The Impact of Nostalgia and Probability Markers on the Effectiveness of Advertising Creative Strategies

Abstract The current study intends to explore the boundaries of nostalgia-themed advertising with probability markers. An experimental design was implemented with 575 valid responses obtained. The findings validate nostalgia-themed advertising as a creative strategy associated with positive effects on brand attitude and purchase intentions among consumers with “high” past brand attachment. Past brand attachment emerges as an important construct as the effects significantly differ between “high” and “low” conditions. The use of probability markers, however, did not confirm the expected incremental effects on nostalgia-themed advertising, with results being compared with an advertisement with no probability marker. Moreover, the use of hedges and pledges seem to perform in a similar manner among the different types of purchase motivations and level of tolerance to ambiguity. Implications for practice, limitations and future studies are presented.


Introduction
Advertising (ads) efficacy has taken an important share in the agenda of researches as from the 1970s, with new theoretical models investigating the mechanisms behind consumers persuasion and guiding creative strategies (e.g., Aaker & Maheswaran, 1997;Mehrabian & Russell, 1974;Petty et al., 1983). Among the various creative strategies, nostalgia-themed advertising has started capturing the attention of researchers as from the 1990s (Marchegiani & Phau, 2010;Muehling & Pascal, 2011), triggered by the economic and financial crisis (Baker & Kennedy, 1994;Merchant et al., 2013), with nostalgia appeals evoking good times of the past and, therefore, helping to reduce the uncertainty and insecurity.
Nevertheless, many studies have argued in favor of nostalgia and its impact on brands performance (e.g., Muehling et al., 2014;Muehling & Pascal, 2011;Pascal et al., 2002), the mixed nature of the nostalgia experiences has progressively been acknowledged. Thereby, nostalgia emerges as a bittersweet affective state, mood or emotion, related with experiences that recall positive memories of the past, while simultaneously generate some sadness, connected to the impossibility of returning to it. These emotions may have mixed impact for brands with which they are cognitively associated (Baker & Kennedy, 1994;Holak & Havlena, 1992, 1998. The current study explores the conditions under which the effects of nostalgia-themed advertising may have positive effects on brands. While doing that, it is considered the controversy regarding the possibility of nostalgic ads negatively influence the persuasion of high involvement products, as they may distract the analysis of the central elements of the ad. The reason behind this is that nostalgic appeals are processed peripherally (Muehling et al., 2004;Muehling & Pascal, 2011). Thus, these findings challenge the use of nostalgia-themed advertising in contexts where the central cognitive processing is essential for persuasion (as in high involvement product contexts). Concurrently, they also invite researchers to analyze the effects of nostalgia in contexts where peripheral cognitive processing is dominant (as in low involvement products contexts, e.g., Muehling & Pascal, 2012;Chou & Singhal, 2017). Therefore, the current study evaluates whether the effects of nostalgia as a creative strategy has a positive impact in the context of low involvement products.
Moreover, the current study also analysis probability markers (Tipps et al., 2006). Probability markers are adverbs, verbs or other expression words, usually used in advertising language to serve a certain level of confidence in the message that the advertiser aims to pass on (Banks & De Pelsmacker, 2014;Berney-Reddish & Areni, 2005. These resources are part of the language style and represent a set of pragmatic features of the language, which can change the way the recipient understands it (Blankenship & Holtgraves, 2005). Thus, the assertiveness of probability markers is expected to increase the persuasion of nostalgia-themed advertising, as it may contribute to further reinforce the sense of certainty and security which an intended output when using nostalgia appeals (Baker & Kennedy, 1994;Merchant et al., 2013).
In this vein, the decision of combining both (nostalgia and probability markers) finds support on the common characteristics between nostalgiathemed advertising and probability markers concerning the mechanisms of peripheral cognitive processing, with synergies between creative strategies being expected. As a result, the current study evaluates the effects of nostalgia-themed advertising and the use of probability markers (hedges and pledges) as a language style in the context of low involvement products.
In order to explore the boundary conditions within which the effects of nostalgia-themed advertisement with probability markers are expected to occur, this study considers the type of purchase motivation (hedonic and functional) (Banks & De Pelsmacker, 2014;Berney-Reddish & Areni, 2005), and consumerstolerance to ambiguity (Banks & De Pelsmacker, 2014).
In sum, the aspects above mentioned intend to elaborate on the use of nostalgia as a creative strategy which is evaluated in the current research employing the following research questions: i) are nostalgia-themed ads in low involvement physical product contexts effective in driving purchase intentions and brand attitude; ii) does the use of probability markers enhance the effects of nostalgia-themed ads on purchase intentions and brand attitude; iii) do the level of past brand attachment, type of purchase motivation and level of consumers tolerance to ambiguity influence the effects of nostalgia-themed ads on purchase intentions and brand attitude.

