self-discipline);
(10) authority, the right to lead or command (Authority); (11) influence, having an impact on people
and events (Influence); (12) wealth, material possessions, money (Wealth); (13) a varied life, filled
with challenge, novelty and change (Variety); (14) an exciting life, stimulating experiences
(Stimulation); and (15) curious, interested in everything, exploring (Curiosity).
A capital asset approach to quality of life recognizes that individuals, organizations, and society as a
whole possess or have access to stocks of cultural capital assets and services (flows) that can enhance
Soc. Sci. 2015, 4 317

quality of life [6,49,50]. Six types of assets—natural, manufactured, human, social, cultural, and
financial capital—were used as themes to group indicators. Two potential threats for each of the six
main types of capital assets were presented and respondents were asked how much of a threat those
factors posed to quality of life in Canada. The 12 threats included: (1) global warming/climate change
(Climate); (2) deterioration of water resource quality and/or quantity (groundwater, rivers, lakes,
wetlands) (Water); (3) unforeseen consequences of rapid technological change (e.g., genetic engineering,
nanotechnology) (Technology); (4) deterioration of Canadian public infrastructure such as roads,
bridges, communications, hospitals, schools (Infrastructure); (5) human health problems relating to
unhealthy lifestyle choices and/or environmental contaminants (Health); (6) declining standards in
education and Canada’s capacity for creativity and innovation (Innovation); (7) declining safety and
security at the community, national, and/or international level (Security); (8) stress on, and the
breakdown of, Canadian families (Families); (9) loss of traditional ways of life, culture, and heritage
(Heritage); (10) a decline in creativity, artistic endeavors, and the cultural industries in Canada
(Culture); (11) globalization and its impact on the Canadian economy (Globalization); and (12) high
prices, unaffordable housing, and losses in real purchasing power for Canadians (Economy).
Survey respondents were asked to indicate whether they were members of any of the selected seven
types of clubs or organizations, including: (1) community service club or organization (e.g., Rotary,
Kinsmen, etc.); (2) environmental or natural history organization; (3) outdoor recreation or fitness club
(e.g., running, hiking clubs); (4) community performing arts organization; (5) political party (i.e.,
“card-carrying member”); (6) internationally oriented club or organization (e.g., Amnesty International,
Oxfam); and (7) local sports club or team.
Respondents were also asked whether they had participated in any of the listed six types of social or
cultural activities over the past 12 months: (1) attended a local church service; (2) attended a concert,
play or musical event; (3) visited a local museum or heritage site; (4) purchased a book written by a
Canadian author; (5) purchased a piece of art by a Canadian artist; and (6) attended a public talk or
lecture at a local college or university.
Standard demographic data collected included age category, income category, gender, marital
status, educational attainment, citizenship, primary languages spoken, and residential postal code. Postal
codes were used to assign respondent’s place of residence to five regions: Atlantic Canada
(Newfoundland and Labrador, Nova Scotia, Prince Edward Island and New Brunswick); Quebec;
Ontario; Prairies (Manitoba, Saskatchewan and Alberta, including Northwest Territories); and British
Columbia (including Yukon). Information was also collected on the version (English or French)
completed, survey completion time (minutes), whether respondents were “protesters” in the
accompanying valuation survey [24] (i.e., they would not support hypothetical investments that
improved Canadian quality of life for reasons such as their distrust of government), and self-reported
awareness of HASS research in Canada.
A HASS awareness indicator was constructed by aggregating scores from four questions regarding
respondents’ awareness of the breadth and scope of HASS research in Canada prior to starting this
survey. The questions related to: (1) what research is and how it works; (2) the three Canadian research
councils that fund research activities nationally; (3) the size of Canadian public and private sector
investments in arts, humanities and social sciences research and training; and (4) the diversity of arts,
humanities, and social science research in Canadian universities (full wording for the questions is
Soc. Sci. 2015, 4 318

included in supplementary material S1). Responses were coded as 1 (not at all aware); 2 (somewhat
aware); and 3 (very aware) and were summed across the four categories, giving a range of scores from 4
(completely unaware of research) to 12 (extremely high level of awareness of HASS research).
Cronbach’s alpha was 0.696, indicating that internal consistency among the four items was acceptable.
An international market research firm, GMI, was contracted to supply a sample, drawn from an
Internet panel of 140,000+ households generally representative of the Canadian population. The
internet survey was designed and implemented with Sawtooth Software’s SSI interviewing software.

2.2. Data Analysis

Latent class (LC) cluster analysis can be used to identify unobserved, or latent, class
membership based on observed variables or indicators that imperfectly measure underlying true class
membership [51,52]. Functionally, LC clustering reduces dataset complexity by identifying a low
number of proxy variables that are predictive of a larger number of dependent indicator variables.
Let X represent the latent variable and Yl one of the L observed indicator variables, where 1 ≤ l ≤ L,
C be the number of latent classes, and Dl be the number of levels of Yl. A particular latent class is
enumerated by an index x, where x = 1, 2, ..., C, and a particular value of Yl by yl, where yl = 1, 2, ...,
Dl. The probability of obtaining any specific response pattern y, P(Y = y), is a weighted average of the
C class-specific probabilities P(Y = y | X = x).

P Y=y = P X = x Y = y|X = x (1)

where P(X = x) denotes the proportion of respondents belonging to latent class x. Traditional LC
analysis can be generalized from its assumption that each observation is a mutually independent
member of only one LC when latent variable X is included as an additional dimension in a log-linear
model of P(X = x | Y = y) [51,53,54]. Latent Gold software was used to estimate all LC models;
Vermunt and Magidson [55] provide technical details on the maximum likelihood estimation
methodology. In this study, all latent classes were defined solely on response patterns;
sociodemographic covariates were not used for LC classification.
Bayesian Information Criteria (BIC = −2L + [LogN][npar], where N = sample size, L = log-likehood
and npar = number parameters) was used to initially identify the model that was most parsimonious [56].
Local independence between indicators was tested in the initial model using the bivariate residual
(BVR) Pearson χ2 statistic. Functionally, a significant BVR statistic (i.e., χ2 > 3.84 for 5% significance
level) indicated a degree of redundancy between indicators. When there were significant interactions
between indicators, I deleted indicators from the LC cluster models until all significant interactions
were eliminated. Deleting indicators does not mean that they are unimportant to respondents, only that
the information that they provide does not help differentiate clustering patterns. Posterior class
membership probabilities from the LC analyses were subsequently used as dependent variables in the
demographically based CHAID segmentation.
The Chi-Squared Automatic Interaction Detection (CHAID) segmentation technique is an
exploratory tree-based segmentation process that separates a sample into mutually exclusive and
exhaustive subgroups [57,58]. Chi-squared goodness of fit tests were used to identify significant
Soc. Sci. 2015, 4 319

predictors of LC membership patterns and merge predictor categories that did not differ in their
prediction of the dependent variables. The CHAID procedure aids the development of predictive
models by screening out extraneous variables and generating interpretable diagrams of mutually exclusive
population segments with similar sociodemographic or survey-specific characteristics [56,58,59].
Predictors with the lowest Bonferroni-adjusted p value defined sub-groupings at each branch of the
CHAID dendogram. Throughout the balance of the paper, I refer to first-stage latent class partitions
as “classes” and the second-stage CHAID groupings as “segments”. I used SI-CHAID software [60]
for all analyses.

