CONSUMER NOSTALGIA LITETURE REVIEW AND AN ALTERNATIVE MEASUREMENT PERSPECTIVE

During the last two decades consumer nostalgia literature has experienced the growing amount of research, nonetheless, the nomological network in the area is still poorly established and fundamental questions of generalizability and measurement of nostalgia e!ects remain unanswered. "is paper represents an a#empt to comprehensively assess extant research in consumer nostalgia $eld, distinguish developments in the literature by summarizing the main $ndings of previous research and establishing theoretical trends. "e analysis reveals that a number of demographic, social and psychological nostalgia antecedents, moderators and outcomes remain at the propositions level or lack the accumulated empirical quantitative support and validation %om other studies. "erefore, speci$c recommendations regarding the development of nostalgia nomological network are provided to aid the continued theoretical and methodological improvements in the area. Since 1991 research in nostalgia has assumed that the correct measurement approach is a re&ective one. "is paper o!ers an alternative perspective for viewing and operationalizing nostalgia construct as a formative construct. Guidelines are summarized that aim to assist researchers with decision rules on whether to employ formative or re&ective nostalgia measurement for future research. One of the main contributions of this study is to show the need for researchers to explicitly justify their choice of re&ective or formative measurement models by supporting it with theoretical arguments and empirical evidence.


Introduction
Emerging economies' share in global output has increased from less than 20% in the early 1990s to more than 30% at present (ECB, 2013, measured at market exchange rates) and these regions are a racting considerable a ention both from businesses and theorists. Taking into account promising growth prospects of emerging countries, businesses seek to expand to those new and li le discovered territories. However, emerging countries may have speci c features that have to be taken into consideration and rms among other important factors need to evaluate consumer di erences that may be deciding for the success of entrance strategies. Consumer nostalgia is one of the factors that are relevant for emerging markets, because in some se ings this individual feature may interplay with another well researched phenomenon, such as consumer ethnocentrism. For example, due to a large number of newly established countries, e ects of nostalgia may be carried-over across state borders and become relevant for international marketing practitioners and researchers. us, the ndings of this paper have considerable implications for companies seeking to enter or strengthen the position in the emerging international markets. Furthermore, researchers need be er understanding of the individual di erences in emerging markets. Despite growing amount of research on emerging international contexts, this topic still remains underresearched and there are signi cant generalizability gaps in the marketing literature.
is study seeks to explore one of the individual factors that is relevant for emerging markets, namely, the consumer nostalgia, de ned as "a preference (general liking, positive a itude, or favorable a ect) toward objects (people, places, or things) that were more common (popular, fashionable, or widely circulated) when one was younger (in early adulthood, in adolescence, in childhood, or even before birth)" (Holbrook & Schindler, 1991, p. 330). e concept of consumer nostalgia has received much a ention with almost exponential increase in the amount of research during the last decade. Surprisingly, li le has been published on consumer nostalgia in terms of literature review. is paper represents an a empt to comprehensively assess the research in the eld of consumer nostalgia literature and to contribute to the consumer nostalgia theory in two major ways.
First, developments in the literature are distinguished by summarizing the main previous research ndings, data, samples, methods and techniques, geographic distribution, thus establishing trends regarding the theoretical evolution of the consumer nostalgia eld and creating a knowledge digest on the subject. Speci c recommendations regarding the development of nostalgia nomological network are provided to aid the continued theoretical and methodological improvements in the area and suggest prospects for the future investigation in nostalgia research domain.
Second, the paper seeks to provide reconsideration of conceptualization and measurement perspective of nostalgia, thus delineating guidance toward capitalizing on the methodological strengths and avoiding the pitfalls of the nostalgia research.
is study discusses the conceptualization and dimensionality of the most widespread nostalgia proneness scale developed by Holbrook (1993) and proposes an alternative measurement perspective.
Although this paper is of particular importance and relevance for emerging economies, due to its conceptual and methodological nature the research ndings are not limited to speci c regions or countries and may be as well useful for overall consumer nostalgia research domain.
e study is organized as follows. First, it provides conceptual background of nostalgia research by reviewing studies in this eld published since 1991. Next, it presents existing measurement instruments of nostalgia and discusses their dimensionality, reliability, and validity. ird, it provides alternative conceptual and theoretical discussion in order to demonstrate why formative measurement may be suitable for nostalgia research. Finally, it discusses implications of this research and delineates guidelines for decision when to employ formative versus re ective indicators. e following literature review is a narrative review of papers published from 1991 (when the de nition of consumer nostalgia was rst used by Holbrook & Schindler) to 2013. e period chosen not only marks the rst conceptualization a empts of nostalgia, but also provides a possibility to summarize and explore more than two decades of nostalgia research. To select the studies for the review, the articles were rst identi ed by searching in scienti c databases such as EBSCO, Emerald, JSTOR, Sage Publications. As there are not many publications on nostalgia, additional search for all relevant scienti c articles mentioned in existing research was performed using the internet search engines. Search categories "nostalgia" and "nostalgic" were used. As in this paper I do not intend and cannot provide a comprehensive review of all nostalgia works, analysis is limited to research in consumer behavior and marketing area and on studies addressing conceptual or theoretical developments in the consumer nostalgia research eld. us, studies that addressed purely psychological or sociological issues of nostalgia were excluded from further analysis. e papers were categorized according to whether they addressed the conceptual foundations (antecedents), implications and outcomes, or conceptualization and operationalization of the nostalgia construct. As measurement issues of the nostalgia construct have been widely discussed over decades, separate analysis of nostalgia measurement is performed. In total 26 papers are reviewed, which represent 17 scholarly journals and proceeding papers from academic conferences. According to the study se ings, most nostalgia studies are concentrated in the USA, and only several studies were carried out in France, the United Kingdom, the Republic of South Africa, Taiwan and the Russian Federation.
Initially nostalgia was identi ed by physicians as a cerebral disease, and was seen as a mental illness that caused depression-like symptoms (McCann, 1941;Rosen, 1975). Davis (1979), who was the rst to explore nostalgia from the sociological point of view, found that many positive sentiments are expressed in regard to nostalgia, and nostalgia began to be viewed in a more positive light. Nostalgia is associated with increased social bonds, increased positive self-regard, as a coping mechanism in which individuals respond to negative moods with positive memories (Wildschut, Sedikides, Arndt, & Routledge, 2006). Sierra & McQui y (2007) explored nostalgia e ects from a di erent angle and applied social identity theory (SIT) perspective. Individuals collectively de ne themselves in terms of unique characteristics (e.g., being raised during a certain time period) and nostalgia is evoked when consumers reminisce about their past social identity (Sierra & McQui y, 2007). Both tangible and intangible stimuli can evoke nostalgia and have the capacity to in uence consumer behavior such as the purchase of nostalgic products (Sierra & McQui y, 2007).
Conceptualization of nostalgia was provided and extensively researched by consumer behavior studies. Holbrook & Schindler (1991) de ned nostalgia as "a preference (general liking, positive a itude, or favorable a ect) toward objects (people, places, or things) that were more common (popular, fashionable, or widely circulated) when one was younger (in early adulthood, in adolescence, in childhood, or even before birth)". Havlena & Holak (1991) and Stern (1992) have clari ed this conceptualization by proposing that nostalgic thoughts may be generated from either a personally remembered past (personal nostalgia) or from a time in history before one was born (historical/communal nostalgia).
Nostalgia a aches to an object related experiences that have been lost and are not available anymore and there may be many reasons, for instance, the relevant objectrelated experiences have become di cult to obtain or the consumer had to change consumption pa ern due to changes in tastes, geographical displacements, or even losses caused by res, earthquakes or other natural disasters Sayre, 1994).  visually illustrate this a ribute with a penchant for eating ice cream -if this is the same ice cream that was available in one's childhood, this experience can't be considered nostalgic. Truly nostalgic sentiment would be longing for the vanilla-avoured rennet custard that one's mother used to cook on the stove before the days of prepackaged pudding and which no longer appears on sale.
e emergent amount of literature on nostalgia provides some propositions and ndings regarding the nature of the nostalgic experience. Among these are that nostalgia occurs in response to negative mood and the discrete a ective state of loneliness (Wildschut et al., 2006), nostalgia is distinct from homesickness, which is a longing for one's home during a time of absence (Van Tilburg, Vingerhoets, & Van Heck, 1996), individuals' nostalgic memories tend to be selective and generally positive, that is, they are ltered through "rose-colored glasses" (Belk, 1991;Havlena & Holak, 1991;Holak & Havlena, 1992;Stern, 1992).