Theoretical background
The S (stimuli)-O(organism)-R(response) framework offers theoretical support for elaborating on the research questions proposed. The S-O-R expresses that the stimuli affect the individual' emotional and cognitive states, whose response may result in approach or avoidance behaviors. In the adaptation from psychology proposed by Mehrabian and Russell (1974), stimuli are designed as environmental cues, organisms comprise cognitive and emotional states and responses are accepted as approach or avoidance related types of behaviors. Interpreted in the context of the current research, the stimuli comprise the nostalgia and non-nostalgia themed ads with and without probability markers, the organism aggregates the understanding, liking and acceptance toward the ad (Defever et al., 2011;MacKenzie & Lutz, 1989;Madden et al., 1988) and the response is captured by means of consumers' purchase intentions and brand attitudes (e.g., Singh & Banerjee, 2019;Yu & Chang, 2013). In line with this rational, in the following paragraphs, the literature in nostalgia and probability markers is presented and hypotheses linking these stimuli with respective responses are proposed.

Nostalgia in advertising
The term nostalgia is first coined by Johannes Hofer (1688), to denote a medical illness involving a range of symptoms related to suffering (algos) (e.g., homesickness, anxiety, insomnia, and irregular heartbeat) and associated to the desire of returning home (nostos) (Hepper et al., 2014;Sedikides et al., 2008;Wildschut et al., 2006). Until the middle of the 20th century, nostalgia remained the diagnosis for a disease frequently associated with extreme depression, melancholy, and homesickness (Baker & Kennedy, 1994). Since then, the idea of nostalgia has become a subject of much more pleasant connotations, associated to the feelings that individuals' biographical memories may evoke (Hepper et al., 2014;Juhl et al., 2010;Sedikides et al., 2008;Wildschut et al., 2006). This rational supports the adoption of nostalgia in advertising and simultaneously reveals the origins of its mixed nature and consequent controversy (Baker & Kennedy, 1994;Holak & Havlena, 1992, 1998 As a creative strategy meant for driving persuasion, nostalgia-themed advertising is accepted for being peripherally processed (Chou & Singhal, 2017;Muehling et al., 2004;Muehling & Pascal, 2011). This understanding builds from the Elaboration Likelihood Model (ELM), which distinguishes between the central and the peripheral processing route for persuasion (Petty et al., 1983;Petty & Cacioppo, 1986). ELM suggests that in the decision making of high involvement products, consumers tend to dedicate more time and effort to understand the core of the advertising message (e.g., quality of the argument, the number of arguments), and process its understanding accordingly (central route) (Friestad & Wright, 1994). Whilst in contexts of a low level of involvement, consumers focus their thoughts on the aspects not related to the core merits of the product, but to the peripheric aspects of the message (Petty et al., 1983). In these contexts, persuasion results from consumers doing inferences and reacting to simple positive or negative cues used in the advertising, expressed using aspects like language style or creative theme (Petty & Cacioppo, 1986).
In the case of nostalgia-themed adverting while evoking personal past experiences, they draw upon consumers biographical memories (Meyers, 2009;Stern, 1992) and not on the merits of the product itself. This argument helps to support the proposition that the effects of nostalgia-themed advertising are more effective in low involvement product contexts. In these contexts, the advertising message is processed peripherally, and nostalgia operates as a cue for consumers to elaborate on the meanings of the product (Chou & Singhal, 2017). Effects on persuasion are expected to occur in response to inferences associated with past memories evoked. Mechanisms of image transfer support the effects (e.g., Muehling et al., 2004;Muehling & Pascal, 2011).
The effects of nostalgia-themed advertising may be expected to occur more efficiently when consumers acknowledge their emotional link with the brand in the past, captured by past brand attachment (e.g., Loureiro & Kaufmann, 2012;Muehling et al., 2014). This effect results from the compatibility between the past of the brand and the nostalgic themed appeal, facilitating the mechanisms of image transfer (Hoeffler & Keller, 2002;Bloom et al., 2006;Keller, 2003). The effects above mentioned are captured in the following hypotheses: H1a: The effects of consumers' exposure to the nostalgia-themed advertisement in low involvement product contexts are likely to produce favorable brand attitudes, especially among consumers who have higher past brand attachment.
H1b: The effects of consumers' exposure to the nostalgia-themed advertisement in low involvement product contexts are likely to have a positive impact on purchase intentions, especially among consumers who have higher past brand attachment.

