Abstract
Situational-enchantment is a hypothesized arousal state encompassing a potent sense of connection or oneness with a “transcendent power or ultimate reality.” Qualitative research previously suggested that this individual difference involves dissonance around ideations with competing “Emotional, Sensorial, Timeless, Rational, and Transformative” themes. We tested this presumed phenomenology via an online convenience sample of 79 men and 101 women who reported memorable ghostly experiences during a paranormal tour within the last 12 months. Respondents provided a global enchantment rating of their anomalous experiences, as well as selected specific descriptors from a set of 30 items on an adjective checklist (ACL). Analyses revealed that 21 items on the Enchantment-ACL formed a Rasch hierarchy of generally “pleasant” themes that was free of response biases related to age, sex, and latency (time since the “enchanting” experience). This structured sequence contained all five experiential themes, and the resulting Enchantment-ACL measure of this phenomenon showed good internal reliability (Rasch reliability = .82) and a positive correlation with global enchantment ratings (r = .51, p < .001). The other nine items formed a separate factor containing overtly “unpleasant” ideations. We discuss the results within a cognitive dissonance framework for situational-enchantment, although future research must explore potential nuances related to the construct’s dimensionality and the specific role of pleasant versus unpleasant ideations.
The world will never starve for want of wonders; but only for the want of wonder.
He who can no longer wonder, no longer feel amazement, is as good as dead, a snuffed-out candle.
Introduction
Part I (Drinkwater et al., 2022) of this two-phase study reviewed and synthesized the limited literature on the nature, experience, and relevance of “enchantment” as an individual difference or psychological construct in service-hospitality consumerism. We subsequently proposed the term situational-enchantment to denote an apparently complex arousal state that occurs when a person becomes immersed within a melee of “pleasant” ideations and emotions (e.g., excitement, surprise, awe, and wonder), simultaneously mixed with more “unpleasant” ideations and emotions (e.g., uneasiness, disorientation, tension, or unpredictability). This juxtaposition ostensibly results from a person-environment enaction (cf. Jawer et al., 2020; Jelić et al., 2016) that disrupts an individual’s normal waking experience with a sudden, unexpected, or profound awareness which seeds a transformative feeling of connection to a “transcendent agency or ultimate reality.”
Consequently, and contrary to common wisdom that enchantment purely reflects positive psychology (i.e., efficacious emotions and experiences; e.g., Bennett, 2001; Filep & Pearce, 2014; Hosany et al., 2015; Kushner, 2018), the conceptualization above emphasizes the role of dissonance or “dis-ease” (i.e., the natural state of “ease” being imbalanced or disrupted) in the formation and maintenance of this hypothesized mental state. Indeed, a qualitative thematic analysis by Drinkwater et al. (2022) characterized the experience of enchantment in terms of five distinctive “target” features: (a) emotional, (b) sensorial, (c) timeless, (d) rational, and (e) transformative. Dissonance can therefore occur on multiple levels, that is, globally there are competing themes (e.g., emotional vs. rational) with specific aspects that are sometimes diametrically opposed (e.g., pleasant vs. unpleasant). Such incompatible ideations are apparently resolved, as needed, by invoking purposeful agents (cf. Valdesolo & Graham, 2014). Overall, this proposed structure and process arguably imply a phenomenon that is fundamentally more akin to a state of shock or surrealism (or even hyperrealism, Santos, 2018) than what might be described as a pleasant surprise or “emotional rapture” (Bermudez, 2008, p. 4).
Psychometric verification of this proposed phenomenology poses challenges; however, as Lange (2017; Lange et al., 2019) has repeatedly shown that serious measurement issues can occur with data that are (a) quantified or modeled using standard approaches in Classical Test Theory (CTT) and (b) derived from surveys on altered states of consciousness or exceptional human experiences. Accordingly, Lange, Irwin, and Houran. (2000) introduced a series of statistical analyses grounded in item response theory (IRT) and particularly Rasch (1960/1980) scaling. This “top-down purification” approach facilitates the identification and removal of age- or sex-related response biases at the item and test levels to yield unbiased instruments with interval measurement properties and rigorously tested factor structures (see, for example, Bond & Fox, 2015).