3. Results and Discussion

3.1. Respondent Demographics

A total of 3555 people visited the survey website over an eight-day period in March 2008. In total,
1920 valid completed surveys (n = 266 French respondents; n = 1654 English respondents) were
available for the final analysis (54.0% of respondents that visited the survey website). Only surveys in
which all fields were completed, including all valuation questions, were used for this analysis.
Province of residence was aggregated into five geographic areas: Atlantic Canada (n = 246),
Quebec (n = 331), Ontario (n = 807), Prairie Provinces (n = 292), and British Columbia (“BC”, n = 244).
There were significant differences in region of residence between the Canadian population (>15 yrs)
and the sample (χ2 = 126.4, 4 d.f., p < 0.0001), with under-representation of Quebec, Prairie Provinces,
and BC respondents, and over-representation of Atlantic Canada and Ontario respondents.
Gender (χ2 = 16.9, 1 d.f., p < 0.0001) and age category (χ2 = 337.5, 3 d.f., p < 0.0001) for the
sample were also significantly different than those for the population, with younger respondents and
females over-represented. As a result, 40 sampling weights (five regions × four age classes × two genders)
were used in all subsequent analyses to correct for imbalances in the sample.
Other demographic covariates used in the CHAID analyses included marital status (654 single,
1073 married or common law, 193 divorced, separated, or widowed), language most often spoken at
home (1495 primarily English, 290 primarily French, 71 fully bilingual, 64 other languages), educational
attainment (425 high school graduation or less, 962 some college or university, 359 bachelor’s degree
or equivalent graduate, 174 with graduate-level education), and citizenship (1602 born in Canada,
203 naturalized Canadian citizens, 115 non-Canadian citizens).
The 108 respondents designated as “protestors” in the accompanying valuation analysis [24] were
flagged with a dummy variable. Average time to complete the survey was 64.4 min, while median time
was 27.5 min. For both the LC and CHAID analyses, survey completion time was grouped into three
categories (fast 15.4 min average; medium 27.8 min; and slow 150.2 min) and treated as a nominal
variable due to the upward skew in average completion times.

3.2. Cultural Values

Respondents were presented with a series of 15 statements that they were asked to rate as to their
importance as a guiding principle in the respondent’s life (Table 1). Models with two to 13 LCs were
estimated. A 12-class LC cluster model minimized BIC and was chosen for further refinement.
Soc. Sci. 2015, 4 320

Numerous BVRs were significant at the 5% level, indicating that there was substantial redundancy
amongst the indicators with regards to the information they provided relevant for cluster discrimination.
Nine indicators were dropped sequentially (arts, harmony, culture, wealth, stimulation, respect, authority,
variety, and influence) to eliminate all significant BVRs. With the reduced indicator set, BIC was
minimized using an eight-class model. The final model (n = 1920 respondents, 91 parameters) performed
well (LL = −16,281.7, entropy R2 = 0.6747, classification error = 20.3%), cleaving the national sample
into eight distinct LCs based on response patterns for six of the 15 Schwartz short scale items. None of
the three culturally oriented indicators were retained as explanatory indicators of core values.

Table 1. Perceived importance of modified Schwartz brief scale value statements.

Perceived Importance a of Factors as Guiding Principles
Threat 0 1 2 3 4 5 6 7
Family security b 3 12 22 38 145 243 468 989
Peace b 7 19 26 78 259 365 427 739
Equality b 15 20 35 68 239 414 468 661
Respect 3 21 33 69 235 373 535 651
Justice b 9 14 24 93 262 448 503 567
Harmony 15 36 65 109 348 491 460 396
Curiosity b 2 24 45 128 356 567 428 370
Self-discipline b 6 25 68 177 443 531 370 300
Stimulation 3 29 43 121 384 567 478 295
Culture 9 75 111 211 469 479 322 244
Variety 2 36 51 157 438 570 427 239
Arts 10 91 143 285 512 469 234 176
Influence 9 44 91 214 556 526 316 164
Wealth 22 64 145 260 483 502 283 161
Authority 19 85 140 313 564 450 210 139
a
0 = opposed to personal guiding principles; 1 = not at all important to 7 = extremely important; b indicator
variables retained in latent class cluster analysis.

Figure 2 illustrates the differences between the LCs. For example, cluster VAL LC01 (column 1),
the single largest class from the sample, was composed of respondents who viewed peace, justice,
equality, and family security as important to extremely important. Clusters LC03 and LC05 illustrate
the power of LC cluster analysis to differentiate response patterns. Respondents in both clusters almost
all viewed peace, justice, and family as extremely important but, unlike respondents in VAL LC03,
respondents in VAL LC05 view equality as much less important.
Soc. Sci. 2015, 4 321

Values Cluster Values Cluster Values Cluster Values Cluster Values Cluster Values Cluster Values Cluster Values Cluster
VAL LC01 VAL LC02 VAL LC03 VAL LC04 VAL LC05 VAL LC06 VAL LC07 VAL LC08
n = 737 (38.4%) n = 549 (28.6%) n = 286 (14.9%) n = 214 (11.1%) n = 50 (2.6%) n = 38 (2.0%) n = 31 (1.6%) n = 15 (0.8%)
100% 100% 100% 100% 100% 100% 100% 100%

75% 75% 75% 75% 75% 75% 75% 75%

Peace 50% 50% 50% 50% 50% 50% 50% 50%

25% 25% 25% 25% 25% 25% 25% 25%

0% 0% 0% 0% 0% 0% 0% 0%
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

100% 100% 100% 100% 100% 100% 100% 100%

75% 75% 75% 75% 75% 75% 75% 75%

Justice 50% 50% 50% 50% 50% 50% 50% 50%

25% 25% 25% 25% 25% 25% 25% 25%

0% 0% 0% 0% 0% 0% 0% 0%
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

100% 100% 100% 100% 100% 100% 100% 100%

75% 75% 75% 75% 75% 75% 75% 75%

Equality 50% 50% 50% 50% 50% 50% 50% 50%

25% 25% 25% 25% 25% 25% 25% 25%

0% 0% 0% 0% 0% 0% 0% 0%
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

100% 100% 100% 100% 100% 100% 100% 100%

75% 75% 75% 75% 75% 75% 75% 75%

Family 50% 50% 50% 50% 50% 50% 50% 50%

25% 25% 25% 25% 25% 25% 25% 25%

0% 0% 0% 0% 0% 0% 0% 0%
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

100% 100% 100% 100% 100% 100% 100% 100%

75% 75% 75% 75% 75% 75% 75% 75%

Discipline 50% 50% 50% 50% 50% 50% 50% 50%

25% 25% 25% 25% 25% 25% 25% 25%

0% 0% 0% 0% 0% 0% 0% 0%
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

100% 100% 100% 100% 100% 100% 100% 100%

75% 75% 75% 75% 75% 75% 75% 75%

Curiosity 50% 50% 50% 50% 50% 50% 50% 50%

25% 25% 25% 25% 25% 25% 25% 25%

0% 0% 0% 0% 0% 0% 0% 0%
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

Figure 2. Value-based cluster summary. X-axis labels: 0 = opposed to item as a personal guiding principle; 1 = not at all important to 7 =
extremely important.
Soc. Sci. 2015, 4 322

In the subsequent CHAID analysis, age was the most powerful predictor of LC class membership
patterns for the eight value-based LCs identified (χ2 = 73.23, 14 d.f., p = 3.9 × 10−8). Three sub-groups,
consisting of respondents <40 yrs (weighted n = 804), 40–54 yrs (n = 515), and 55+ yrs (n = 601),
showed clear patterns of increasing (VAL LC01, LC03, LC05) and decreasing (VAL LC02, LC04)
membership with increasing respondent age (Figure 3). For example, membership in VAL LC02 fell
from 32.3% to 23.0% with increasing age. The CHAID dendogram split further along survey
completion time for the youngest group and along gender lines for two older groups. Membership was
higher in VAL LC01, LC03, and LC05 for older females, and higher in VAL LC02 and LC03 for older
males. Older male responders generally seemed to be intermediate in their membership between the
<40 yr fast responders and 55+ yr females, the two groups with the most distinctively different value
cluster memberships. Note that HASS research awareness was not a significant predictor of any core
value membership patterns.