Probability markers in advertising
The effective use of written and spoken language is key in creating persuasive ads (Blankenship & Craig, 2011;Blankenship & Holtgraves, 2005). In the current research, the features of the language style under review are the probability markers. Among the multiple examples of probability, markers are the pledges (e.g., certainly, undoubtedly, always and definitely) and the hedges (e.g., probably, perhaps, may and help). Areni (2002) explains that the pledges suggest an absolute truth, being considered as a powerful language to the extent of conferring the absolute confidence on what is advertised. By contrast, Hedges signalize a probable truth and are used by advertisers in situations where claiming the absolute truth would evoke low credibility. Although both types of probability markers intend to increase the assertiveness of the message, hedges represent a less powerful language, whose function is to instill a probability.
The use of pledges and hedges in advertising is a common practice, despite some contradictory findings (Areni, 2002;Berney-Reddish & Areni, 2005. Results in favor of the use of probability markets argue that they lead to an increasing acceptance of advertising appeals (e.g., truth rating) (Banks & De Pelsmacker, 2014;Harris et al., 1993) and consequently a more favorable attitude toward the brand and greater purchase intention.
Other studies, however, claim different results, with hedges being associated to negative effects on source credibility and pledges working only under some limited situations (Berney-Reddish & Areni, 2005Blankenship & Holtgraves, 2005;Durik et al., 2008;Holtgraves & Lasky, 1999). In reaction to the contradictory results, the use of probability markets is proposed as conditional to specific characteristics associated to the level of product involvement, the type of purchase motivation (hedonic vs. utilitarian), and consumers' tolerance to ambiguity (Banks & De Pelsmacker, 2014).
Relatively to the level of consumer involvement, Banks and De Pelsmacker (2014) find that the probability markers affect the advertising effectiveness in low involvement services, but not in high involvement services. Thus, in the context of probability markers the Elaboration likelihood model (ELM) also provides a plausible explanation to the lack of consensus relatively to their effects in advertising (e.g., Banks & De Pelsmacker, 2014;Gelbrich et al., 2012;Mart ın-Santana & Beerli-Palacio, 2013). This implies that when the individual has no motivation to understand carefully the arguments present in the message, the usage of probability markers can have stronger effects on consumers' attitude and purchase intentions, as it brings more assertiveness to perceived takeout.
In the current research, it is proposed that probability markers are combined with nostalgia-themed ads in low involvement product contexts. The rational for this proposition is supported on the understanding that the assertiveness evoked by probability markers may help nostalgia appeals on accommodating consumers' sense of insecurity. In psychology, nostalgia integrates individuals mechanism of defense helping on facing uncertainties, vulnerabilities and fear (Baker & Kennedy, 1994;Merchant et al., 2013;Zhou et al., 2013). Thus, as individuals' memories evoke a sense of security, the use of probability markers is expected to reinforce that.
In addition, synergies are expected to result from the common characteristics between nostalgia-themed advertising and probability markers concerning the mechanisms of peripheral cognitive processing. Thus, as both creative resources are processed at the same level, consumers may associate nostalgic messages with probability markers while doing their inferences (Petty & Cacioppo, 1986). This understanding supports the suggestion of the following hypotheses: H2a: The effects of consumers' exposure to nostalgia-themed advertisement in low involvement product contexts are more likely to produce favorable brand attitudes when the ad has probability markers than when the ad has no probability markers.
H2b: The effects of consumers' exposure to the nostalgia-themed advertisement in low involvement product contexts are more likely to produce higher purchase intention when the ad has probability markers than when the ad has no probability markers.
As in Banks and De Pelsmacker (2014), the type of purchase motivation and personality traits are also associated with the conditional effectiveness of probability markers. Thus, regarding the type of purchase motivation, two types are considered: hedonic and utilitarian. Hedonic motivations are generated by affective experiences with a certain product (e.g., pleasure, entertainment, and fantasy), while utilitarian product purchase motivations are associated with the product total utility perceived (e.g., functions, Batra & Ahtola, 1991;Voss et al., 2003).
In the context of probability markers, the use of hedges is expected as more adequate to hedonic motivations due to their subjectivity. Therefore, the relative assertiveness of hedges is expected to reinforce the message, while the absolute assertiveness of pledges would be expected to evoke lower message/source credibility. Utilitarian motivations, however, are more objective, as they are based on concrete terms. Thus, in these contexts, the use of pledges may legitimate the message and increase its effects, as the absolute assertiveness is established over objective and comparable terms (Banks & De Pelsmacker, 2014).
In what concerns personality traits, Banks and De Pelsmacker (2014) propose that tolerance to ambiguity is considered. Tolerance to ambiguity is a relatively stable personality trait that distinguishes the individuals, particularly in what concerns their preference for stimulus more or less ambiguous (e.g., Furnham & Marks, 2013;Furnham & Ribchester, 1995;Herman et al., 2010;McLain et al., 2015).
In advertising, a more tolerant consumer, not only accepts better the most ambiguous stimulus (e.g., hedges) -less familiar or even less congruent-but he also considers it desirable, interesting and enthusiastic (Furnham & Ribchester, 1995). Meantime, individuals who are less tolerant of ambiguity tend to prefer a clearer language (e.g., pledges), entirely without ambiguities or uncertainties (Furnham & Ribchester, 1995).
While combining both conditions, the use of hedges in ads -which are more linked to pleasure (hedonic purchase motivation)-is an effective way to boost the attitude toward the brand and purchase intentions, especially when the brand target is more tolerant to ambiguity (Banks & De Pelsmacker, 2014). By contrast, the use of pledges in ads -that are more functional (utilitarian purchase motivation)-tends to be effective, particularly when among those consumers less tolerant to ambiguity. The following hypotheses capture these effects: H3a: The effects of consumers' exposure to the nostalgia-themed advertisement of hedonic products are more likely to produce favorable brand attitudes when the ad has hedges than when it has pledges, especially among consumers that are more tolerant to ambiguity.
H3a: The effects of consumers' exposure to the nostalgia-themed advertisement of hedonic products are more likely to produce higher purchase intentions when the ad has hedges than when it has pledges, especially among consumers that are more tolerant to ambiguity.
H4a: The effects of consumers' exposure to the nostalgia-themed advertisement of utilitarian products are more likely to produce more favorable brand attitudes when the ad has pledges than when it has hedges, especially among consumers that are less tolerant to ambiguity.
H4b: The effects of consumers' exposure to the nostalgia-themed advertisement of utilitarian products are more likely to produce higher purchase intentions when the ad has pledges than when it has hedges, especially among consumers that are less tolerant to ambiguity.