The Present Study
Part II aims to validate the formerly proposed phenomenology for enchantment (Drinkwater et al., 2022) via an empirical analysis of self-reported “enchanting experiences” during paranormal tours. Like Houran et al. (2020), this study neither considers/endorses nor relies on the ontological reality of so-called “paranormal or supernatural” phenomena. Rather, the social science literature contains many studies based on data from participants in ghost tours, legend-tripping, or related excursions. In this way, researchers have been able to explore various orthodox issues such as belief formation and maintenance or the quality of immersive visitor experiences within ecologically valid settings (e.g., Eaton, 2019; Langston & Hubbard, 2019; Pharino et al., 2018).
This approach is equally an important data pipeline for the present study, as paranormal tourism and related activities arguably facilitate experiences of “enchantment” (Herrmann, 2014; Holloway, 2010; Houran et al., 2020; Schneider, 1993). Thus, data on visitor experiences collected in these contexts are expected to emphasize the five target features noted above, as opposed to referencing terms that contradict these core themes. We therefore tested three main hypotheses using an adjective checklist (ACL) approach to scrutinize empirically the phenomenology of situational-enchantment:
An ACL to Gauge the Phenomenology of Enchantment
The ACL method has a respected legacy in social science. Gough and Heilbrun (1965, 1983/2007; cf. Gough, 1960) were perhaps the first researchers to popularize this technique as a systematic way to identify common psychological traits or tendencies in study samples. The original ACL has been one of the most widely used personality inventories (Buros, 1978), with inclusion in more than 1,000 studies and research reports and continued influence today (e.g., Hill et al., 2002; Matthews et al., 1990; Piedmont et al., 1991). Here, we used a two-step procedure to adapt the basic ACL method to quantify the characteristics of experiences that involve a putative state of enchantment.
First, Table 4 in Drinkwater et al. (2022) was our starting point to develop a set of “target” words that we hypothesized would be endorsed by individuals who reported an “enchanting” experience during a paranormal tour or related excursion. We reviewed, revised, and shortened the original collection of items (n = 64) in that Table to enhance its readability across a wide range of respondents. We developed a list of five target terms for each of the five phenomenological features (n = 25) and added brief descriptions to clarify and reinforce their intended meanings.
Second, we created a parallel set of antonyms to serve as “control” words (n = 5). These were terms that we would not expect people to endorse as descriptors for their “enchanting” experiences. The row entries of Table 1 present the original pool of items in the Enchantment-ACL, whereas Table 2 summarizes its readability statistics calculated at https://readable.io/text/. Text analysis indicates that the items meet a college-level of reading comprehension (for an introduction to readability metrics, see Kouamé, 2010).
Rasch Scaling Results for the Original Item Pool in the Enchantment–Adjective Checklist Measure.
Note. E = Emotional; S = Sensorial; I = Timeless; R = Rational; T = Transformative; C = Control words.
Outfit values exceeding 1.40 cutoff are shown in bold.
Readability Metrics for the Enchantment–Adjective Checklist Measure.
Method
Rasch Scaling
The following provides a summary of the basic features of the (binary) Rasch (1960/1980) model used in the present study (for more detailed overviews, see Bond & Fox, 2015; Lange, 2017). This model focuses around the Pij,―the probability that person i will endorse item j—given that the person has trait level Ti (also called the person location) and the item assesses the trait at level Dj. The latter is also referred to as the item’s difficulty or the item’s location. The quantities are related as follows:
Items’ fit to Equation 1 are assessed via their outfit values, and values in the range of 0.5 to 1.4 generally deemed acceptable (Linacre, 2020). To determine a variable’s semantics, it is very useful to plot items according to their locations in a map-like format, and because Ti and Dj are expressed as units along the same dimension, persons’ trait levels can be shown as well (Bond & Fox, 2015). We provide an example of such a “Wright map” in the section “Results” (cf. Lange, 2017).