Sample
VAL LC01 (%) 38.41 “Balanced, all important“
VAL LC02 (%) 28.59 “Balanced, all somewhat important“
VAL LC03 (%) 14.92 “Extreme importance except discipline, curiosity“
VAL LC04 (%) 11.12 “Middle of scale“
VAL LC05 (%) 2.60 “Peace, justice, family extremely important“
VAL LC06 (%) 1.99 “Peace, equality, curiosity priorities“
VAL LC07 (%) 1.61 “Family extremely important, opposed to equality“
VAL LC08 (%) 0.76 “Most unimportant“
n = 1920

χ 2 = 73.23, 14 d.f.
p = 3.9 x 10–8
Age (yrs) < 40 40-54 55+
VAL LC01 (%) 35.67 38.84 41.73
VAL LC02 (%) 32.28 29.34 23.00
VAL LC03 (%) 10.82 17.32 18.35
VAL LC04 (%) 14.84 9.67 7.39
VAL LC05 (%) 1.16 1.88 5.15
VAL LC06 (%) 2.17 2.07 1.69
VAL LC07 (%) 1.82 0.89 1.93
VAL LC08 (%) 1.15 0.00 0.76
n = 804 n = 515 n = 601
χ 2 = 43.89, 7 d.f. χ 2 = 14.56, 7 d.f. χ 2 = 19.84, 7 d.f.
p = 6.7 x 10–7 p = 0.042 p = 0.006
Survey
Completion Time Fast Med-Slow Gender Female Male Female Male
VAL LC01 (%) 29.98 39.52 43.05 34.80 45.48 38.66
VAL LC02 (%) 34.54 30.75 26.17 32.38 17.17 27.77
VAL LC03 (%) 8.08 12.67 18.73 15.96 23.11 14.47
VAL LC04 (%) 20.61 10.94 6.47 12.73 3.02 10.96
VAL LC05 (%) 0.62 1.52 3.07 0.74 8.13 2.71
VAL LC06 (%) 1.54 2.60 2.03 2.11 0.83 2.39
VAL LC07 (%) 2.27 1.51 0.49 1.28 1.10 2.61
VAL LC08 (%) 2.36 0.50 0.00 0.00 1.16 0.44
n = 325 n = 480 n = 252 n = 263 n = 270 n = 331
Segment Segment Segment Segment Segment Segment
VAL SG01 VAL SG02 VAL SG03 VAL SG04 VAL SG05 VAL SG06

Figure 3. Value-based CHAID segmentation. Bold indicates the highest single value for latent
class membership; underlined indicates the single lowest value for latent class membership.

Six demographically distinct sample segments (terminal nodes) were identified, each of which
can be described according to exogenous predictors and their LC membership patterns (Table 2). The
six segments were ranked by index scores that compared membership rate for a single segment to
Soc. Sci. 2015, 4 323

membership rate for the sample as a whole. Index scores were calculated by comparing membership
proportions for LC clusters and the overall sample. For example, in CHAID segment 1 (VAL SG01),
34.5% of respondents were from latent class VAL LC02 versus 28.6% for the overall sample (see
Figure 3), giving an index score of 121 (=34.5/28.6) and signifying that members of this segment were
21% more likely than average to belong to the “balanced, all somewhat important” latent class.
Index scores provide insight into the relative level and diversity of LC membership across CHAID
segments and allow rankings of LC membership across CHAID segments.

Table 2. Summary of segmentation of value-based latent class clusters.

VAL VAL VAL VAL VAL VAL
SEG 01 SEG 02 SEG 03 SEG 04 SEG 05 SEG 06
Segment size (n) 325 480 252 263 270 331
Proportion of sample (%) 16.9 25.0 13.1 13.7 14.0 17.2
Segment characteristics
Age (yrs) <40 <40 40–54 40–54 55+ 55+
Gender - - Female Male Female Male
Survey completion time a 1 2–3 - - - -
Latent class membership (%)
Balanced, all important 29.98 39.52 43.05 34.80 45.48 38.66
Index 78 103 112 91 118 101
Rank 6 3 2 5 1 4
Balanced, all somewhat important 34.54 30.75 26.17 32.38 17.17 27.77
Index 121 108 92 113 60 97
Rank 1 3 5 2 6 4
Extremely important except discipline, curiosity 8.08 12.67 18.73 15.96 23.11 14.47
Index 54 85 126 107 155 97
Rank 6 5 2 3 1 4
Middle of scale 20.61 10.94 6.47 12.73 3.02 10.96
Index 185 98 58 114 27 99
Rank 1 4 5 2 6 3
Peace, justice, family extremely important 0.62 1.52 3.07 0.74 8.13 2.71
Index 24 58 118 29 313 104
Rank 6 4 2 5 1 3
Peace, equality, curiosity important 1.54 2.60 2.03 2.11 0.83 2.39
Index 77 130 102 106 42 120
Rank 5 1 4 3 6 2
Family extremely important, opposed to equality 2.27 1.51 0.49 1.28 1.10 2.61
Index 141 94 30 79 69 163
Rank 2 3 6 4 5 1
Most unimportant 2.36 0.50 0.00 0.00 1.16 0.44
Index 309 65 0 1 152 57
Rank 1 3 6 5 2 4
a
Time: 1 = fast < 21.2 min (mean = 15.4); 2 = medium 21.2 to 35.9 min (mean = 27.8); 3 = slow 36+ min
(mean = 150.2); bold indicates the highest single value for latent class membership; underlined indicates the
lowest value for latent class membership).

Each segment was described in terms of the demographic profile of its members and their LC
membership patterns. Segment 1, for example, which was comprised of younger respondents who
Soc. Sci. 2015, 4 324

completed the survey quickly, had the highest index scores for latent class VAL LC02 (“balanced, all
somewhat important”), VAL LC04 (“middle of scale”), and VAL LC08 (“most unimportant”)
memberships (the highest index scores are bolded in all CHAID diagrams). On the other hand, this
group also had the lowest index scores for latent classes VAL LC01 (“balanced, all important”), VAL
LC03 (“extreme importance except discipline, curiosity”), and LC05 (“peace, justice, family extremely
important”) memberships.

3.3. Threat Perceptions

Respondents were presented with 12 potential threats to Canadian quality of life and were asked to
rate their importance (Table 3). Threats to the economy and environment were viewed as the most
important by respondents, while threats to heritage and culture, in addition to threats due to
technological advances, were perceived as having relatively low importance.

Table 3. Perceived importance of potential threats to Canadian quality of life.

Perceived Importance of Potential Threats to Quality of Life in Canada
Threat Very Low Low Moderate High Very High No Opinion
Water 49 136 350 547 804 34
Economy a 45 125 384 552 790 24
Climate a 148 202 383 418 733 36
Health a 60 159 480 591 609 21
Infrastructure a 60 187 467 638 537 31
Families a 112 235 521 516 502 34
Innovation a 100 259 570 519 429 43
Security a 126 344 636 465 301 48
Globalization a 167 306 661 446 259 81
Heritage a 304 436 528 354 251 47
Technology a 303 469 588 320 149 91
Culture a 420 543 566 232 86 73
a
Indicator variables retained in latent class cluster analysis to define classes.

A 5-class LC cluster model initially minimized BIC and was chosen for further refinement. Only
one BVR was significant, indicating that there was low overlap in clustering-relevant information that
indicators provided. Water was dropped from the model to eliminate all significant BVRs. With the
reduced indicator set, BIC was minimized using a five-class final model that performed very well
(n = 1920 respondents, 91 parameters; LL = −29,986.1, entropy R2 = 0.7720, classification error = 11.8%).
Figure 4 illustrates the indicator breakdowns.
Soc. Sci. 2015, 4 325

Threats Cluster Threats Cluster Threats Cluster Threats Cluster Threats Cluster
TH LC01 TH LC02 TH LC03 TH LC04 TH LC05
n = 859 (44.8%) n = 531 (27.7%) n = 307 (16.0%) n = 170 (8.9%) n = 53 (2.8%)
100% 100% 100% 100% 100%