Methodology
In view of the research objectives, an experiment with factorial design 2 (types of product: hedonic and utilitarian) Â 2 (creative strategies: nostalgic and non-nostalgic/control condition) Â 3 (language styles: pledge, hedge and no probability marker) was implemented. In total, 12 different conditions (split into 6 groups as all respondents were presented to the two types of products) were prepared according to the following criteria. Two product categories were chosen for representing hedonic and utilitarian product types (beer and toothpaste respectively) in low involvement product categories (Banks & De Pelsmacker, 2014;Kim et al., 2019;Muehling et al., 2004Muehling et al., , 2014Voss et al., 2003;Zaichkowsky, 1985). Both brands selected (Colgate and Heineken) represent an important market share in the geographic area where data collection occurred (Portugal) and comply with the criteria established in the pretest. The two probability markers selected were probably (as a hedge) and certainly (as a pledge) (Banks & De Pelsmacker, 2014). For developing the nostalgic and non-nostalgic ads, guidelines from former studies were adopted with nostalgic ads using dates (e.g. in the '90s), words (e.g., relive, remember, recall) and specific background colors (e.g., warm colors: red, sepia, yellow) as cues that evoke nostalgia (Muehling et al., 2004;. On the other hand, the non-nostalgic ads were a more traditional type of appeal, evoking positive and pleasant references of the present or the future (e.g., year, technologies) . Final stimuli tested are presented in Figure 1 and 2.