Equation 1 can be extended to the multidimensional case, whereby several distinct factors are fitted simultaneously (for more information, see Wu et al., 2007). We note that the multivariate approach allows for competitive tests between nested models with different numbers of factors. It also estimates attenuation-corrected correlations between the factors; however, no statistical tests are available for such correlations. Additional facets can be included in the model to assess their impact either as main effects (e.g., men vs. women) or as interaction effects. Of particular interest are interactions involving the Dj, as these indicate that items’ difficulties vary across subgroups, thus creating different item hierarchies. Such interactions are also referred to as “biases.” For instance, we might find that some items are easier (or harder) to endorse for men than for women with equal trait levels; and these items are said to be biased. The bias notion can be extended to include interactions between groups of items and respondent groups. One might ask, for instance, whether the combined endorsement rates of the control items (combined) are biased across men and women.
In Rasch scaling, raw sums are sufficient statistics to estimate respondents’ trait levels, and estimates of T can be provided in a lookup table. Rasch scaling defines reliability (R) analogous to the approach in CTT (see, for example, Linacre, 2020), but reliability computations use test-takers’ estimated trait levels T in Equation 1 rather than raw scores. As a result, R can also be estimated in the event of missing data. In general, R tends to be lower than KR-20 or coefficient alpha. We performed all univariate Rasch analyses using the Winsteps software (Linacre, 2020) whose “joint maximum likelihood estimation” (JMLE) algorithm makes no assumptions about the distribution of trait levels or item difficulties. Accordingly, the shapes of the item and person location distributions do not materially affect the estimates of D and T. Issues of dimensionality are addressed here via factor analysis of items’ Rasch residuals (Linacre, 2020). All multivariate analyses use the R-based TAM software (Robitzsch et al., 2020).
Respondents
A convenience sample of 79 men and 101 women (age: M = 38.6 years, SD = 14.2 years, range = 18–70 years) was recruited from the SurveyMonkey® Audience paid research panel. Such online samples are commonly used in social science for their ease of access and low cost (e.g., Jamnik & Lane, 2017; Pollet & Saxton, 2019). The respondents were pre-screened for having had reported a “particularly memorable ‘ghostly’ experience during a paranormal tour or excursion within the past 12 months.” Participation was voluntary and could be withdrawn at any point by the respondents.
Procedure
Respondents completed demographic questions about age, sex, and latency (i.e., “number of months” elapsed since the memorable paranormal experience). The aim of the study was then explained via the following instruction set: “People often use the word ‘enchantment’ to describe experiences or events that made them feel ‘charmed or under a spell’. We are researching the characteristics of such experiences.” Next, respondents provided a global assessment of their experience, that is, “Overall, how ‘enchanting’ would you rate your memorable paranormal experience?” (0 = not at all enchanting, 1 = barely enchanting, 2 = somewhat enchanting, and 3 = very enchanting). Finally, the 30 Enchantment-ACL items (binary with no reverse scoring) were presented in random order for each individual survey with the following instruction set: “Now carefully consider which (if any) of the words below match the thoughts and feelings you remember having during your memorable ‘paranormal experience’. Check all the words that match what you felt.” Our procedure further strived to mitigate potential response biases by ensuring that (a) respondents’ data were anonymous and (b) no appraisals or value judgments were either indicated or made about the veracity of respondents’ experiences or their subsequent effects.
Results
Preliminaries
The sampling outcomes revealed some interesting findings. First, the binary screening question of “having had a ghostly experience” yielded an incidence rate of 49.73%, which exceeds previously reported metrics on haunt-type experiences (e.g., Haraldsson, 2011; McClenon, 2012; Rice, 2003). Second, most of the respondents rated their experiences as “somewhat” (46%) or “very” (27%) enchanting, followed by 18% who described them as “barely” enchanting. Only 9% of the respondents indicated that they were “not at all enchanted” by their anomalous experiences. These patterns strongly support previous suggestions that haunt-type experiences are associated with enchantment (Holloway, 2010; Houran et al., 2020; Schneider, 1993). As such, the results underscore the viability of using paranormal tourists to study the nature of relevance of enchantment in naturalistic settings.