Construct conceptualization and categorization studies
A number of consumer nostalgia studies focused on the conceptualization of the construct. e roots of the construct conceptualization in consumer research can be traced back to Holbrook and Schindler article of 1989 in which the authors found that preferences toward popular music appear to re ect tastes acquired during late adolescence or early adulthood. e development of tastes for popular music follows an inverted U-shaped pa ern that reaches a peak in about the 24th year. e term "nostalgia" was later de ned by Holbrook & Schindler (1991) and further conceptualization of nostalgia was provided by Havlena & Holak (1991), Stern (1992), Baker & Kennedy (1994), Rousseau & Venter (1999, Goulding (2001), Pascal, Spro , and Muehling (2002) and others.
For example, Rousseau & Venter (1999) proposed the nostalgia model that a empts to incorporate cognitive, a ective and action tendency components in the following categories: the in uencing variables (e.g., individual, demographic factors), impact areas (which relate to arts, culture, consumer products, fashion, etc.), manifestations (which imply quality, aesthetics, acquaintance) and outcome based actions (e.g. consumer preference, purchase, consumption pa erns). e most extensive categorization of nostalgia was proposed by Holak, Havlena, & Matveev (2006) Baker & Kennedy (1994) call "real nostalgia" and it has been the subject of most psychological and sociological analysis. 2. Interpersonal nostalgia (indirect individual experience) refers to nostalgic experience based on direct experience and the memories of other individuals, for example, intergenerational nostalgia may be communicated from parents or grandparents. Interpersonal nostalgia evokes less intense feelings and may produce a less complex emotional pro le with regard to the original stimulus than personal nostalgia. 3. Cultural nostalgia (direct collective experience) involves direct experience that is common across members of the group, for instance, presence of reminiscences of Woodstock or similarities across families in celebrations of anksgiving and Christmas. 4. Virtual nostalgia (indirect collective experience) is based upon fantasy and indirect experience and may originate from books, video materials, or conversations with experts and scholars (who themselves have no direct experience with the object of the nostalgia). Holak et al. (2006) conclude that due to di erences in their origins, the four classes of nostalgia may involve substantially di erent responses. Personal and cultural nostalgia are likely to be much richer, complex experiences than interpersonal or virtual nostalgia. On the other hand, cultural and virtual nostalgia, because of their collective emphasis, will probably be much more consistent across individuals than personal or interpersonal nostalgia. As a result, most business uses of nostalgia in advertising and product design emphasize subjects likely to evoke cultural or virtual nostalgia (Havlena & Holak, 1991;Stern, 1992).
Numerous studies were conducted to research nostalgia impact in advertising and it has been found that advertisement eliciting nostalgic reactions are capable of generating a more favorable perception of an ad and advertised brand and of contributing to greater purchase likelihood (Pascal et al., 2002), nostalgic cues in advertising in uence the type of thoughts consumers have during ad exposure, and that these thought processes appear to have an in uence on a itudes toward the advertisement and advertised brand (Muehling & Spro , 2004), positive relationship of ads was found between individual nostalgia proneness and the nostalgia intensity towards advertisement and brand (Reisenwitz, Iyer, & Cutler, 2004), previously heard old songs have positive ad e ects due to evoking consumers' good moods or by generating more favorable nostalgiarelated thoughts (Chou & Lien, 2010).