Development of the stimuli
In order to develop and validate the 12 advertisements used in the experiment, three sequential pretests were implemented. In the first study (an online survey with 55 valid responses), one of the main objectives was to validate the brands selected for being known (brand awareness) and liked (brand liking). Moreover, the product category was inspected for being low in involvement and complying with the hedonic (beer) and utilitarian (toothpaste) criteria (based on Zaichkowsky, 1994). Remaining questions were intended to evaluate the stimuli for their ability to evoke acceptable ad liking (Madden et al.,1988), pleasantness (Broach et al., 1995), visual appeal (Cox & Cox, 1988), vividness (Petrova & Cialdini, 2005), processing flow (Lee & Aaker, 2004), and nostalgia Pascal et al., 2002).
In the second and third pretests (an online survey with 64 and 50 valid responses respectively), probability markers were also included in the stimuli. For the analysis of results, ANOVAs (IBM-SPSS 25) were implemented and averages were compared in-between groups, with the objective of identifying needs for improvement and balance the various stimuli.

Development of the questionnaire
For the final data collection, an online survey was created. In total, six constructs were measured. The two first ones were used to control for the stimuli, with the credibility of the message (two items from Darley & Smith, 1995) and nostalgic feelings evoked (ten items from Pascal et al., 2002) being measured using endpoints Completely disagree (1) and Completely agree (7). Specific questions to validate the understanding of the stimuli were also introduced. Tolerance to ambiguity (twelve items from Herman et al., 2010) and past brand attachment (four items from Muehling et al., 2014) was measured using endpoints Completely disagree (1) and Completely agree (5). Finally, the two dependent constructs, namely brand attitude (three items from Sengupta & Johar, 2002) and purchase intentions (four items from Dodds et al., 1991) were measured using endpoints Completely disagree (1) and Completely agree (7). The original items were translated from English to Portuguese by two bilingual researchers  and back-translated to English by other two bilingual researchers, so the meaning intended in the original items was assured in their translation.

Sampling procedures
In total 2000 emails were sent to university students (from bachelor to master's degree) from 18-35 years. This age range was pre-established so that the sample profile would be susceptible to having comparable nostalgic thoughts or experiences (Muehling et al., 2004;. The emails were randomly selected from a list of contacts provided by two universities. An URL with the link to the questionnaire was provided in the email. As respondents accessed the link, they were initially exposed to common questions related to their demographics and tolerance to ambiguity. At this stage, consumers were also asked if they knew both brands used in the research. All consumers who knew both brands were then randomly allocated to one of the six groups. Once allocated to the specific group, respondents were exposed to ads from the two brands (Colgate and Heineken). The order of brands was randomized and presented separately, so respondents would first answer to questions regarding the first ad presented (as e.g., Heineken) and only afterwards were presented to the second ad (e.g., Colgate) and asked the respective questions.

Data analysis procedures
Initial statistical analyses were performed for validating the scales regarding their unidimensionality and reliability (Anderson et al., 1987;Churchill, 1979), using IBM-SPSS statics V.25. The constructs were checked for their reliability considering a minimum value of 0.65 for Cronbach's alpha. The value of the Cronbach's alpha if item deleted was also inspected. For validating unidimensionality a six-factor solution was considered with factor loadings being required to be high loaded (above 0.5) in only one component.
The stimuli were inspected using ANOVAs, with differences between nostalgic and non-nostalgic stimuli being expected to differ in the level of nostalgia evoked. The credibility of the message, on the other hand, was also inspected, with results being expected not to differ among stimuli. After constructs and stimuli were inspected, the hypotheses were tested using ANOVAs (Huberty & Morris, 1989).