Hypothesis 1a: Basics
Our predictions were largely confirmed. Specifically, the top two sections of Table 1 show the basic Rasch scaling analyses as applied to the 25 non-control items in the top two sections (“Pleasant” and “Unpleasant”). It can be seen that the four unpleasant items (i.e., 3, 5, 9, and 10) showed outfit in excess of 1.4, indicating noticeable presence of noise. However, acceptable fit is obtained for the 21 items (“Pleasant”) in the top section. In addition, none of the pleasant items showed statistically significant biases associated with respondents’ age, sex, or latency (time elapsed since the “paranormal” experience). In the preceding, we set the Type I error level at 0.01 given the large number (i.e., 63) of pairwise comparisons. With some exceptions, these findings thus support the existence of a single (i.e., unidimensional) Rasch hierarchy of ideations that captures a probabilistic hierarchical set of enchantment-related ideations.
The semantics of the 21 pleasant items are captured by the Wright map (see Linacre, 2020) in Figure 1, which shows the persons’ trait-level locations on the left and the item location on the right of the Rasch dimension represented by the central vertical line. Note that the higher locations denote greater trait levels of apparent enchantment, as higher item locations reflect greater item difficulty (lower endorsement). The probabilistic item progression on the right side of Figure 1 indicates that the state of enchantment commonly starts with awareness and surprise, followed by feelings of excitement, amazement, and finally transcendent insights and higher-order ideations such as joy, beauty, and goodness. This basic sequence is consistent with the key junctures of Drinkwater et al.’s (2022) “epiphanic-type” process model for enchantment. In summary, the fit of the Rasch model implies that the sequence is a legitimate and interconnected hierarchy such that items occurring higher in the hierarchy always (i.e., regardless of respondents’ trait levels) have a lower probability of being endorsed than items those occurring at lower levels. For instance, individuals reporting awe and delight and being inspired are also likely to report being aware or surprised or being lost-in-the-moment.

Wright-Style Person and Item Map in Logits for the Enchantment–Adjective Checklist Measure.
Respondents’ estimated trait levels (locations) are shown as a histogram on the left side of Figure 1. Notice that the persons’ locations (M = 0.42, SD = 1.68) largely overlap with the items’ locations (M = 0.00, SD = 0.71 logits). Accordingly, best measurement occurs near the middle of the dimension. The overall Rasch reliability is quite high, R = .82. A series of independent t-tests on the 21-item Rasch scaled Enchantment-ACL revealed no statistically significant differences between women and men (MWomen = 0.40 vs. MMen = 0.44 logits), younger and older respondents (MYounger = 0.47 vs. MOlder = 0.35 logits), and time elapsed since the “paranormal” experience (MShorter = 0.28 vs. MLonger = 0.59 logits; all p > .10). As mentioned above, no item shifts were associated with these variables. Hence, the absence of statistical significance cannot be attributed to response biases.
Scaled scores
To obtain a convenient reporting scale, the persons’ parameters in the present sample were transformed to lie inside the range 1 to 99 with M = 50 and SD = 15. Table 3 provides the corresponding “raw-sum to scaled-score” transformation, and we offer investigators this Enchantment-ACL instrument to quantify situational-enchantment in future research (see Appendix A). Table 3 also gives the local standard errors of measurement (SE), which achieve a minimum (i.e., 4) for raw sum scores of 7 through 14. Note that we also designed a generic version of the Enchantment-ACL measure for studies outside of paranormal tourism. Interested readers may contact the authors for details on obtaining a copy for non-commercial purposes.
Raw-Score to Scale Score Transformation for the Enchantment–Adjective Checklist Measure (M = 50, SD = 15).
Answers are coded as 0 (not endorsed) and 1 (endorsed).