Antecedents
Despite the facts that literature still lacks consistency regarding a possible relation between nostalgia and its antecedents, two broad categories of nostalgia antecedents may be summarized that are either mentioned theoretically or empirically tested in previous research. e categories involve demographic and socio-psychological antecedents. e following section provides a summary of sometimes con icting results of previous work on the antecedents and outcomes of nostalgia.
Age. Age as a demographic variable has been among most widely studied antecedents of nostalgia.
ere are two streams of research into the age variableone stream focuses on evaluating age as a variable in uencing nostalgic consumption preferences (age as nostalgia antecedent), the other explores how age and nostalgia proneness are connected with each other (temporal aspect -age as a chronological variable when nostalgic preferences are formed). e research into the age antecedent dates back to the seminal work of Davis (1979), who developed several hypotheses regarding the intensity of nostalgia-proneness over the individual's life cycle. Nostalgia was viewed as an adaptive capacity during transition, and individuals should be more prone to experience it during transitional periods, for example, people who move into a "mid-life crisis", retirement, cope with the loss of a loved one, divorce, or change their careers would be particularly prone to nostalgia. Although Davis proposed that nostalgia-proneness is an individual trait that would be in uenced by these individual and demographic factors, his use of a small convenience sample of twelve interviewees did not allow for testing of these hypotheses in a systematic, scienti c manner (Holak et al., 2006). e further research into age and nostalgic preferences was carried out by Holbrook and Schindler (1989), who found that preferences toward popular music appear to re ect tastes acquired during late adolescence or early adulthood. e development of tastes for popular music follows an inverted U-shaped pa ern that reaches a peak in about the 24th year. It is worth paying a ention to the fact that age was discovered rather to be an antecedent of nostalgic songs preferences than nostalgia as a construct.
is relationship was later re ned in a study by Holbrook in 1993. Both age (as a chronological variable) and nostalgia proneness (as an individual characteristic) are logically connected to nostalgia-related preferences. ese two measures represent di erent constructs and individual propensity towards nostalgia proneness operated independently of the aging process (Holbrook, 1993). e author concludes that even when age varies over a wide range in the sample of interest, the e ect of nostalgia proneness works independently of age. us, in general, older respondents relatively tend to prefer earlier lms, whereas those higher in nostalgia proneness show di erential preferences for tender musicals. Both phenomena are nostalgic, but the former refers to temporally related aspects of age, the la er to sentiment aspects of nostalgia proneness as a psychographic variable. One cannot capture consumption phenomena related to nostalgia by looking at either age or nostalgia proneness in isolation. Similar results were replicated by Holbrook &Schindler in 1994 andin 2003. However, somewhat contrary ndings were obtained by Rousseau & Venter (1999). Using Holbrook's nostalgia index, they found that age is signi cantly related to consumer nostalgia. Older Xhosa speaking respondents in the lower middle income group with a primary or secondary school education scored the highest on nostalgia, while young English speaking respondents in the upper income group with a tertiary education scored the lowest on nostalgia. Similar results were replicated in a study by same authors in 2000.
Moreover, mixed results were obtained by Reisenwitz et al. (2004), who used Holbrook's nostalgia index and concluded that no positive relationship between age and societal nostalgia proneness was discovered, but the relationship between individual nostalgia proneness and age existed.
In contrast to Holbrook & Schindler ndings, in an extensive perfume consumption study in France, Lambert-Pandraud & Laurent (2010) found that nostalgia, in the sense of maintaining a preference for perfumes encountered during a person's formative stage, has relatively li le in uence, contrary to Holbrook & Schindler's (1989, 1994 assertion that people's preferences peak for cultural and hedonic products they encounter during their formative years. Only a minority of consumers older than 30 years of age have nostalgic perfume preferences for perfumes they encountered before they were 30. erefore, nostalgia does not o er a main explanation of older consumers' observed tendency to use older perfumes (Lambert-Pandraud & Laurent, 2010). Summarizing, no rm relationship among age and nostalgia can be established as mixed and sometimes contradictory results were obtained in previous research. It remains unclear if nostalgia phenomenon is pertinent to more senior consumers reminiscing about the good old days, or is a stable individual disposition that some people develop early in life.
Gender. e empirical research into gender relationship towards nostalgia proneness was less extensive and in earlier studies it was hypothesized that males tend to be more nostalgic than females (Davis, 1979). Holbrook & Schindler (1989) concluded that each gender may experience and/or express nostalgia in di erent ways. Stern (1992) proposed that this issue relates to the socially-constructed nature of gender and each sex may evaluate di erent stimuli as nostalgic and may articulate nostalgia responses with di ering intensity. Men and women may respond di erently to di erent nostalgic elements in advertisements just as they respond di erently to di erent "romance" or fantasy literature (Stern, 1992). Similarly, Baker & Kennedy (1994) proposed that men and women di er in the items which evoke feelings of nostalgia and those di erences may depend on the product category and the time in the person's life that is being considered. Holbrook & Schindler (1993) researched a itudes towards movies and found that women appear to be marginally more nostalgia-prone than men scoring on Holbrook's nostalgia index. Rousseau & Venter (1999) found no signi cant relationship among gender and nostalgia proneness. However, in their later study (2000) females scored on average higher than males on nostalgia and consumer nostalgic preference (a new construct developed by authors), although the di erences were non-signi cant. Muehling & Spro (2004) found that analyses regarding the potential moderating e ects of gender on individuals' brand and ad a itudes yielded no signi cant interaction results. Reisenwitz et al. (2004) found women are more nostalgia prone than men from both an individual as well as a societal nostalgia proneness perspective. us, gender di erences are associated with stronger or weaker nostalgia, for some products expressed more strongly in men, for other -in women, and the relationship needs to be further researched more profoundly.
Other demographic variables. Other demographic variables such as language, income, education were analyzed in several studies by Rousseau & Venter. In a study in 1999 they analyzed the relationship between language, income and education on one side and nostalgia on the other and found that older Xhosa speaking respondents in the lower middle income group with a primary or secondary school education scored the highest on nostalgia, while young English speaking respondents in the upper income group with a tertiary education scored the lowest on nostalgia. Socio-demographic variables language and income are signi cantly related to nostalgia. Researchers conclude that di erences between levels of nostalgia would emerge amongst respondents in the sample, thus con rming the previous ndings by Mc Cracken (1998) that di erences in nostalgia may occur at a cultural level (it is assumed that language is an indicator of culture) (Rousseau & Venter, 1999). ese results were replicated by the study of the same authors in 2000. In this study the authors distinguished another construct -consumer nostalgic preference and found that demographic variables have similar e ects on both nostalgia and consumer nostalgic preferences. Older Afrikaans speaking female respondents from lower income category with a school education scored highest on nostalgia, while young Xhosa speaking respondents in the upper middle income category with university education scored highest on progressiveness. With regard to consumer nostalgic preference, older Afrikaans speaking female respondents with college/technical education in the lower income category scored highest on this factor. (Rousseau & Venter, 2000). Rousseau & Venter (1999) summarized the possible antecedents of nostalgia by proposing that the variables in uencing nostalgia can be individual (learning perception, personality, resistance to change), environmental (culture, social factors), demographics (age, income), psychographics (lifestyle, values, AIOs). Regre ably, few of the antecedents mentioned in these propositions were tested empirically (with exception of demographic variables). Such personality trait as materialism was studied together with nostalgia by Rind eisch et al. (2000). For products with a high degree of public symbolism and consensually recognizable meanings such as the new VW Beetle and the Lexus GS300, materialism is a stronger predictor of preference and choice than nostalgia, however, neither materialism nor nostalgia in uence the preference or choice for products that a empt to combine both materialistic (i.e., luxury and status) and nostalgic appeals. Rousseau & Venter (1999) tested relationship among nostalgia and propensity for vintage-antiques, anti-modern art and fashion, pro-modern technology/entertainment constructs. Holbrook's 20-item nostalgia scale was factor analyzed and two factors were extracted, namely, nostalgia and progressiveness. Moderate signi cant positive correlations occur between nostalgia and vintage-antiques, as well as between nostalgia and anti-modern art and fashion. However, signi cant and positive relationship among progressiveness and pro-modern technology/entertainment occurred, which is somehow contrary to the general understanding of nostalgia construct. In later study Rousseau & Venter (2000) re ned the scales and used not only Holbrook's scale (divided into nostalgia and progressiveness), but also newly developed consumer nostalgic preferences scale and vintage/antiques propensity scale. Strong positive correlation was found among nostalgia and consumer nostalgic preferences, but no relationship among progressiveness and consumer nostalgic preferences. e positive correlation was also observed among nostalgia and vintage/antiques propensity, as well as among consumer nostalgic preferences and vintage antique propensity.

Social and psychological antecedents
Furthermore, prospective avenues for nostalgia antecedents research were o ered in a qualitative study of living museum visitors by Goulding (2001), who found that for those who did use the museum as a platform for nostalgia, there were di erences in the source of the reaction, and its personal signi cance. Consequently, two separate categories were developed to di erentiate between behaviors: existential and aesthetic. e ndings have identi ed that nostalgic experience is based on four major themes relating to the nostalgic reaction: the number and nature of roles occupied by the individual, the degree of alienation experienced in the present, the quality of and desire for social contact, and the ability to selectively recall the past, which results in either rst-order or vicarious nostalgia (Goulding, 2001).

Outcomes and moderators
Rousseau & Venter (1999) claim that from marketing perspective the nostalgia outcomes are of utmost importance in understanding consumer preference, purchase and consumption pa erns. Nevertheless, previous research has scarcely addressed nostalgia outcomes from the marketing perspective and an overwhelming amount of literature focused either on preferences for nostalgic products or perceived outcomes (such as a itudes towards the ad or brand). For instance, preference (liking) as nostalgia outcome was researched by Holbrook & Schindler (1989, 1991, 1993, 1994, 2003, and Rind eisch et al. (2000). Another stream of research concentrated on the a itudes towards nostalgic advertising (e.g., Baker & Kennedy, 1994;Pascal et al., 2002;Muehling & Spro 2004;Reisenwitz et al., 2004;Muehling & Pascal, 2011). Baker & Kennedy (1994) made proposition that the a itude or a ect associated with the ad is independent from the nostalgia associated with the ad, and the nostalgic feeling evoked by the ad is not entirely mediated by the a itude towards the ad. Pascal et al. (2002) provided empirical evidence that advertisements eliciting nostalgic reactions are capable of generating more favorable perceptions of an ad and advertised brand, and of contributing to greater purchase likelihood. ey found that evoked nostalgia was a signi cant predictor of a itude towards ad and advertised brand. e results regarding likelihood of purchase were mixed and the hypothesis that the more nostalgia an ad evokes, the greater is the purchase likelihood was only partly con rmed. e hypothesis was supported for the brand of Kodak, but only marginally supported for Toshiba. e analyses revealed that ad a itude did mediate the relationship between ad-evoked nostalgia and brand a itude. In 2004 Muehling & Spro found similar results -individuals who were exposed to the nostalgic ad held more favorable ad a itudes and more favorable brand a itudes than did individuals exposed to the non-nostalgic ad. Later Muehling & Pascal (2011) empirically con rmed the hypothesis that exposure to a personal nostalgic ad would generate more self-directed thoughts among study participants than would a historical nostalgic ad or a non-nostalgic ad. Reisenwitz et al. (2004) empirically con rmed that positive relationship exists between nostalgia proneness and nostalgia intensity towards the advertisement and towards the advertised brand. Ford & Merchant (2010) found that appeals for charity that evoke personal nostalgia will have an e ect on the charitable-donation intentions of consumers. In Study 1, nostalgic charity appeals evoke higher levels of emotions and donation intentions than non-nostalgic appeals. Study 2 indicates that this e ect is moderated by the consumer's propensity towards being nostalgic. In Study 3 the e ect of nostalgia emotions and intentions is moderated by the importance of the memory evoked.
Regre ably, only several studies focused on actual purchases of nostalgic goods. Lambert-Pandraud & Laurent (2010), who studied nostalgia outcomes and dependent variables -preference for brand and actual ownership, concluded that nostalgia, in the sense of maintaining a preference for perfumes encountered during a person's formative stage, has relatively li le in uence, in contrast with Holbrook & Schindler's (1989, 1994 assertion that people's preferences peak for cultural and hedonic products they encounter during their formative years. At the same time a study by Loveland et al. (2010) found that increased preference for nostalgic products is experienced by the consumers for whom the need to belong is an active goal experience. Moreover, this research demonstrates that the consumption of nostalgic products, rather than the exposure to or the mere selection of nostalgic products, successfully satiates the need to belong.