Results
In total 575 valid responses were obtained (290 respondents were submitted to nostalgia-themed ads and others to non-nostalgia themed), with most respondents in the age group between 18 and 26 years old (81%). The split between groups is shown in Table 1.

Measurement scales
As mentioned in the data analysis procedures, the scales were checked for their reliability and unidimensionality. The results are presented in Table 2, with all original constructs being accepted for their reliability, by means of Cronbach alpha, and unidimensionality, by means of exploratory factor analysis, except for tolerance to ambiguity. After having some items reversed, the original scale of tolerance to ambiguity had obtained no acceptable levels of reliability (Cronbach alpha <0.65) (Churchill, 1969), with items falling out of the unidimensional criteria in the exploratory factor analysis (EFA). Item-to-total correlations were used for considering the items for deletion. Cronbach alfa, EFAs and item-to-total correlations were recomputed each time and one item was deleted (Flynn & Pearcy, 2001). The item deletion was sequentially processed until the criteria for reliability and unidimensionality was achieved.
After the constructs were validated, averaged summations were prepared for each of the brands. Participants were also classified into "high" (values > 3.25) and "low" tolerance to ambiguity, according to the means obtained in the summation scale. The same procedure was adopted for past brand attachment, with participants being classified into "high" (values > 2.29 for Heineken and 3.51 for Colgate) and "low" past brand attachment for each of the brands.

Manipulation check
The stimuli were checked for their evoked nostalgia and credibility using one-way ANOVAs (as above mentioned). As expected, nostalgic stimulus  (M NOST : 3.55) was perceived significantly different than non-nostalgic stimulus (M NOST : 2.31) (F: 14.83; p < .001). Credibility between advertisements with and without probability markers did not differ (F: 1.33; p: .25, ns) between stimuli with pledge (M CRED : 3.52), with hedges (M CRED : 3.69) and no probability marker (M CRED : 3.85).

Hypotheses test
In hypotheses H1a and H1b, it was proposed that the effects of consumers exposure to the nostalgia-themed advertisement in low involvement physical product contexts are likely to produce favorable brand attitudes and positive impact on purchase intentions, particularly among consumers who have a high past brand attachment. For analyzing these hypotheses, the sample submitted to no probability markets was used and respondents exposed to nostalgic and non-nostalgic ads were compared with each other. Both past brand attachment conditions were considered. The rational pursued was that if nostalgic and non-nostalgia themed ads obtained comparable effects on purchase intentions and brand attitudes, hypotheses would be accepted. Analyses were performed with two ways ANOVA after data were inspected for non-significant variances in Levene's Test.
In view of the results, hypotheses H1a and H1b were accepted with nostalgia-themed advertising evoking positive effects on brand attitude and purchase intentions. Moreover, past brand attachment imprints significant differences on the effects, independently of the type of stimuli.
In hypotheses H2a and H2b, it was proposed that nostalgia-themed advertisements in low involvement product contexts are more likely to produce favorable brand attitudes and purchase intentions when the ad has probability markers than when the ad has no probability markers. In order to evaluate the results, only respondents submitted to nostalgic ads were considered. Moreover, the analyses were focused on comparing the effects on brand attitude and purchase intentions between groups exposed to stimuli with a hedge (probably), pledge (certainly), and no probability marker (control). Both "high" and "low" past brand attachment were considered. Data were analyzed using two ways ANOVA after data were inspected for non-significant variances in Levene's Test.
The results revealed no significant differences on brand attitude, F BA ( .51, p: .60, g 2 : 0.004. Thus, hypotheses 2a and 2 b were not accepted, with stimulus with probability markers obtaining a comparable performance to stimulus with no probability markers. In hypotheses H3a and H3b, it is proposed that nostalgia-themed advertisement of a hedonic product is more likely to produce favorable brand attitudes and purchase intentions when the ad has hedges than when it has pledges, especially among consumers that are more tolerant to ambiguity. In order to evaluate the results, respondents exposed to nostalgia-themed advertisement were considered. The analysis regarding Heineken (hedonic product) in past brand attachment, tolerance to ambiguity and probability markers conditions. Data were analyzed using three ways ANOVA, even after data were inspected and significant variances were identified in Levene's Test. ANOVA's superior robustness for multi-factor analyses has justified this decision (Huberty & Morris, 1989).
In hypotheses H4a and H4b, it is proposed that the effects of consumers exposure to the nostalgia-themed advertisement of a utilitarian product are more likely to produce more favorable brand attitudes and purchase intentions when the ad has pledges than when it has hedges, particularly among consumers that are less tolerant to ambiguity. To evaluate the results, respondents exposed to nostalgia-themed advertisement were considered. The analysis considered Colgate (utilitarian product) in both past brand attachment and tolerance to ambiguity conditions. Data were analyzed using three ways ANOVA, even considering the significant variances identified in Levene's Test (Huberty & Morris, 1989).