75% 75% 75% 75% 75%

Climate Change 50% 50% 50% 50% 50%

25% 25% 25% 25% 25%

0% 0% 0% 0% 0%
-2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK

100% 100% 100% 100% 100%

75% 75% 75% 75% 75%

Health 50% 50% 50% 50% 50%

25% 25% 25% 25% 25%

0% 0% 0% 0% 0%
-2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK

100% 100% 100% 100% 100%

75% 75% 75% 75% 75%

Innovation 50% 50% 50% 50% 50%

25% 25% 25% 25% 25%

0% 0% 0% 0% 0%
-2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK

100% 100% 100% 100% 100%

75% 75% 75% 75% 75%

Security 50% 50% 50% 50% 50%

25% 25% 25% 25% 25%

0% 0% 0% 0% 0%
-2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK

100% 100% 100% 100% 100%

75% 75% 75% 75% 75%

Families 50% 50% 50% 50% 50%

25% 25% 25% 25% 25%

0% 0% 0% 0% 0%
-2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK

100% 100% 100% 100% 100%

75% 75% 75% 75% 75%

Heritage 50% 50% 50% 50% 50%

25% 25% 25% 25% 25%

0% 0% 0% 0% 0%
-2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK

100% 100% 100% 100% 100%

75% 75% 75% 75% 75%

Arts and Culture 50% 50% 50% 50% 50%

25% 25% 25% 25% 25%

0% 0% 0% 0% 0%
-2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK

100% 100% 100% 100% 100%

75% 75% 75% 75% 75%

Globalization 50% 50% 50% 50% 50%

25% 25% 25% 25% 25%

0% 0% 0% 0% 0%
-2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK

100% 100% 100% 100% 100%

75% 75% 75% 75% 75%

Economy 50% 50% 50% 50% 50%

25% 25% 25% 25% 25%

0% 0% 0% 0% 0%
-2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK

100% 100% 100% 100% 100%

75% 75% 75% 75% 75%

Technology 50% 50% 50% 50% 50%

25% 25% 25% 25% 25%

0% 0% 0% 0% 0%
-2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK

100% 100% 100% 100% 100%

75% 75% 75% 75% 75%

Infrastructure 50% 50% 50% 50% 50%

25% 25% 25% 25% 25%

0% 0% 0% 0% 0%
-2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK -2 -1 0 1 2 DNK

Figure 4. Threat-based cluster summary. X-axis scale: −2 = not at all important, to +2 =

very important; DNK = do not know.
Soc. Sci. 2015, 4 326

For latent class TH LC01 (column 1), which comprised 44.8% of the sample, respondents viewed
climate change and threats to the economy as the most important, and threats to the arts and cultural
sector as least important. Cluster TH LC02 showed a very similar pattern of threat perception but their
responses were more skewed to the left of the scale, indicating they viewed all threats as somewhat
less important than TH LC01 respondents. Respondents in latent class TH LC03 exhibited a much
different pattern, with relatively high emphasis on the economy and infrastructure, and a low emphasis
on threats to arts and heritage, climate, and innovation. Respondents in cluster TH LC04 tended to
view all threats as very important (although with slightly less emphasis on arts and culture, heritage,
and technology) and respondents in cluster TH LC05 exhibited a high degree of uncertainty.
HASS research awareness was the most powerful predictor of LC class membership (χ2 = 74.25,
8 d.f., p = 1.9 × 10−11), with a clear breakdown of the sample into three groups (for brevity, Figure 5
shows only the first level of the CHAID tree) comprised of respondents with: no awareness of research
or the HASS research process (score = 4); low to intermediate (score = 5–10) awareness; and high
(score = 11–12) awareness. There is a particularly strong trend for cluster TH LC04 (“all threats
important”) to increase with increasing HASS research awareness.

Sample

TH LC01 (%) 44.76 “Balanced threats, economy focus“

TH LC02 (%) 27.68 “Lower threats, arts not very important“
TH LC03 (%) 15.97 “Arts, heritage not at all important“
TH LC04 (%) 8.85 “All threats important“
TH LC05 (%) 2.75 “High uncertainty about threats“
n = 1920

χ 2 = 74.25, 8 d.f.
p = 1.9 x 10–11

HASS Research
Awareness 4 5-10 11-12
TH LC01 (%) 39.10 46.32 23.06
TH LC02 (%) 26.65 28.22 15.97
TH LC03 (%) 18.59 15.34 23.24
TH LC04 (%) 5.42 8.46 35.54
TH LC05 (%) 10.24 1.66 2.19
n = 239 n = 1627 n = 54

Figure 5. Threat-based sample segmentation (first level of the CHAID dendogram only).
HASS research awareness: 4 = no awareness to 12 = highest awareness.

After the first-level, there were complex splits based on varying demographic and survey-specific
indicators. Thirteen distinct CHAID segments were identified for the national sample (Table 4). For
respondents with the lowest level of research awareness, educational attainment and—for respondents
with more than secondary education—region of residence, were significant predictors of segmentation.
Respondents with low HASS research awareness and low education attainment had the highest level of
uncertainty about threats to quality of life. In segments TH SG02 and TH SG03, there were strong
differences in LC membership by region with the “arts, heritage not at all important” being more
prevalent in Ontario and western Canada relative to Quebec and Atlantic Canada.
The choice experiment “protest” dummy was significant for TH SG07. Over half of the members of
this segment (intermediate HASS research awareness, fast completion time, male, some college
education or above, protesters) belonged to TH LC03, the “arts, heritage not at all important” class.
Soc. Sci. 2015, 4 327

Table 4. Summary of segmentation of threat-based latent class clusters.

CHAID Perceived Threats Segment
TH TH TH TH TH TH TH TH TH TH TH TH TH
SG01 SG02 SG03 SG04 SG05 SG06 SG07 SG08 SG09 SG10 SG11 SG12 SG13
Segment Size (n) 90 72 77 215 40 185 23 158 400 493 114 18 36
Proportion of sample (%) 4.7 3.8 4.0 11.2 2.1 9.6 1.2 8.0 20.8 25.7 5.9 0.9 1.9
Segment characteristics
HASS research awareness a 4 4 4 5–10 5–10 5–10 5–10 5–10 5–10 5–10 5–10 11–12 11–12
Educational attainment b 1 2–4 2–4 - 1 2–4 2–4 1 2–4 - - - -
Survey completion time c - - - 1 1 1 1 2–3 2–3 2–3 2–3 1 2–3
Region d - 1–2 3–5 - - - - - - - - - -
Gender - - - F M M M F F M M - -
Citizenship e - - - - - - - - - 1 2 - -
Protest f - - - - - 1 2 - - - - - -
Latent class membership (%)
Balanced threats, economy focus 47.2 42.0 26.8 45.6 35.2 35.2 28.4 52.3 58.8 45.7 36.5 31.6 18.8
Index 106 94 60 102 79 79 63 117 131 102 81 71 42
Rank 3 6 12 5 9 8 11 2 1 4 7 10 13
Lower threats, arts not important 18.6 34.8 28.6 31.8 14.8 39.9 20.5 19.6 26.2 31.0 28.8 21.7 13.1
Index 67 126 103 115 53 144 74 71 95 112 104 78 47
Rank 11 2 6 3 12 1 9 10 7 4 5 8 13
Arts, heritage not at all important 12.1 8.1 36.0 14.2 35.5 19.9 51.1 10.7 7.4 16.9 17.1 30.2 19.8
Index 76 51 225 89 222 124 320 67 47 106 107 189 124
Rank 10 12 2 9 3 5 1 11 13 8 7 4 6
All threats important 2.3 12.6 2.4 5.0 12.1 2.5 0.0 14.5 7.5 6.3 15.6 10.1 48.2
Index 26 142 27 57 137 29 0 164 857 71 177 114 544
Rank 12 4 11 9 5 10 13 3 7 8 2 6 1
High uncertainty about threats 19.8 2.5 6.2 3.3 2.4 2.5 0.1 2.9 0.0 0.2 1.9 6.4 0.1
Index 720 92 227 121 87 91 2 105 2 8 71 233 4
Rank 1 6 3 4 8 7 12 5 13 10 9 2 11
a
4 = no HASS research awareness to 12 = highest awareness; b Education: 1 = secondary or less, 2 = some college or university, 3 = bachelor’s, 4 = graduate education;
c
Time: 1 = < 21.2 min, 2 = 21.2 to 35.9 min, 3 = 36+ min; d Region: 1 = Atlantic Canada, 2 = Quebec, 3 = Ontario, 4 = Prairie Provinces, 5 = BC; e Citizenship:
1 = Canadian-born, 2 = non-naturalized residents/non-Canadians; f 1 = non-protestor, 2 = protest voter in valuation survey. Bold = 2 highest membership values;
underlined = 2 lowest membership values.
Soc. Sci. 2015, 4 328