Products and stimuli
Over the past two decades of nostalgia research multiple products, product categories and services have been tested empirically. e research was more extensive into culturally loaded products such as music hits (Holbrook & Schindler, 1989, 1991, movie stars (Holbrook & Schindler, 1994), diseased celebrities (Evans et al., 2010). Such durable goods as automobiles were tested by Rind eisch et al. (2000) and , branded products (perfume) by Lambert-Pandraud & Laurent (2010), heritage (visiting living museum) by Goulding (2001), wide array of retro products and brands such as movies, automobiles, jeans or cereals (Brown, Kozinets, & Sherry, 2003a;, previously popular movies, television programs, cookies, crackers, shower gel, soup, candy, and cars (Loveland, Smeesters, & Mandel, 2010), nostalgia in uence on charity donations was explored by Ford & Merchant (2010).
Another stream of research concentrated on nostalgic e ects in advertising and tested advertisement of durable and non-durable photo products (Pascal et al., 2002;Muehling & Spro , 2004;Muehling & Pascal, 2011), advertising for low-involvement product categories (food and household cleaning products) (Reisenwitz et al., 2004), advertising in general (Baker & Kennedy, 1994;Stern, 1992). e e orts to develop theoretical proposition into products categories were more extensive. Sierra & McQui y (2007) proposed to widen the list to music, toys, literature, movies, artwork, clothing, sports memorabilia, candies, furniture, vehicles, technology, outdoor equipment, reworks, home, perfume. Similarly, Rousseau & Venter (1999) hypothesized that e ects of nostalgia can be expressed in such impact areas as arts, cultural entertainment, consumer products, technology, fashion and clothing, collection of antiques. Holak & Havlena (1992) regarded the phenomenon more deeply and proposed that family, home, persons, objects, events, sights, smells, tastes serve as potent stimuli for nostalgia.  concluded that nostalgic bonding occurs ubiquitously and takes a variety of forms such as sensory experience, homeland, rites of passage, friendships and loved ones, gi s of love, security, breaking away, art and entertainment, performance and competence and creativity. Regre ably, few of the propositions and relationship among nostalgia and speci c product groups were tested empirically in wider quantitative studies.

Methodological background and nomological validity
Out of 24 qualitative studies, the sample of 16 studies included adults or a mixture of adults and students, and 8 studies employed students samples, particularly, when research was concentrated on nostalgic e ects in advertising. Quantitative research samples ranged from 108 to 555 with one extremely large sample of 130,411 consumers. Most of the research was carried out using non-probability convenience samples by selfadministered procedures. Multiple regression, factor analysis or other more traditional statistical techniques were most common in quantitative studies of nostalgia. Only 2 out of 25 analyzed nostalgia papers employed structural equation modeling (SEM) for research results analysis. It is worth noting that SEM has become one of the techniques of choice for researchers across disciplines and increasingly is a 'must' for researchers in the social sciences (Hooper, Coughlan, & Mullen, 2008) that allows the researcher to build, test and con rm models of complex relationships (Gallagher, Ting, & Palmer, 2008). Schumacker & Lomax (2004, p.7) provide four major reasons why to conduct SEM. First, use of multiple observed variables helps be er understand scienti c inquiry and deal with sophisticated theories, statistical models and explore complex phenomena. Second, greater recognition can be expected to be given to the validity and reliability of observed score from measurement instruments. SEM explicitly takes into account measurement error when statistically analyzing the data. ird, SEM techniques have advanced rapidly over the last 30 years and are able to analyze more advanced theoretical models, and give additional analytical capacities to the researchers. Fourth, SEM so ware programs have become increasingly user-friendly.
Use of state-of-the-art statistical techniques for examination of complex marketing constructs is closely related to nomological validity of the research area. Figure 1 shows an integrative model of nostalgia antecedents, moderators, related constructs and outcomes that allows investigating both the theoretical relationships between di erent constructs and the empirical relationships between measures of those constructs (Netemeyer, Bearden, & Sharma, 2003, p. 82). It is evident that nomological validity can be further re ned and established in consumer nostalgia eld of research.
Summarizing the studies of nostalgia antecedents, a ention should be drawn to the fact that no rm consensus on antecedents e ects has yet been established, for example, some authors argue that age is an important predictor of nostalgia phenomenon, others provide contrary ndings. e research into other demographic antecedents is at its initial development stage and only several authors provide empirical evidence on nostalgia relationship with such antecedents as gender, education or income. A number of social and psychological antecedents remain at the propositions and hypotheses level or lack the accumulated empirical quantitative support and validation from other studies, countries and researchers. Rousseau & Venter (1999) proposed that not only consumer preference, but also actual purchases and consumption pa erns can be nostalgia outcomes. However, during the last two decades most of nostalgia outcomes investigation remained focused on perceived consequences. Strong paths have been established and con rmed by many authors between nostalgic advertisement and positive a itude towards brand or advertising, nostalgia proneness and preferences. However, a empts to establish the path among nostalgia proneness and purchase intent were not so successful and only one study (Lambert-Pandraud & Laurent (2010) empirically tested the relationship between nostalgia proneness and actual perfume ownership and found relatively li le in uence. us, despite the extensive amount of research that has been conducted, a fundamental question of the generalizability of nostalgia e ects remains unanswered. is might be caused by substantial di erences in research and methodological designs, di erent nostalgia measurement scales used, di erent products studied, data collection modes, or respondent bases. erefore future marketing research should concentrate on nostalgia antecedents and outcomes that might help to understand the phenomenon more deeply and gain additional insights. Nostalgia e ects may be context dependent and exist only under certain conditions, therefore, instead of seeking generalizations and universality, researchers should be more focused on the identi cation of variables that explain di erential e ects. Additionally, reliable and valid measures should be further re ned and developed as it seems that no consensus on nostalgia measurement has been reached so far.