Discussion
Hypotheses H1a and H1b address the controversy existing around the effects of nostalgia-themed advertising regarding its mixed nature and level of product involvement. For evaluating these hypotheses nostalgia-themed advertising was compared with non-nostalgia themed (control condition) on its effects on brand attitudes and purchase intentions. The findings reveal that in low involvement product contexts consumers react positively to nostalgia-themed advertising, with effects on brand attitudes and purchase intentions were similar between both creative routes. Thus, findings confirm the predominance of positive effects over negative, as such, it supports previous studies which suggest that cues of nostalgia may help consumers on making positive inferences in low involvement products contexts (Chou & Singhal, 2017;Muehling et al., 2004;Muehling & Pascal, 2011).
Moreover, the findings also contribute to raising attention over the relevance of past brand attachment as a moderating construct. This result was acknowledged when nostalgia-themed advertising was compared for its effects on a "high" vs. "low" past brand attachment, with significant differences obtained. Thus, when consumers acknowledge their attachment to the brand in the past, the mechanisms of image transfer are facilitated. This effect is supported by the perceived compatibility between the brand perception and the nostalgic themed appeal, so consumers can more easily associate the nostalgic images when they acknowledge the past of the brand (Keller, 2003;Hoeffler & Keller, 2002;Bloom et al., 2006).
In hypotheses, H2a and H2b, the use of probability markers in nostalgiathemed advertising were tested for their potential effects. For analyzing these hypotheses stimuli with hedges (probably), pledges (certainly) and no probability markers were compared. Results indicate that the presence of probability markers (pledges and hedges) does not generate incremental effects on brand attitudes or purchase intentions compared to advertisements with no probability markers.
In the remaining hypotheses, specific conditions were inspected with the type of product and the level of consumers tolerance to ambiguity being expected to influence the effects of probability markers. Therefore, in hedonic products, advertisements with hedges were expected to have more positive effects than with pledges, especially among consumers with high tolerance to ambiguity (H3a and H3b). On the other hand, in utilitarian products, advertisements with pledges were expected to perform better than hedges, especially among consumers with less tolerance to ambiguity (H4a and H4b). The results show no significant differences between the conditions, with hedges and pledges obtaining comparable effects on brand attitude and purchase intentions among different levels of tolerance to ambiguity and types of product.