Hypothesis 1a: Dimensionality
Our analyses identified three subsets of the 30 administered items using qualitative criteria and insights, that is, the control words (Items 26–30), the unpleasant ideations (Items 3, 5, 9, 10), and the remaining 21 items that refer to generally pleasant ideations and which define the unidimensional Enchantment-ACL measure described above. To check the validity of these partitions, we compared the fit of a three-dimensional Rasch model comprising these factors with the one-dimensional model comprising all 30 items using the TAM software (Robitzsch et al., 2020). As is shown in Table 4, competitive model test indicated that the three-factor formulation indeed provided greater statistical fit (p < .001), thereby validating our scaling approach.
Competitive Model Tests for Enchantment–Adjective Checklist With Varying Numbers of Rasch Factors.
In addition, we factor-analyzed the response residuals of the 21 Enchantment-ACL items using the Winsteps’ (Linacre, 2020) software, and Table 5 shows the items’ loadings along the first residual factor. The pattern loadings suggest that, in addition to varying along the probabilistic Rasch scale, the 21 Enchantment-ACL items also vary along bipolar distinction of “Attention” (positive loadings) versus “Connection” (negative loadings). Therefore, a four-dimensional Rasch model was defined by subdividing the 21 pleasant items into two groups according to Table 5, and we found that the data provided a superior fit to the four-dimensional model (p < .001, see Table 4).
First Residual Factor in Pleasant Items on the Enchantment–Adjective Checklist Measure.
Table 6 gives the disattenuated correlations among these four apparent factors. It can be seen that the highest correlation (rdirect = .62) occurs between Attention and Connection factors. In this context, Evans et al. (2019) argued that correlations of this magnitude might indicate multidimensionality. If validated, these putative factors might be in line with previous arguments (e.g., David, 2010; Lange et al., 2019; Wahbeh et al., 2020) for two, potentially separate processes: (a) the occurrence of anomalous experiences or altered states, versus (b) the interpretation of these experiences or states. This clarification of the factors would therefore perhaps agree with Bermudez (2009) who differentiated the nature of “extraordinary architectural experiences” (e.g., emotions or sensual perceptions) from their outcomes on percipients (e.g., a sense of peace or insight). Given the exploratory nature of the preceding analyses, we recommend that researchers provisionally treat and use the 21 items in the Enchantment-ACL as a one-dimensional scale until the present findings have been sufficiently replicated. Alternatively, one might consider excluding the Connection items.
Attenuation-Corrected Correlations for the Four-Factor Rasch Model.
Hypothesis 2
Our expectation regarding the control items was confirmed. The bottom section of Table 1 includes the five “control” words, that is, our hypothesized antonyms to the defining features of enchanting experiences. These were calibrated while holding constant the difficulties of the 21 “normal” (i.e., pleasant) items listed in the top section of Table 1. As is indicated by their greater average difficulty (MControl = 1.83) relative to the pleasant items (MNormal = 0.00), respondents selected the control words with considerably smaller frequency, t(28) = −4.63, p < .001. That is, individuals in an ostensible state of enchantment tend not to feel bored, tired, disappointed, ordinary, or alone. Simply put, these descriptors rarely applied and for this single reason are irrelevant. In addition, we note that two control words (i.e., Items 27 and 29, see Table 1) showed poor fit to the Rasch model.
Hypothesis 3
The Pearson correlation between respondents’ four-category global enchantment ratings and scaled scores on our 21-item Enchantment-ACL measure was .51 (p < .001). This supports the hypothesis that the items on the checklist captured the respondents’ overall experiences of enchantment.
Discussion
The present results lend strong credence to situational-enchantment as a quantifiable construct (or individual difference) with a uniform phenomenology that aligns to the components of Drinkwater et al.’s (2022) conceptual map. Rather than an experience defined by several independent factors as suggested by prior research on the experience economy (e.g., De Geus et al., 2016; Mehmetoglu & Engen, 2011) and extraordinary architectural experiences (e.g., Bermudez, 2009; Ro & Bermudez, 2015), our findings suggest that enchantment comprises a reliable progression of competing emotional, sensorial, timeless, rational, and transformative perceptions. Furthermore, these sequential themes proved stable irrespective of respondent age, sex, or latency. Our results therefore argue that enchantment reflects a state of “surrealism” that expands the boundaries of consumers’ beliefs or expectations. Rather than giving consumers what they literally expect or want, the idea is to facilitate experiences that situate consumers “betwixt and between” reality and fantasy. This more complex and nuanced view undermines generic characterizations of enchantment as “memorable or emotionally-engaging experiences” (e.g., Kawasaki, 2011; Santos, 2018) or “a feeling of being connected in an affirmative way to existence” (Bennett, 2001, p. 156).