Measurement overview
Out of 18 analyzed quantitative studies, 11 studies used Holbrook's nostalgia scale (8 or 20 item), ve studies used advertising evoked nostalgia scale (developed by Pascal et al. (2002) using a 10-item scale adapted from Holbrook's scale), and two studies used the advertising evoked nostalgia scale proposed by Baker & Kennedy (1994). Other studies either employed newly developed scales or combined Holbrook's scale with new scales. Rousseau & Venter (2000) used a newly developed consumer nostalgic preference scale, Holak et al. (2006) proposed another 31 item scale of nostalgia proneness index. A new nostalgia measurement instrument was employed by Sierra & McQuity (2007), including yearning for the past and a itudes about the past. Ford and Merchant (2010) used Batcho nostalgia scale (for details see Annex 1, Table 3). In this study we concentrate on Holbrook's scale as this scale is more widely recognized in marketing and consumer behavior literature. Holbrook's (1993) Nostalgia scale has both a 20-item long form and an 8-item short form and is designed to represent the phenomenon of nostalgia proneness (Holbrook, 1993).  Holbrook's 20-item Nostalgia index (1990, 1993 Holbrook's 8-item Nostalgia index (1990, 1993) 1. ey don't make 'em like they used to 2. Newer is almost always be er 3. In the future, people will have even be er lives 4. ings used to be be er in the good old days 5. I believe in the constant march of progress 6. Yesterday, all my troubles seemed so far away 7. Products are ge ing shoddier and shoddier 8. Compared to our parents, we've got it good 9. Technological change will insure a brighter future 10. When I was younger, I was happier than I am today 11. Today's new movie stars could learn from the old pros 12. I must admit it's ge ing be er, be er all the time 13. e truly great sports heroes are long dead and gone 14. History involves a steady improvement in human welfare 15. Today's standard of living is the highest ever a ained 16. Sometimes, I almost wish that I could return to the womb 17. We are experiencing a decline in the quality of life 18. Steady growth in GNP has brought increased human happiness 19. Compared to the classics, today's music is mostly trash 20. Modern business constantly builds a be er tomorrow 1. ey don't make 'em like they used to 4. ings used to be be er in the good old days 7. Products are ge ing shoddier and shoddier 9. Technological change will insure a brighter future 14. History involves a steady improvement in human welfare 17. We are experiencing a decline in the quality of life 18. Steady growth in GNP has brought increased human happiness 20. Modern business constantly builds a be er tomorrow Holbrook twenty statements were originally generated to represent the domain of the construct. Ten of these items were reversed items. Exploratory and con rmatory factor analyses, using two samples and movies as stimuli, were performed by Holbrook (1993). Initial con rmatory factor analysis showed poor t and a stepwise procedure was employed to eliminate low loading items and a er this procedure, an eight-item nostalgia scale was developed. With the rst sample it showed adequate unidimensionality, coe cient alpha, summated scale construct reliability estimates of internal consistency of 0.78, factor loadings ranged from 0.49 to 0.76 (p<0.01). e second sample replicated unidimensionality of the 8-item scale (coe cient alpha, construct reliability estimates of 0.73, factor loadings ranging from 0.34 to 0.60 (p<0.01)) (Holbrook, 1993).

Dimensionality of nostalgia scale
Holbrook's scale is widely recognized and used in many marketing and consumer behavior studies to measure nostalgia, indeed numerous authors reported about multidimensionality of the scale. For example, Rind eisch et al. (2000) provided evidence that nostalgia scale appeared to be multi-dimensional in nature. e authors nd that both exploratory and con rmatory factor analysis procedures reveal that the 8-item scale appears to consist of two separate dimensions. e rst dimension consists of such items as " ings used to be be er in the good old days", while the second dimension includes such items as "Modern business constantly builds a be er tomorrow. " Rind eisch et al. (2000) term the rst dimension "Product-Nostalgia" as the items in this dimension appear to re ect nostalgic feelings regarding products or objects. Likewise, they term the second dimension "Life-Nostalgia" as the items in this dimension appear to re ect nostalgic feelings about life in general.
Spro & Silverman (2000) provide additional con rmation of this proposed dimensional structure by nding identical pa erns of factor loadings in applications of the scale in two of their studies. In addition, researchers nd that the correlation between these two dimensions is relatively modest (i.e., r≤.30), and that the reliability of the two dimensions is superior to the reliability of the overall scale. Rousseau & Venter (1999 apply the scale in Eastern Cape and conclude that factor analysis revealed two factors of the Holbrook nostalgia scale, namely, a itudes towards nostalgia and a itudes towards progressiveness. Reisenwitz et al. (2004) use an eight-item Holbrook's nostalgia scale and nd that items load on 2 dimensions with eigenvalues greater than 1 -macro and micro. e reliability of each factor was 0.81 and 0.78 respectively. Each component was represented by four statements. e macro (or reverse-scored) statements included the following: "Technological change will insure a brighter future", "History involves a steady improvement in human welfare", "Steady growth in GNP has brought increased human happiness", "Modern business constantly builds a be er tomorrow". e micro statements included the following: " ey don't make 'em like they used to", " ings used to be be er in the good old days", "Products are ge ing shoddier and shoddier", "We are experiencing a decline in the quality of life". In Reisenwitz et al. (2004)study the macro factor is called societal nostalgia proneness and the micro factor is referred to as individual nostalgia proneness. Borges & Boulbry (2003) apply the 8-item scale to France and nd 2 dimensions which are called present and future temporal orientation and past temporal orientation. e rst dimension comprises Present and future temporal orientation and includes statements "Technological change will insure a brighter future", "Steady growth of GNP (Gross National Product) has brought increased human happiness", "History involves a steady improvement in human welfare" and "Modern business constantly builds a be er tomorrow". In their work the rst factor is referred to as nostalgia-tradition scale. e second dimension is called past temporal orientation and comprises items: " ings used to be be er in the good old days", " ey don't make 'em like they used to", "Products are ge ing poorer and poorer in quality", and "We are experiencing a decline in the quality of life". e authors conclude that "the application of the unidimensional nostalgia proneness seems to be inappropriate in France". Evans et al. (2010) used principal components factor analysis to assess the convergent and discriminant validity properties and revealed that the 8-item scale comprises 2 dimensions: the factors are identi ed as a nostalgia-tradition Scale and a Nostalgia-Progress Scale. Items "Technological change will insure a brighter future", "Steady growth of GNP (Gross National Product) has brought increased human happiness", "History involves a steady improvement in human welfare" and "Modern business constantly builds a be er tomorrow" comprise the rst factor that is named nostalgia tradition scale. Items " ings used to be be er in the good old days", " ey don't make 'em like they used to", "Products are ge ing poorer and poorer in quality", and "We are experiencing a decline in the quality of life" comprise the second factor and are named nostalgia-progress scale. Valid measures of nostalgia-proneness are necessary both for the identi cation of nostalgic consumers for market segmentation and for the testing of hypotheses regarding the nature and determinants of the trait (Holak et al., 2006). Rind eisch & Spro (2000) conclude that while the Holbrook nostalgia scale appears to exhibit both convergent and discriminant validity, this measure appears to be multidimensional in nature, as consumers' nostalgic sentiments for products may be conceptually and empirically distinct from their nostalgic sentiments toward life in general: "As nostalgia research advances in the coming years, we believe that the speci cation and measurement of how consumers react to nostalgia will take on increased importance" (Rind eisch & Spro , 2000). Rind eisch et al. (2000) suggest that future research on nostalgia should pay close a ention to the dimensionality and reliability of the nostalgia scale itself.