3.4. Memberships

Binary yes-no responses regarding participation in seven various types of clubs or organizations
were used for the memberships-based LC cluster analysis. The most popular organizations were
recreation/fitness (n = 408, 21.3%) and sports (n = 403, 21.0%), followed by community service
(n = 278, 14.5%), political (n = 138, 7.2%), internationally oriented (n = 130, 6.8%), environmental
(n = 128, 6.7%), and community arts organizations (n = 109, 5.7%). A four-class LC cluster model
minimized BIC but had one significant BVR due to a service-politics interaction. Rather than drop one
of these indicators, the five-class model, which had no significant BVRs and only marginally higher
BIC, was chosen as the final model (n = 1920 respondents, 39 parameters; LL = −4429.4, entropy
R2 = 0.5194 classification error = 14.4%). Figure 6 shows the indicator breakdowns, which are simpler
due to the binary membership indicators.
MEM LC01 MEM LC02 MEM LC03 MEM LC04 MEM LC05
n = 1433 (74.7%) n = 216 (11.2%) n = 148 (7.7%) n = 109 (5.7%) n = 13 (0.7%)
98.6% 97.5% 98.6%
100% 100% 100% 100% 89.9%
100% 95.8%

77.4% 78.6% 80.5% 80.9%

73.3%
75% 75% 75% 75% 75%

50% 50% 50% 50% 46.4% 50%

36.1%
32.3% 33.8%
28.1% 25.7% 26.8% 31.5%
24.2% 21.8%
25% 25% 25% 25% 25%
13.0% 13.0% 11.2%
6.5% 9.7% 8.2%
5.3% 1.7% 2.1% 3.8%
1.4% 1.6% 0.6% 0.0% 0.0%
0% 0% 0% 0% 0%
SE EV FT AR PO IN SP SE EV FT AR PO IN SP SE EV FT AR PO IN SP SE EV FT AR PO IN SP SE EV FT AR PO IN SP

Figure 6. Membership-based cluster summary. X-axis labels: SE = community service

clubs; EV = environmental; FT = fitness and outdoor recreation; AR = community arts;
PO = political; IN = internationally-oriented; SP = sports.

There were very clear differences between the classes, with 75% of the sample belonging to the
“inactive” latent class MEM LC01. MEM LC02 members were moderately active in all organizations
except for sports teams. MEM LC03 members were fitness and sport focused, while members of MEM
LC04 were most involved with community service clubs. Members of the small MEM LC04 (n = 13)
latent class were active in all organizations.
The first level of the CHAID dendogram for the membership analysis is shown in Figure 7 (the full
chart is available from the author upon request). Again, HASS research awareness was the most
powerful predictor of LC class membership (χ2 = 47.46, 8 d.f., p = 3.5 × 10−6). The survey respondents
showed a sharp trend of declining inactivity with increasing HASS research awareness.
The low-medium awareness group (score = 6–7) split further on marital status, with higher MEM
LC02 membership amongst single and married respondents. The group with medium to high HASS
awareness (score = 8–12) could be further cleaved according to educational attainment and, for those
with the highest levels of educational attainment, by citizenship at the 5% significance level. Six
distinct CHAID segments were identified (Table 5). Respondents with the lowest levels of HASS
research awareness (score = 4–5) were largely (82.9%) members of latent class MEM LC01, the
inactive class. They also had the lowest proportion of MEM LC02 (moderately active, no sports) and
MEM LC03 membership (fitness and sports) classes. At the other end of the spectrum, respondents
who had a high level of HASS research awareness and educational attainment, and were not born in
Canada had the highest level of membership in MEM LC03 and LC05 (full spectrum).
Soc. Sci. 2015, 4 329

Sample

MEM LC01 (%) 74.68 “Inactive“

MEM LC02 (%) 11.23 “Moderately active, no sports“
MEM LC03 (%) 7.73 “Fitness and sports“
MEM LC04 (%) 5.68 “Local community service orientation“
MEM LC05 (%) 0.69 “Full spectrum“
n = 1920

Χ 2 = 47.46, 8 d.f.
P = 3.5 x 10–6

HASS Research
Awareness 4-5 6-7 8-12
MEM LC01 (%) 82.89 73.97 67.81
MEM LC02 (%) 7.78 11.63 13.98
MEM LC03 (%) 5.30 7.85 9.86
MEM LC04 (%) 3.80 6.35 6.61
MEM LC05 (%) 0.23 0.20 1.74
n = 565 n = 757 n = 598

Figure 7. Membership-based sample segmentation (first level of the CHAID dendogram

only; HASS research awareness: 4 = no awareness to 12 = highest awareness).

Table 5. Summary of segmentation of membership-based latent class clusters.

CHAID Membership Segment
MEM MEM MEM MEM MEM MEM
SG01 SG02 SG03 SG04 SG05 SG06
Segment Size
n 565 316 441 384 163 51
% of total sample 29.43 16.46 22.97 20.00 8.49 2.66
Segment characteristics
HASS research awareness a 4–5 6–7 6–7 8–12 8–12 8–12
Marital status b - 1,3 2 - - -
Educational attainment c - - - 1–2 3–4 3–4
Citizenship status d - - - - 1 2
Latent class membership (%)
Inactive 82.89 73.37 74.40 70.53 63.05 62.50
Index 111 98 100 94 84 84
Rank 1 3 2 4 5 6
Moderately active, no sports 7.78 15.67 8.73 14.28 13.51 13.29
Index 69 140 78 127 120 118
Rank 6 1 5 2 3 4
Fitness and sports 5.30 6.73 8.66 8.58 11.83 13.26
Index 69 87 112 111 153 172
Rank 6 5 3 4 2 1
Local community service orientation 3.80 3.89 8.12 6.01 9.47 2.00
Index 67 69 143 106 167 35
Rank 5 4 2 3 1 6
Full spectrum 0.23 0.35 0.09 0.61 2.14 8.95
Index 34 30 13 89 311 1300
Rank 5 4 6 3 2 1
a
HASS research awareness: 4 = no awareness to 12 = highest awareness; b Marital status: 1 = single;
2 = married or equivalent; 3 = divorced, separated, or widowed; c Educational attainment: 1 = secondary or less,
2 = some college or university, 3 = Bachelor level graduate, 4 = graduate education; d Citizenship: 1 =
Canadian born citizens, 2 = non-naturalized residents and non-Canadians; bold indicates the highest single
value for latent class membership; underlined indicates the single lowest value for latent class membership.
Soc. Sci. 2015, 4 330

3.5. Cultural Activities

Binary yes-no responses regarding participation in six various types of social or cultural activities
were used for the cultural activities LC cluster analysis. In the last 12 months, 1052 respondents
(54.8%) had attended a concert or musical event, 850 (44.3%) had visited a museum or heritage site,
815 (42.4%) had purchased a book by a Canadian author, 720 (37.5%) had attended a local church
service, 397 (20.7%) had attended a public lecture, and 248 (12.9%) had purchased a piece of
Canadian art. After deleting Canadian book purchases (two significant BVRs), a three-class LC cluster
model minimized BIC and was chosen as the final model (n = 1920 respondents, 17 parameters;
LL = −5303.6, entropy R2 = 0.4916 classification error = 20.8%). The LC membership breakdown is
shown in Figure 8. The first level of the CHAID dendogram for the activities analysis is shown in
Figure 9 (the full chart is available from the author upon request). HASS research awareness was a
very powerful predictor of LC class membership (χ2 = 138.55, 8 d.f., p = 3.3 × 10−24), with increasing
trends in cultural participation as HASS research awareness increased.
ACT LC01 ACT LC02 ACT LC03
n = 903 (47.0%) n = 795 (41.4%) n = 222 (11.6%)
100% 100% 100% 95.0%
86.6% 88.0%

75% 70.9%
65.2% 75% 75%
57.8%
50% 45.7%
50% 50%
27.2% 30.3%
26.2%
25% 15.6% 25% 25%
10.3%
6.2% 8.1%
3.4%
0% 0% 0%
CH CO MU AR LE CH CO MU AR LE CH CO MU AR LE

Figure 8. Activities-based cluster summary. X-axis labels: CH = attended church service;

CO = attended concert; MU = museum or heritage site visit; AR = purchased Canadian art;
LE = attended public lecture.