Conclusions
Although many studies have argued in favor of nostalgia-themed advertising and its impact on brand performance (e.g., Bambauer-Sachse & Gierl, 2009;Muehling et al., 2014;Muehling & Pascal, 2011;Pascal et al., 2002), the encapsulated sadness of nostalgia has generated some controversial results regarding the associations transferred to brands (Baker & Kennedy, 1994;Holak & Havlena, 1992, 1998. The controversy around nostalgia is also associated with the type of cognitive processing of nostalgic messages, with studies positioning it as a distracting element in advertisements of high involvement products and simultaneously suggesting that further studies evaluate its effectiveness in low involvement product contexts (Chou & Singhal, 2017;Muehling & Pascal, 2011;Sherman & Quester, 2005). The current study focuses on addressing this opportunity, exploring the boundaries of nostalgia as a creative strategy to drive persuasion in advertising of low involvement products.
From the theoretical perspective, the current study extends the literature in advertising as it contributes to clarifying two controversial topics. First, regarding the use of nostalgia-themed advertising the findings help to position it as a creative route of positive impact in low involvement products. Therefore, even if previous studies indicate that nostalgia may carry some negative tonality and generate distraction in the decision making processes, in the context of low involvement products nostalgia generates positive effects on brand attitude and purchase intentions when compared with non-nostalgia themed advertising. These effects are associated to the emotional meanings transferred to the advertiser (Muehling et al., 2004;Muehling & Pascal, 2011), with similar results obtained for products associated to hedonic and utilitarian motivations. This result corroborates the idea that nostalgia may perform well as an alternative creative strategy for advertisers.
Past brand attachment emerges as a determinant aspect, so the effects of nostalgia are increased if consumers perceive that their relations with the brand are routed in past experiences. In that sense, nostalgia appeals seem to be more effective when are complementary to associations already existing, so consumers may find the brand within the nostalgic appeal presented. The relevance of past brand attachment had already been raised in previous studies , but yet not further explored.
The second topic of controversy is regarding the use of probability markers. The initial proposition was that as probability markers are accepted as an effective heuristic cue in peripheric processing of advertising messages (Banks & De Pelsmacker, 2014), their use could be beneficial for the assertiveness of nostalgia-themed advertising (Muehling et al., 2004;Muehling & Pascal, 2011). However, the results do not support this rational, contributing to reinforcing the idea that the use of probability markers is of limited interest as they do not drive incremental effects to nostalgia-themed advertising. This result is consistent across different purchase motivations (hedonic and utilitarian) and individuals' personality traits (tolerance to ambiguity). The effects obtained may be related to the relative position of probability markers compared to nostalgia in the processing of the message, with nostalgia overtaking the presence of probability markers, instead of interacting with it. As nostalgia is considered a deep emotional phycological state important for individuals' own self-esteem and rapport with others (Hepper et al., 2014;Juhl et al., 2010;Sedikides et al., 2008;Wildschut et al., 2006) it may deviate, consumers, attention from the markers introduced in the message.
From the managerial point of view, the results suggest that advertisers can make use of nostalgia-themed advertising as a creative alternative as long as their brand has some sort of built-in ground for exploring nostalgia, in the sense of having a relevant number of consumers who personally relate to the past of the brand. The findings also show that probability markers used as a standalone language style (as in the stimuli), do not add value to nostalgia-themed advertising. In the current research, the markers were intentionally applied as add-ons, which were not completely integrated with the core of the creative idea. From this perspective, the results should not be interpreted by managers as a straightforward recommendation for not using probability markers, but to avoid their use as add-ons not integrated at the core of the creative idea.
Furthermore, the findings above presented need to be interpreted within some limitations. Among them is the choice for exploring nostalgia based on two already existing brands (Heineken and Colgate). This decision has allowed that past brand attachment was analyzed; however, it may have had an impact, with brand familiarity potentially compensating for effects of low tolerance to ambiguity. The choice for the nostalgia appeals used in the stimuli was focused on the personal type of nostalgia, which is usually better evaluated for its effects than other types (e.g., historic nostalgia) (Chou & Singhal, 2017;Muehling et al., 2014;Muehling & Pascal, 2011. Moreover, the study is limited to two types of probability markers and within that to two words (certainly and probably). Therefore, findings should be interpreted within the personal type of nostalgia and specific probability markers.
Future studies could replicate and extend the findings by means of broadening the scope of nostalgic types of appeals, using different markers and exploring advertising executions where probability markers are built-in as an essential part of the creative idea. Moreover, future studies could further explore the boundary conditions within which nostalgia-themed advertising impacts low involvement products (e.g., by type of media) and extend the analyses beyond the impact on brand attitude and purchase intentions (e.g., brand love). Finally, future studies could deepen those findings by means of exploring the mechanism that mediates the effects of nostalgia-themed advertising on brand results.