Reconsideration of nostalgia construct measurement
Nostalgia research literature review provides evidence that since 1991 work in this eld has been based on traditional measurement theory (classical test theory) -nostalgia measurement model is represented as e ects (re ective) indicators. is re ective approach assumes (1) that equally reliable indicators are interchangeable and they can be substituted for one another without a ecting construct de nition, (2) indicators of the same factor have positive intercorrelations, (3) factors are conceptualized as unidimensional latent variables (Kline, 2011, p. 280). Indeed, for some constructs, it makes more sense conceptually to view causality owing from the measures to the construct, rather than vice versa ( Jarvis, MacKenzie, & Podsako , 2003;Bagozzi, 1981Bagozzi, , 1984Fornell & Bookstein, 1982). Diamantopoulos, Rie er, & Roth (2008) refer to Bollen's (1989, p. 65) statement "[M]ost researchers in the social sciences assume that indicators are e ect indicators. Cause indicators are neglected despite their appropriateness in many instances" and make a call to encourage the thoughtful application of formative models (Diamantopoulos et al., 2008). Jarvis et al. (2003) give a warning that potentially serious consequences of measurement model misspeci cation exist, and researchers need to think carefully about the direction of causality between constructs and their measures. Such measurement model misspeci cation can create measurement error, which in turn a ects the structural model ( Jarvis et al., 2003;MacKenzie, Podsako , & Jarvis, 2005), can have a dramatic impact on one's understanding of theory (Edwards & Bagozzi, 2000) and can lead to both Type I and Type II errors (Pe er, Straub, & Rai, 2007). Diamantopoulos et al. (2008) summarize studies empirically examining the consequences of measurement model misspeci cation on parameter estimates and report serious under-or overestimation of parameters as a consequence of misspeci ed causality, wrongly adopted puri cation procedures, or a combination of both. Diamantopoulos et al. (2008) point out that misspeci cations are not detected by poor t index values and such biases may lead to incorrect conclusions on tested relationships, thus pu ing many empirical results into question. For instance, Jarvis et al. (2003) report that 29 percent of studies published in the top four journals during a 24-year period improperly speci ed formative and re ective constructs and this is by far the most common type of measurement model speci cation error. e di erence between re ective and formative indicators depends on causal priority between indicators and the latent variable in question (Diamantopoulos & Siguaw, 2000, p.21), while the construct causes variance in its re ective indicators, the direction of causality is reversed such that the formative indicators cause variance in the construct (Cenfetelli & Bassellier, 2009); in re ective measures a change in the construct a ects the underlying measures, whereas in formative constructs, changes in the formative measures cause changes in the underlying construct ( Jarvis et al., 2003). Multicollinearity among indicators can be a signi cant problem for formative measurement model, but it is a virtue when the indicators are re ective ( Jarvis et al., 2003). Formative measurement items are designed to tap into the di erent subconstructs and multicolinearity is safeguarded by ensuring that the items do not tap into the same aspects (Pe er et al., 2007). Internal consistency is of minimal importance in formative indicators, and reliability methods based on internal consistency do not apply (Netemeyer et al., 2003, p. 93). Variables that might even be negatively related can both be as meaningful indicators of a formative construct (Diamantopoulos & Winklhofer, 2001;Nunnally & Bernstein, 1994, p. 489). Dropping a causal indicator that possesses low item-to-total correlations may omit a unique part of the formative construct and change the meaning of the variable ( Jarvis et al., 2003;Pe er et al., 2007) and could make the measure de cient by restricting the domain of the construct (Churchill, 1979). Items used as formative indicators must cover the entire scope of the latent variable as described under content speci cation (Diamantopoulos & Winklhofer, 2001).
To the contrary, in the case of re ective construct all the measures are assumed to be equally valid indicators, measures must be internally consistent and are interchangeable ( Jarvis et al., 2003) and unidimensionality is a key assumption within covariance-based SEM for re ective constructs (Pe er et al., 2007). Construct validity is unchanged when a single indicator is removed from re ective construct, although reliability estimates (e.g., Cronbach's alpha) of the set of indicators can be lower if fewer indicators are included in the measurement model (Bollen & Lennox, 1991). In the formative model error is represented at the construct level, whereas in the re ective model -at the individual item level ( Jarvis et al., 2003) and formative models minimize "the trace of the residual variances in the 'inner' (structural) equation" (Fornell & Bookstein, 1982, p. 442). Coltman, Devinney, Midgley, & Venaik (2008) summarize that uncritical and universal application of a re ective structure leads to oversimpli cation of broad, diverse and complex real-world constructs and exposes scholars to the risk of reducing the rigor of business theory and research and its relevance for managerial decision making.
Bearing in mind the above, the next section, in line with Jarvis et al. (2003) suggestions, provides conceptual discussion regarding measurement approach of consumer nostalgia: the possible direction of causality between nostalgia construct and its indicators, interchangeability and covariation among indicators and nomological network.

1) Direction of causality between the construct and its indicators.
For formative measurement models, the direction of causality ows from the measures to the construct, and it ows from the construct to the measures for re ective measurement models.
In nostalgia case the conceptual domain of nostalgia is described as "a preference (general liking, positive a itude, or favorable a ect) toward objects (people, places, or things) that were more common (popular, fashionable, or widely circulated) when one was younger (in early adulthood, in adolescence, in childhood, or even before birth)" (Holbrook & Schindler, 1991, p. 330) (emphasis added). Nostalgia a aches primarily to object related experiences that have somehow been lost either because the relevant object-related experiences have become di cult to obtain or because the consumer in question has moved on to a consumption pa ern in which they are no longer included due to changes in tastes, geographical displacements, or even losses caused by res, earthquakes or other natural disasters Sayre, 1994, emphasis added). Holak et al. (2006) proposed a four-way classi cation of nostalgic experience: personal nostalgia (direct individual experience), interpersonal nostalgia (indirect individual experience), cultural nostalgia (direct collective experience) and virtual nostalgia (indirect collective experience). e authors state that due to di erences in their origins, the four classes of nostalgia may involve substantially di erent responses and that nostalgia experiences are elicited in response to "object," "person", and "event" prompts (Holak et al., 2006) (emphasis added). e study by Holbrook & Schindler (2003a, p. 121) tries to illustrate possible types of nostalgic bonding and concludes that nostalgic bonding occurs ubiquitously and takes a variety of forms: " e informants' reports demonstrate the wide range of object-related experiences that can be connected to memories of the past. e object itself can be large (such as an antique table) or small (such as an engraved gold medal). It can be costly (an 18th century cello) or inexpensive (a candy bar), decorative (an 'atomic' lamp)

or functional (a briefcase), handmade (a picture ame) or massproduced (a television). e object can be edible (cloves), branded (Tropicana orange juice), musical (a clarinet), out of style (old glasses) or unique (family photographs).
Apparently, there is no limit to the types of object that can carry nostalgic feelings". ...... " e informants' vigne es and stereographs provide a glimpse of those life events that are particularly likely to lead to such strong feelings. Among these, relations with other people are certainly a very common source of powerful emotions. Speci cally, the love felt towards a parent, grandparent, child, spouse or signi cant other leads to nostalgic bonding. Furthermore, the comfort and security of a place whether a family environment or a geographic homeland is a classic source of nostalgic feelings. Conversely, the thrill of then new things for example, a rst irtation, an introduction to poetry or a novel consumer purchase appears to be a potentially strong target for nostalgic bonding. Finally, the joys of accomplishment whether relating to tennis victories, to prowess at sewing or to musical performances in the subway are capable of generating nostalgic experiences"(emphasis added). 20-item Holbrook's nostalgia scale was originally developed as potential facet of individual character -a psychographic variable, aspect of life style, or general customer characteristic -that may vary among consumers, independent of time-or age related factors (Holbrook, 1993). Scale development was based on large literature review and developed in explaining preferences towards a large set of 125 products. e operationalization of the scale follows conventional scale development procedures and is based on the assumption that nostalgia proneness is re ected in preference towards objects, people or places. However, what if the conceptualization were another way round -nostalgia proneness is caused by existence of objects that evoke this feeling? Nostalgia proneness occurs and is made of preferences towards objects, people or places. is can be supported by the way  describe nostalgia as "object-related experiences", "leads to nostalgic bonding", "source of nostalgic feelings" etc. Occurrence of these constituents would positively or negatively impact the degree of nostalgia proneness. Following the logic, nostalgia can't occur if there are no objects. Objects cause nostalgia, and not vice versa. Following this proposition, the direction of causality is from items to construct. Changes in indicators should cause changes in the construct, but changes in the construct do not cause changes in indicators.
2) Interchangeability of the indicators. e indicators need not be interchangeable for formative measurement models, but should be for re ective measurement models. e re ective approach requires that individual items share a common theme. Doubts can be raised whether individual items in Holbrook's nostalgia scale domain ful ll this requirement. For example, would an item designed to measure a itude towards business "Modern business constantly builds a be er tomorrow" necessarily be related with one designed to measure a itude towards music or movie stars "Today's new movie stars could learn from the old cinema pros" or "Compared to classics, today's music is mostly trash"?
Misspeci ed measure can lead to a neglect of a key aspect of the focal construct. More speci cally, if one wants to explore nostalgic music preference and erroneously drops an item "Compared to the classics, today's music is mostly trash" from a measure of nostalgia proneness due to not meeting conventional standards for re ective items (i.e., low factor loadings), we lose a key aspect of nostalgia assessment. A number of nostalgia objects may underlie the nostalgia proneness, nostalgia may be personal, interpersonal, historical and virtual (Holak et al., 2006). e diversity of nostalgia phenomena suggests that the formative viewpoint may be more appropriate.