Sample
ACT LC01 (%) 47.04 “Partially Engaged“
ACT LC02 (%) 41.39 “Culturally Disengaged“
ACT LC03 (%) 11.57 “Highly Engaged“
n=1920

χ 2 = 138.55, 8 d.f.
p = 3.3 x 10–24

HASS Research
Awareness 4 5 6 7–8 9–12
ACT LC01 (%) 35.12 43.24 49.57 51.14 47.98
ACT LC02 (%) 61.74 51.12 42.25 34.86 25.96
ACT LC03 (%) 3.14 5.65 8.19 14.00 26.07
n=239 n=326 n=420 n=679 n=256

Figure 9. Activities-based segmentation (first level of the CHAID dendogram only).

HASS research awareness: 4 = no awareness to 12 = highest awareness.

The segments with the two lowest levels of research awareness could not be further split but all
other groups with higher HASS awareness could be further divided based on educational attainment,
with a general trend of increasing activity as educational level increased. Eleven demographically
distinct sample segments were identified (Table 6). The lowest rates of participation (i.e., highest ACT
LC02 membership at 61.7%) in cultural activities occurred in segment ACT SG01, the group with the
lowest level of HASS awareness. Membership in the inactive LC02 class fell to 21.6% in ACT SG10,
where respondents had the highest level of HASS research awareness, slow survey completion times,
but lower levels of educational attainment (some college or university or less) than for ACT SG11.
Soc. Sci. 2015, 4 331

Table 6. Summary of CHAID segmentation of activity-based latent class clusters.

CHAID Cultural Activities Segment
ACT ACT ACT ACT ACT ACT ACT ACT ACT ACT ACT
SG01 SG02 SG03 SG04 SG05 SG06 SG07 SG08 SG09 SG10 SG11
Segment Size (n) 239 326 171 165 84 146 337 197 85 61 110
Proportion of sample (%) 12.4 17.0 8.9 8.6 4.4 7.6 17.5 10.3 4.4 3.2 5.7
Segment characteristics
HASS research awareness a 4 5 6 6 6 7–8 7–8 7–8 9–12 9–12 9–12
Educational attainment b - - 1–2 1–2 3–4 1 2 3–4 1–2 1–2 3–4
Gender - - F M - - - - - - -
Survey completion time c - - - - - - - - 1–2 3 -
Latent class membership (%)
Partially engaged 35.1 43.2 47.9 48.8 54.5 43.9 55.0 50.5 49.2 51.5 45.1
Index 75 92 102 104 116 93 117 106 105 109 96
Rank 11 10 7 6 2 9 1 4 5 3 8
Culturally disengaged 61.7 51.1 42.6 48.0 30.2 51.7 32.2 26.9 39.1 21.6 18.2
Index 149 123 103 116 73 125 78 65 94 52 44
Rank 1 3 5 4 8 2 7 9 6 10 11
Highly engaged 3.1 5.6 9.5 3.1 15.3 4.5 12.8 23.2 11.7 26.9 36.7
Index 27 49 82 27 132 39 110 200 101 232 317
Rank 11 8 7 10 4 9 5 3 6 2 1
a
HASS research awareness: 4 = no awareness to 12 = extremely high level of awareness; b Educational attainment: 1 = secondary or less, 2 = some college or university,
3 = Bachelor level graduate, 4 = graduate education; c Time: 1 = fast < 21.2 min (mean = 15.4), 2 = medium 21.2 to 35.9 min (mean = 27.8), 3 = slow 36+ min
(mean = 150.2); bold indicates the two highest values for latent class membership; underlined indicates the two lowest values for latent class membership.
Soc. Sci. 2015, 4 332

3.6. Correlations between Latent Classes

Posterior probabilities were generated for each respondent in the values, threats, memberships, and
activities LC cluster analyses. Respondents were assigned to the cluster for which their probability of
membership was highest and correlations between each of these groups of LCs tested (Spearman rank
correlation). The values-based latent class assignments were significantly correlated with both the
threats-based (rs = 0.15, p < 0.01) and cultural activities-based (rs = 0.08, p < 0.01) latent class
assignments. Organizational membership latent class assignments were not significantly correlated
with those for any of the other groups.

3.7. Segmentation Summary

The results of this relatively large national survey highlight the large degree of heterogeneity
amongst Canadians with regards to their core values, potential threats to quality of life, community
engagement, and participation in, and consumption of, cultural services and goods. For core values, six
of 15 items for the Schwartz brief scale [46] were used to identify eight distinct LCs nationally. In
subsequent CHAID segmentations, age, gender, and survey completion time were statistically
significant predictors of six distinct segments that varied according to respondents’ probability of
membership in the eight LCs.
Similarly, 11 of 12 potential quality of life indicators were used to identify five LCs within which
sample respondents’ threat perception patterns were statistically similar. HASS research awareness,
educational attainment, survey completion time, region of residence, and gender all played some role
as significant predictors of membership for the various LCs. A total of 13 segments were identified
with these predictor variables. All seven membership indicators were used to identify five LCs of
community engagement. HASS research awareness, marital status, educational attainment, and
citizenship status were predictors of six segments, within which individuals’ probability of membership
in the five LCs describing level of engagement were similar. Finally, five of six indicators of cultural
engagement and consumption were used to identify three LCs with similar patterns. The probability
that an individual belonged to any one (of 11) particular cluster could be predicted using HASS
research awareness, educational attainment, gender, and survey completion time.
It is likely that a small proportion of the sample were inattentive when answering the survey
quickly. In the core values segmentation (recall Table 2), the segment VAL SEG 01 (16.9% of the
sample) was comprised of respondents aged <40 yrs who completed the survey quickly. While it may
well be the case that many younger respondents in this segment are technology savvy and can answer
Internet-based survey questions quickly, it is also possible that some are completing the survey
in a flippant manner. This group tended to belong to LCs where all responses were middle-of-the-scale
or unimportant.
In the threat perceptions segmentation, both survey completion time and the dummy variable for
choice experiment protestors were significant predictors of LC membership probabilities for respondents
with a moderate level of HASS research awareness. Males with intermediate HASS awareness, higher
levels of education, and fast survey completion times (10.8% of the sample) were most likely to belong
to the LCs placing low priority on arts and heritage. These results imply that these segments may be
Soc. Sci. 2015, 4 333

identifying a mixture of respondents who really are middle-of-the-scale and who view arts and heritage
as unimportant and those, perhaps around 5% of the sample, who simply have been grouped with the
“true neutrals” due to their quick and inattentive responses.

4. Conclusions

The key finding of this exploratory study was that HASS research awareness acted as a powerful
predictor of threat perceptions, levels of community activity, and cultural engagement at the local level
(note that I use the term predictor in a modeling sense here; a strong relationship was identified but
that does not necessarily imply causality). It was not, however, a significant predictor of core values.
From a theoretical perspective, this is in line with a priori expectations as international values [48,61]
and VBN theories [44] suggest core values are a precursor to worldviews, threat perceptions, and
behaviors (also see [62]).
The strength of the predictive power of HASS research awareness was striking: it was the most
powerful predictor for national patterns of quality of life threat perceptions, community organization
membership, and cultural engagement and consumption. It was a more powerful predictor of those
outcomes than educational attainment alone, but in all three analyses, the combination of HASS
research awareness and educational attainment provided a degree of discriminatory power that HASS
research awareness alone could not.
Values and threat perception latent classes, and values and cultural engagement latent classes were
significantly correlated with each other, while threat perception and cultural engagement latent classes
were not. Community-based membership patterns were not correlated with membership patterns from
any of the other analyses. As the membership indicators included a mixture of environmental, sports
and fitness, service, political, and arts organization indicator variables, it is not particularly surprising
that community-based membership patterns are not correlated.
What was, however, surprising was that HASS research awareness was such a strong predictor of
community membership and engagement. Canadian respondents were largely unengaged culturally at
their community level and the majority was physically inactive. Among the more active and engaged,
HASS research awareness was significantly higher relative to the unengaged portion of citizens.
O’Brien [5] provided a telling 2011 quote from then-UK Culture Minister that alluded to the
importance of free museum access as a means to create future consumers of cultural commodities. The
results from this survey suggest a broad and strong relationship between HASS research awareness and
citizen engagement and well-being, where individuals with strong awareness of the role that HASS
research plays in society themselves play a much more important role in their communities and
society. This survey, which was not designed to specifically focus on this issue, is not able to provide
evidence regarding the causal relationships among those factors but does highlight that there is a
relationship. That is, we cannot say, based on the current results, that HASS awareness boosts
engagement or if individuals that are engaged are more likely to follow issues that expose them to the
scope of HASS research. Whichever way the relationship runs, the results strongly suggest that HASS
exposure does far more than simply create future consumers of cultural commodities.
There are very different policy prescriptions for increasing HASS research awareness and, by
extension, Canadian citizens’ cultural and physical engagement, depending on causality and how
Soc. Sci. 2015, 4 334