Accordingly, this study substantiates both prior theoretical and empirical interest in this largely unexplored aspect of psychology (e.g., Boje & Baskin, 2011; Hartmann & Brunk, 2019; Ostergaard et al., 2013). We refer readers to Appendix B for some preliminary recommendations on engineering experiences of enchantment in general consumer settings based on our cumulative findings. Obviously not every service-hospitality offering can (or should) be enchanting for consumers, but our Enchantment-ACL inventory offers a viable measurement approach for continued research on people’s motivations and experiences across a wide array of service-hospitality contexts (for use with permission and attribution, see Appendix A). In this way, businesses have the beneficial capability to measure, compare, and monitor the level of putative enchantment associated with current offerings or new product developments.
Such testing can be conducted separately or in conjunction with other consumer research, including Net Promoter Scores, mystery shopping, social media feedback, or consumer satisfaction surveys (for reviews, see, for example, Ares & Varela, 2018; Peighambari et al., 2016). This line of research certainly includes the potential for gaining a deeper understanding of employee engagement and the meaning and determinants of “enchanted workplaces” (e.g., Boje & Baskin, 2011; Endrissat et al., 2015; Michaelson et al., 2014). We should further stress that the enchantment construct likely has similar importance and application for other disciplines, such as anomalistic, clinical, cognitive, and transpersonal psychologies (for reviews, see Cardeña et al., 2014; Lankton & Lankton, 1986; Rabeyron & Loose, 2015; Sagher et al., 2019).
Although our Rasch analyses supported a process involving aspects of dis-ease (e.g., shock or confusion), the finding that overtly unpleasant items did not reliably fit the Enchantment-ACL Rasch hierarchy might seem to contradict Drinkwater et al.’s (2022) model of situational-enchantment as an expression of “cognitive dissonance.” This term refers to situations involving conflicting attitudes, beliefs, or behaviors that produce mental discomfort and thereby lead to an alteration in thoughts or behaviors to restore psychological balance (Festinger, 1957, 1962; Metin & Camgoz, 2011). It has been further argued that dissonance is remedied by invoking purposeful agents (Valdesolo & Graham, 2014), which manifests in the context of enchantment as the idea of connection or oneness with a “transcendent agency or ultimate reality” (Bermudez, 2009; Drinkwater et al., 2022).
However, the mis-fitting unpleasant items arguably support Drinkwater et al.’s (2022) position. The key to the apparent discrepancy could be the inherent latency of the paranormal tour experiences in our survey design combined with cognitive dissonance being inherently revisionist in nature. From the early work with Festinger (1957)—who created a conflict between behavior and attitudes—to later research on the cognitive dissonance between roles and statuses (e.g., Matz & Wood, 2005; Rydell et al., 2008; Van Dijk & Brown, 2006) and experiences of paranormal tourism that can create distinctly different emotions at the same time (Holloway, 2010; Pharino et al., 2018), the unavoidable consequence of cognitive dissonance is for individuals to select one state over the other as more important or better-fitting to their existing schemas.
Likewise, Festinger’s (1957, 1962) work was controversial in the sense that behaviors would dictate attitudes and not the other way around (see, for example, Elliot & Devine, 1994; Harmon-Jones & Harmon-Jones, 2007; Hinojosa et al., 2017; Rabin, 1994). Following this rationale, we might infer that the Enchantment-ACL measure showed a unified dimension of generally “pleasant” items because we collected the ideations or impressions post-experience. In this circumstance, any cognitive dissonance would already have been resolved for the respondents. Consequently, and consistent with Lange and Houran’s (1998, 1999, 2000) anxiolytic model of paranormal belief and experience, we propose that conditions of low-enchantment can involve negative or unpleasant aspects, but that such ideations quickly diminish (or are discounted or reinterpreted) to produce a condition of high-enchantment.