3) Covariation among the indicators.
Covariation among the indicators is not necessary or implied by formative indicator models, but covariation among the indicators is a necessary condition for re ective indicator models.
Conceptually, a change in one of the nostalgia indicators doesn't necessarily imply a change in other indicators. Having high nostalgia proneness in regard to progress perception does not lead and imply higher nostalgia proneness in regard to other objects, for example, favorite music from ones youth. A person may score high on nostalgia regarding sports heroes, but score low on nostalgia regarding products. Formative construct can be represented by mutually exclusive types of behavior ( Jarvis et al., 2003). For example, we have several nostalgia indicators "History involves a steady improvement in human welfare", "Today's new movie stars could learn from the old cinema pros", " e truly great sports heroes are long dead and gone", "Technological change will insure a brighter future", "Modern business constantly builds a be er tomorrow", "When I was younger, I was happier than I am today". ese indicators may be mutually exclusive. For example, a person may think that great sport heroes are gone or modern business builds be er tomorrow, but he doesn't necessarily thinks both.

4) Nomological net of construct indicators.
For the re ective indicator model, since all of the indicators re ect the same underlying construct and are assumed to be interchangeable, they should all have the same antecedents and consequences. However, for the formative indicator model, because the measures do not necessarily capture the same aspects of the construct's domain and are therefore not necessarily interchangeable, there is no reason to expect them to have the same antecedents and consequences.
Again if we look at nostalgia items, we can observe that one group of antecedents and outcomes can be expected to result for nostalgic products perception. A person who thinks that products were be er in good old days will more likely tend to score higher on purchase intent or actual purchase. If we take culturally loaded items (like "Compared to the classics, today's music is mostly trash"), one can expect that these items would in uence outcomes for speci c cultural products like music records purchase, but this doesn't mean that people who score high on cultural nostalgia will tend to buy all nostalgic products and avoid technological and innovative items.
eoretically, if we look more deeply into antecedents, the proposition can be made that for di erent product categories di erent antecedents can be important.
Moreover, nostalgia construct is context speci c: not only country speci c, but also generation speci c, and speci c at the individual level. For example, Woodstock festival can have one meaning to US consumers and completely di erent meaning to China consumers. Or soviet era symbols can have positive virtual nostalgic e ects in a country that was not a ected by occupation, and negative in a country that was a ected by occupation. Objects that cause nostalgia can vary between countries, individuals, and are not universal. is implies that direct comparison of nostalgia indexes may not be possible across countries and di erent generation samples.

Discussion and Conclusions
On measurement. Over the past two decades nostalgia research has assumed that the correct measurement model is a re ective one. is paper o ers an alternative approach for viewing and operationalizing nostalgia construct as a formative construct. Propositions and theoretical reasoning is provided that in some instances the re ective assumption may not be theoretically or empirically justi ed. Diamantopoulos (2011) stresses that constructs themselves are not inherently formative or re ective and formative or re ective constructs are only intended as a shorthand description actually referring to constructs-once-measured (originally stressed words). e choice of the measurement perspective and use of the formative or re ective measurement should be based on the "auxiliary theory" (Diamantopoulos, 2011;Diamantopoulos & Siguaw, 2006;Bagozzi, 1982). Latent constructs are not inherently formative or re ective and the choice of measurement rests on theoretical considerations (Cenfetelli & Bassellier, 2009;Bollen, 2007;Howell, Breivik, & Wilcox, 2007). e substantive theory (which addresses the underlying conceptual properties of constructs), as well as the auxiliary measurement theory (which explains the nature of the relationships between constructs and their measures), should also be considered when deciding on formative versus re ective measurement (Hardin, Chang, & Fuller, 2008;Howell et al., 2007). For example, psychological constructs are best measured using re ective indicators, while constructs determined by an explanatory combination of variables are best measured using formative indicators (Bagozzi, 2007;Fornell & Bookstein, 1982;Howell et al., 2007). e choice to model and analyze a construct as unidimensional (i.e., re ective), formative, or multidimensional depends largely on the construct under study and "the generality or speci city of one's theoretical interest" (Pe er et al., 2007;MacKenzie et al., 2005, p. 713).
us, if the aim of the research is to measure a speci c aspect of nostalgia, represent subjective nostalgic a itudes or self perceived nostalgia, a re ective nostalgia measurement may suit be er. Re ective nostalgia scale was appropriately employed by Baker and Kennedy (1994) as they took original Holbrook's scale and remade it for advertising measurement purpose. eir scale includes items " is ad reminds me of an experience from the past", " is ad makes me think of an experience which I feel sad about because it is over, yet it is a happy memory", " is ad does not make me have any feelings about the past", "I wish I could relive the experience(s) this ad makes me think of ", "I do not think about the past when I look at this ad", "I associate this ad with a happy experience, yet it makes me feel sad". Note that in Baker & Kennedy case the aim is not to measure general nostalgia on all objects. Baker & Kennedy scale measures nostalgia towards only one object -advertising.
However, if the research aim is to explore complex nostalgia phenomenon, to build a holistic index to understand in general what makes a consumer nostalgic, to gain insights what objects cause nostalgia, to cover the whole domain of nostalgia, a formative view might be more appropriate. In the table below the guidelines and propositions are summarized that aim to assist researchers on decision rules whether to employ formative or re ective nostalgia measurement for future research. Nostalgic a itudes towards speci c advertising or speci c products or product groups, towards concrete historical events.
Indexes that capture many nostalgic stimuli, products or objects is ad makes me think of an experience which I feel sad about because it is over, yet it is a happy memory. 3. is ad does not make me have any feelings about the past. 4. I wish I could relive the experience(s) this ad makes me think of. 5. I do not think about the past when I look at this ad. 6. I associate this ad with a happy experience, yet it makes me feel sad.