HASS research awareness affects citizens’ threat perceptions, values, and behavior. VBN theory
suggests that threat salience is influenced by worldview, which is itself based on core values,
sociocultural factors, and upbringing (recall Figure 1). Following that logic, HASS research awareness
may be more influenced by factors likely to be encountered later in life (e.g., news, advertising, media
campaigns, university education, etc…). In order to increase HASS research awareness, and
community and cultural engagement, interventions that focus on providing individuals (perhaps from
the early teens) with better information about the role of HASS in well-being and measures to remove
institutional barriers (norms and formal rules—[62]) that prevented people acting on intentions to
become more engaged. In this situation, the credibility of HASS researchers and research on HASS
research impact may become particularly important as a credible evidence base may be important in
influencing adults. Stronger evidence would lead to people to put an increased level of importance on
threats to arts and culture, and increase citizens’ propensity to act or support action to address threats.
Community- and culturally-engaged citizens may thus be more willing to politically and financially
support HASS research investments, potentially leading to a self-reinforcing virtuous cycle.
If, on the other hand, a high level of HASS research awareness was a result of the early-life family
and cultural environment a person was brought up within, early life experiences should be the focus of
efforts for policy. Experiential learning, rather than formal evidence, would be more important early in
life. In the environmental field “nature deficit disorder” (referring to the lack of experience children
gain as a result of not being exposed to nature in modern society) has been highlighted as being
responsible for a range of phenomena from citizens’ lack of understanding of food supply systems to
general lack of concern for environmental conservation [63–65]. If HASS awareness, as well as
cultural and community engagement were, at their core, a challenge relating to childhood experience, it
may be that society is, to echo Louv’s book Last Child in the Woods [63], facing a situation of “last
child in the museum”. Given differences in access policy among comparable countries (e.g., the UK’s
no-fee policy for access to major museums versus fee-based access in Canadian museums), it may well
be possible to develop testable hypotheses and conduct research that empirically identifies the relative
importance of these two possible pathways to citizen engagement.
This research project, while not designed specifically to assess the conceptual impacts of HASS
research on the general public, suggests this may be a potentially important issue to explore in more detail
in the future. The linkages between research awareness and social and cultural engagement will need
to be more thoroughly examined, both with regards to their causality and geographic scope, in order to
elucidate the causes, social and economic consequences, and policy implications of the relationship.

Acknowledgments

This research was funded by a grant from the Presidential Fund initiative “Capturing the Outcomes
and Impacts of Canadian Research in the Social Sciences and Humanities” by the Canadian Social
Science and Humanities Research Council and the Canada Research Chairs program while the author
was at Memorial University.
Soc. Sci. 2015, 4 335

Abbreviations

CBA: cost-benefit analysis;

HASS: humanities, arts, and social sciences;
LC: latent class;
STEM: science, technology, engineering and medical;
VBN: Value-Belief-Norm theory.

Conflicts of Interest

The author declares no conflict of interest.

References

1. Ruth Levitt, Claire Celia, Stephanie Diepeveen, Stephanie N. Chonaill, Lila Rabinovic, and Jan
Tiessen. Assessing the Impact of Arts and Humanities Research at the University of Cambridge.
Cambridge: RAND Europe, 2010.
2. Jordi Molas-Gallart, and Puay Tang. “Tracing ‘productive interactions’ to identify social impacts:
An example from the social sciences.” Research Evaluation 20 (2011): 219–26.
3. Paul Benneworth. “Tracing how arts and humanities research translates, circulates and
consolidates in society: How have scholars been reacting to diverse impact and public value
agendas?” Arts and Humanities in Higher Education 14 (2015): 45–60.
4. Helen Small. The Value of the Humanities. Oxford: Oxford University Press, 2013.
5. Dave O’Brien. “Cultural value, measurement and policy making.” Arts and Humanities in Higher
Education 14 (2015): 79–94.
6. Robert Costanza, Brendan Fisher, Saleem Ali, Caroline Beer, Lynne Bond, Roelof Boumans,
Nicolas L. Danigelis, Jennifer Dickinson, Carolyn Elliott, Joshua Farley, and et al. “Quality of
life: An approach integrating opportunities, human needs, and subjective well-being.” Ecological
Economics 61 (2007): 267–76.
7. Heide Hackmann, Susanne C. Moser, and Asuncion Lera St. Clair. “The social heart of global
environmental change.” Nature Climate Change 4 (2014): 653–55.
8. Joanne Kauffman. “Promoting integration and cooperation for sustainability views from the
symposium held at UNESCO headquarters September 19, 2013.” Sustainability Science 9 (2014):
419–30.
9. Gisli Palsson, Bronislaw Szerszynski, Sverker Sörlin, John Marks, Bernard Avril, Carole
Crumley, Heide Hackmann, Poul Holm, John Ingram, Alan Kirman, and et al. “Reconceptualizing
the ‘Anthropos’ in the Anthropocene: Integrating the social sciences and humanities in global
environmental change research.” Environmental Science & Policy 28 (2013): 3–13.
10. Murray A. Rudd. “How research-prioritization exercises affect conservation policy.” Conservation
Biology 25 (2011): 860–66.
11. David Victor. “Climate change: Embed the social sciences in climate policy.” Nature 520 (2015):
27–29.
Soc. Sci. 2015, 4 336

12. Eleonora Belfiore. “‘Impact’, ‘value’ and ‘bad economics’: Making sense of the problem of value
in the arts and humanities.” Arts and Humanities in Higher Education 14 (2015): 95–110.
13. Alan Brinkley. “Landscape of Humanities Research and Funding.” 2009. Available online:
http://archive201406.humanitiesindicators.org/essays/brinkley.pdf (accessed on 27 April 2015).
14. Claire A. Donovan. “Government Policy and the Direction of Social Science Research.” Ph.D.
Dissertation, University of Sussex, Sussex, UK, November 2001.
15. Bruce Gardner, and William Lesser. “International agricultural research as a global public good.”
American Journal of Agricultural Economics 85 (2003): 692–97.
16. Jonathan Grant, Philipp-Bastian Brutscher, Susan E. Kirk, Linda Butler, and Steven Wooding.
Capturing Research Impacts: A Review of International Practice. Cambridge: RAND Europe, 2010.
17. Shyama Kuruvilla, Nicholas Mays, Andrew Pleasant, and Gill Walt. “Describing the impact of
health research: A research impact framework.” BMC Health Services Research 6 (2006): 134.
18. Council of Canadian Academies. The State of Science and Technology in Canada. Ottawa:
Council of Canadian Academies, 2006.
19. Council for the Humanities, Arts and Social Sciences (CHASS). Measures of Quality and Impact
of Publicly Funded Research in the Humanities, Arts and Social Sciences. Canberra: CHASS, 2005.
20. Einar Bowitz, and Karin Ibenholt. “Economic impacts of cultural heritage—Research and
perspectives.” Journal of Cultural Heritage 10 (2009): 1–8.
21. Michelle Reeves. Measuring the Economic and Social Impact of the Arts: A Review. London: Arts
Council of England, 2002.
22. Ståle Navrud, and Richard C. Ready. Valuing Cultural Heritage: Applying Environmental Valuation
Techniques to Historic Buildings, Monuments and Artefacts. Cheltenham: Edward Elgar, 2002.
23. Douglas S. Noonan. “Contingent valuation and cultural resources: A meta-analytic review of the
literature.” Journal of Cultural Economics 27 (2003): 159–76.
24. Murray A. Rudd. “An exploratory analysis of societal preferences for research-driven quality of
life improvements in Canada.” Social Indicators Research 101 (2011): 127–53.
25. Tran H. Tuan, and Ståle Navrud. “Capturing the benefits of preserving cultural heritage.” Journal
of Cultural Heritage 9 (2008): 326–37.
26. Wiktor L. Adamowicz. “What’s it worth? An examination of historical trends and future
directions in environmental valuation.” The Australian Journal of Agricultural and Resource
Economics 48 (2004): 419–43.
27. Mark Sagoff. “Should preferences count?” Land Economics 70 (1994): 127–44.
28. John Dewey. Logic: The Theory of Inquiry. New York: Holt, 1938.
29. Mark Sagoff. “The quantification and valuation of ecosystem services.” Ecological Economics
70 (2011): 497–502.
30. Ricky N. Lawton, and Murray A. Rudd. “Strange bedfellows: Ecosystem services, conservation
science, and central government in the United Kingdom.” Resources 2 (2013): 114–27.
31. John Spoehr, Kate Barnett, Simon Molloy, Sanjugta V. Dev, and Ann-Louise Hordacre. Connecting
Ideas: Collaborative Innovation for a Complex World. Adelaide: Australian Institute for Social
Research, University of Adelaide, 2010.
32. Mark Banks, and Justin O’Connor. “After the creative industries.” International Journal of
Cultural Policy 15 (2009): 365–73.
Soc. Sci. 2015, 4 337