Our conclusions are tempered by several limitations. First, the results are based on our choice of specific words, descriptors, and categorizations as guided by Drinkwater et al.’s (2022) qualitative research. We do not know how the results might change based on different content. In addition, our results might differ with content having greater readability. Future studies might strive, therefore, to revise the Enchantment-ACL over time to meet a “Grade 8” rating to ensure that it can be easily understood by 80% of adult Americans. Second, we explored the phenomenology of enchantment only in retrospective accounts related to paranormal tourism. Thus, it is uncertain whether our Rasch model generalizes to experiences of enchantment across the wider service-hospitality industry and even beyond to other contexts.
Moreover, this study neither specifically nor vigorously inspected Drinkwater et al.’s (2022) process model for enchantment that is rooted in dis-ease or cognitive dissonance, that is, Detection → Absorption → Consternation → Impression → Affirmation. Although the hierarchical sequence of perceptions or ideations in our Rasch model generally paralleled key junctures of Drinkwater et al.’s hypothesized process, a comprehensive validation is a more complicated issue that will likely require creative methodologies to tackle. For instance, one tactic might be to analyze first-hand narratives with interpretive phenomenological analysis (e.g., Drinkwater et al., 2013; Simmonds-Moore, 2016) or conversation analysis (Murray & Wooffitt, 2010; Wooffitt, 1992). These approaches combine hermeneutics and idiography to understand how people construct meaning from their experiences and likewise how experiences affect individuals. Another technique is structural equation modeling (or path analysis) to build and empirically test competing statistical models that aim to describe the direction and strength of relationships among a set of variables in a data set (e.g., Lange & Houran, 1998, 1999; Lawrence et al., 1995).
Further research is plainly needed because the effect of “attitudinal revisionism” in cognitive dissonance can only be supported or rejected by laboratory or field research that adopts a longitudinal design or measures people’s attitude or belief formation in real-time. Finally, we must consider the prospect that what Drinkwater et al. (2022) presented implicitly as a linear process might involve nonlinear relationships among its components, as seen in some instances of “meaning-making” relative to anomalous experiences (e.g., Lange et al., 2000–2001; Lange & Houran, 2000). Identifying these components and understanding their nuances is another challenge. For example, this study identified some ambiguities related to the dimensionality of situational-enchantment (i.e., Attention vs. Connection factors), which will require clarification to develop the most accurate causal models of this construct or individual difference.
On a broader and closing note, our results underscore previous recommendations (Houran et al., 2017, 2019; Lange & Houran, 2015) for research instruments in the industry to have robust scaling properties and corrections for response biases to maximize reliability and validity, particularly in applied contexts. Unfortunately, most empirical research on visitor motivations or outcomes and the factor structures of hospitality-related experiences have traditionally relied on CTT versus Modern Test Theory approaches (e.g., Boley et al., 2011; Choi & Sirakaya, 2005; De Geus et al., 2016; Honea & Dahl, 2005; Hosany et al., 2015; Hosany & Gilbert, 2010; Huang & Pearce, 2019; Kim & Ritchie, 2014; Larsen et al., 2009; Luo et al., 2018; Mehmetoglu & Engen, 2011). It is unclear, therefore, to what extent such literature on industry theory and practice might be distorted or fundamentally flawed. We hope this article accordingly serves as a case study in high-tech psychometric methods for powerful operationalization and quantification of constructs. To be sure, quality science follows from quality measurement (Kornbrot et al., 2018; Lange & Houran, 2015).
Footnotes
Appendix A
Appendix B
Acknowledgements
We thank Neil Dagnall, Ken Drinkwater, Cindy Little, and Brandon Massullo for their assistance with this study. The second author’s participation was made possible by a grant (bursary # 20/18) from the BIAL Foundation (Portugal). Our appreciation also goes to J. Bruce Tracey for encouraging a psychometric analysis of the enchantment concept.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, or publication of this article.