Holbrooks 20 item Nostalgia proneness index
1. ey don't make 'em like they used to 2. Newer is almost always be er. 3. In the future, people will have even be er lives. 4. ings used to be be er in the good old days. 5. I believe in the constant march of progress. 6. Yesterday, all my troubles seemed so far away. 7. Products are ge ing shoddier and shoddier. 8. Compared to our parents, we've got it good. 9. Technological change will insure a brighter future. 10. When I was younger, I was happier than I am today. 11. Today's new movie stars could learn from the old pros. 12. I must admit it's ge ing be er, be er all the time. 13. e truly great sports heroes are long dead and gone. 14. History involves a steady improvement in human welfare. 15. Today's standard of living is the highest ever a ained. 16. Sometimes, I almost wish that I could return to the womb. 17. We are experiencing a decline in the quality of life. 18. Steady growth in GNP has brought increased human happiness. 19. Compared to the classics, today's music is mostly trash. 20. Modern business constantly builds a be er tomorrow.
One of the main contributions of this study is to show the need for researchers to explicitly justify their choice of re ective or formative measurement models by supporting it with theoretical arguments and empirical evidence. Researchers have to bear in mind that a construct measured with re ective indicators is not necessarily the same construct when measured with formative indicators, even if the construct name stays the same (Diamantopoulos, 2010).
On nomological network. Nostalgia literature review demonstrates that its nomological network may be re ned or strengthened by a number of research initiatives. First, new and continuing research is needed into the antecedents of nostalgia in order to transfer the scienti c knowledge from hypotheses level to empirically con rmed, valid and reliable research. e list of antecedents should be further extended by future research. Another less developed area is the moderators' impact on nostalgia relationship with its outcomes. e accumulated knowledge from other consumer behavior areas suggests that nostalgia e ects on preferences or buying intentions might be mitigated by a number of moderators (e.g., perceived quality, individual involvement type and level, brand name). Also, the research into product type impact on nostalgic preferences can be further extended by examining how nostalgia impacts changes for durable versus non-durable goods, for arts, cultural products, for low-involvement versus high involvement products, for technological and complex products versus fast moving or convenience products, etc. Finally, another interesting research stream might concentrate on empirical studies to determine whether nostalgia e ect applies equally well to intangible services. A good example is provided by Goulding (2001) and might be extended to visiting nostalgic music events, restaurants, etc.
Second, as the number of variables and relationship in the nomological network of nostalgia research increases, greater need for more complex testing emerges. Structural modeling provides basis for empirical grounding and testing of this complex relationship and constructing of more holistic models. Moreover, antecedents, manifestations and outcomes testing in multivariate models can provide some indication of the relative strength of these constructs. Traditional statistical techniques could be blamed for lack of nomological validity in nostalgia research, thus, future studies should try to employ more state-of-the-art techniques and try to explore nostalgia phenomenon testing more complexly the relationships among antecedents, related constructs and consequences.
ird, reconsideration of nostalgia measurement is of utmost importance as none of the above propositions can be completed successfully if the conceptual and theoretical consensus in this eld of research is not well established. Researchers have to be aware of the auxiliary nature of nostalgia construct, clearly de ne the study objectives and choose properly re ective versus formative measurement of nostalgia. Development and validation of re ned nostalgia measurement instruments is a priority for the future research. Speci cally, the following steps in measures development should be of the greatest concern: To develop re ned formatively measured nostalgia index that captures the conceptual and complex domain of the construct. To re ne existing re ective measures of nostalgic a itudes to di erent product categories, se ings, and nostalgia types. For example, historical and virtual nostalgia re ective measurement theoretically should be distinct from personal nostalgia re ective measurement. Finally, taking into account the complexity of nostalgia phenomenon and following recommendations by Pe er et al. (2007), multidimensional measurement of nostalgia construct should be explored to determine whether such constructs dimension can be measured using either re ective or formative indicators.

Limitations
As Hardin et al. (2008) note, respeci cation of the indicators as formative or re ective should not be driven only by examining the comparative lists of properties of formative and re ective measures. is should be used only as a tool for identifying misspeci ed measures. While these two types of indicators may share common aspects of the construct, their speci cation is driven by measurement theory and, thus, should not be examined from any perspective other than their original intent (Howell et al., 2007b). e decision to specify indicators as formative or re ective should be made prior to their use, because the theoretical underpinnings of formative versus re ective measurement are incompatible (Diamantopoulos & Siguaw, 2006;Howell et al., 2007a). However, according to Coltman et al. (2008), a counter-argument is that measuring existing re ective scales as formative constructs represents a conservative test of the proposition that formative measurement is worth considering. us, the propositions of current study can be regarded only as exploratory and providing guidance for future research. In order to capture the whole domain of formative nostalgia construct, the new item generation procedures should be employed involving expert screening, focus groups, content analysis, and other measurement generation procedures recommended by measurement literature. Age and nostalgia proneness working together does play a role in shaping consumption preferences by in uencing pa erns of consumer tastes. ose higher in nostalgia proneness tend to prefer musicals and tenderhearted lms. An eight-item nostalgia proneness scale is developed, no signi cant correlation appeared between age and nostalgia.

Movies
Age, gender Six-item nostalgia scale (Baker, Kennedy, 1994) with a 5-point Likert scale measuring the intensity of feeling from strongly agree to strongly disagree.
Holbrook & Schindler (1994), USA, Journal of Marketing Research e existence of an age-related preference peak previously found for the case of music generalizes to the context of visual preferences for photographs of movie stars. Nostalgia is shown to moderate this tendency, and di erences between male and female respondents suggest that the experience of strong positive feelings plays a causal role. Liking for movie stars peaked at the age of 14. Nostalgic e ects are extended to durable and utilitarian products as automobiles.
Despite the common impression that the tendency toward nostalgia increases with age (e.g., Davis, 1979), none of these three nostalgiaproneness indices provide any evidence of this. e preferences of the Holbrook index's high nostalgia peaked at a product-speci c age of 18 and the preferences of the Taylor/Konrad index's high nostalgia peaked at a product-speci c age of 20, the preferences of the McKechnie index's high nostalgia peaked at a product-speci c age of -37. Because of this substantial disagreement between the nostalgia-proneness indices, it appears appropriate to consider the possibility that these measures are tapping di erent aspects of liking for the past.  Muehling & Spro (2004), USA, Journal of Advertising Nostalgic cues in advertising do indeed in uence the type of thoughts consumers have during ad exposure, and these thought processes appear to have an in uence on a itudes toward the advertisement and advertised brand. Follow-up analyses regarding the potential moderating e ects of age or gender on individuals' brand and ad a itudes yielded no signi cant interaction results  Reisenwitz, Iyer, & Cutler (2004), USA, Marketing Management Journal e authors distinguish societal and individual nostalgia proneness. A positive relationship exists between individual nostalgia proneness and nostalgia intensity towards the advertisement and towards the brand. No positive relationship between age and societal nostalgia proneness was discovered, but relationship between individual nostalgia proneness and age exists. Women are more nostalgia prone than men. Previously heard old songs have positive ad e ects due to evoking consumers' good moods or by generating more favorable nostalgiarelated thoughts. High-relevance lyrics facilitate the production of favorable ad execution-related thoughts, which improve ad a itude directly and indirectly through good moods. Appeals for charity that evoke personal nostalgia will have an e ect on the charitable-donation intentions of consumers. In study 1, nostalgic charity appeals evoke higher levels of emotions and donation intentions than non-nostalgic appeals. Study 2 indicates that this e ect is moderated by the consumer's propensity towards being nostalgic. In study 3 the e ect of nostalgia emotions and intentions is moderated by the importance of the memory evoked. Nostalgia, in the sense of maintaining a preference for perfumes encountered during a person's formative stage, has relatively li le in uence, in contrast with Holbrook andSchindler's (1989, 1994) assertion that people's preferences peak for cultural and hedonic products they encounter during their formative years. Increased preference for nostalgic products is experienced by consumers for whom the need to belong is an active goal experience. Consumption of nostalgic products, rather than the exposure to or the mere selection of nostalgic products, successfully satiates the need to belong. Personal nostalgia (a yearning for one's past) generally outperforms both historical and nonnostalgic advertising when measures of selfdirected thoughts, positive a ect, and a itude toward the ad are considered. However, when cognitive measures (i.e., brand/message-related cognitive responses and message recall) are considered, a personally nostalgic ad is shown to be comparable to a historical nostalgic ad, but inferior to a non-nostalgic ad.