33. Joseph L. Koerner, and Lisbet Rausing. “Value.” In Critical Terms for Art History, 2nd ed. Edited
by Robert S. Nelson and Richard Shiff. Chicago: University of Chicago Press, 1996, pp. 419–34.
34. Jonathan Bate. The Public Value of the Humanities. London: Bloomsbury Academic, 2011.
35. Arts Council England (ACE). Cultural Capital: A Manifesto for the Future. London: ACE, 2010.
36. Alan S. Brown, Jennifer L. Novak, and Amy Kitchener. “Cultural Engagement in California’s
Inland Regions.” 2008. Available online: http://www.giarts.org/sites/default/files/Cultural-
Engagement-in-Californias-Inland-Regions.pdf (accessed on 27 April 2015).
37. Kate Clark, and Gareth Maeer. “The cultural value of heritage: Evidence from the Heritage
Lottery Fund.” Cultural Trends 17 (2008): 23–56.
38. Seoyong Kim, and Hyesun Kim. “Does cultural capital matter? Cultural divide and quality of
life.” Social Indicators Research 93 (2009): 295–313.
39. Jane Macnaughton, Mike White, and Rosie Stacy. “Researching the benefits of arts in health.”
Health Education 105 (2005): 332–39.
40. Jude Ortiz, and Gayle Broad. Culture, Creativity, and the Arts: Achieving Community Resilience
and Sustainability through the Arts in Sault Ste. Marie. Saskatoon: Centre for the Study of
Co-Operatives, University of Saskatchewan, 2009.
41. Susan Galloway. “Cultural participation and individual quality of life: A review of research
findings.” Applied Research in Quality of Life 1 (2006): 323–42.
42. David Throsby. “Determining the value of cultural goods: How much (or how little) does
contingent valuation tell us?” Journal of Cultural Economics 27 (2003): 275–85.
43. Andy Choi, Franco Papandrea, and Jeff Bennett. “Assessing cultural values: Developing an
attitudinal scale.” Journal of Cultural Economics 31 (2007): 311–35.
44. Paul C. Stern. “Toward a coherent theory of environmentally significant behavior.” Journal of
Social Issues 56 (2000): 407–24.
45. Riley E. Dunlap, Kent D. van Liere, Angela G. Mertig, and Robert E. Jones. “Measuring
endorsement of the New Ecological Paradigm: A revised NEP scale.” Journal of Social Issues 56
(2000): 425–42.
46. Paul C. Stern, Thomas Dietz, and Gregory A. Guagnano. “A brief inventory of values.”
Educational and Psychological Measurement 58 (1998): 984–1001.
47. Shalom H. Schwartz. “Are there universal aspects in the structure and contents of human values?”
Journal of Social Issues 50 (1994): 19–45.
48. Shalom H. Schwartz. “An overview of the Schwartz theory of basic values.” Online Readings in
Psychology and Culture, 2012. doi:http://dx.doi.org/10.9707/2307-0919.1116.
49. Sao-Wen Cheng. “Cultural goods creation, cultural capital formation, provision of cultural services
and cultural atmosphere accumulation.” Journal of Cultural Economics 30 (2006): 263–86.
50. Claire A. Donovan. “The Australian Research Quality Framework: A live experiment in capturing
the social, economic, environmental, and cultural returns of publicly funded research.” New
Directions for Evaluation 2008 (2008): 47–60.
51. Jay Magidson, and Jeroen K. Vermunt. “Latent class models.” In The Sage Handbook of Quantitative
Methodology for the Social Sciences. Edited by David Kaplan. Thousand Oaks: Sage Publications,
2004, pp. 175–98.
Soc. Sci. 2015, 4 338

52. Jeroen K. Vermunt, and Jay Magidson. “Latent class cluster analysis.” In Applied Latent Class
Analysis. Edited by Jacques A. Hagenaars and Allan L. McCutcheon. Cambridge: Cambridge
University Press, 2002, pp. 89–106.
53. Jacques A. Hagenaars, and Allan L. McCutcheon. Applied Latent Class Analysis. Cambridge:
Cambridge University Press, 2002.
54. Juan Pulido-Fernández, and Marcelino Sánchez-Rivero. “Attitudes of the cultural tourist: A latent
segmentation approach.” Journal of Cultural Economics 34 (2010): 111–29.
55. Jeroen K. Vermunt, and Jay Magidson. Technical Guide for Latent GOLD 4.0: Basic and
Advanced. Belmont: Statistical Innovations Inc., 2005.
56. Nissan Levin, and Jacob Zahavi. “Predictive modeling using segmentation.” Journal of Interactive
Marketing 15 (2001): 2–22.
57. Gordon V. Kass. “An exploratory technique for investigating large quantities of categorical data.”
Applied Statistics 29 (1980): 119–27.
58. Jay Magidson, and Jeroen K. Vermunt. “An extension of the CHAID tree-based segmentation
algorithm to multiple dependent variables.” In Classification: The Ubiquitous Challenge. Edited by
Claus Weihs and Wolfgang Gaul. Heidelberg: Springer Heidelberg, 2005, pp. 176–83.
59. Cathy H. C. Hsu, and Soo K. Kang. “CHAID-based segmentation: International visitors’ trip
characteristics and perceptions.” Journal of Travel Research 46 (2007): 207–16.
60. Jay Magidson. SI-CHAID 4.0 User’s Guide. Belmont: Statistical Innovations Inc., 2005.
61. Shalom H. Schwartz. “Rethinking the concept and measurement of societal culture in light of
empirical findings.” Journal of Cross-Cultural Psychology 45 (2014): 5–13.
62. Elinor Ostrom. Understanding Institutional Diversity. Princeton: Princeton University Press, 2005.
63. Richard Louv. Last Child in the Woods: Saving Our Children from Nature-Deficit Disorder.
Chapel Hill: Algonquin, 2005.
64. Roger Walsh. “Lifestyle and mental health.” American Psychologist 66 (2011): 579–92.
65. Nancy M. Wells, and Kristi S. Lekies, “Nature and the life course: Pathways from childhood nature
experiences to adult environmentalism.” Children, Youth and Environments 16 (2006): 1–24.

© 2015 by the author; licensee MDPI, Basel, Switzerland. This article is an open access article
distributed under the terms and conditions of the Creative Commons Attribution license
(http://creativecommons.org/licenses/by/4.0/).