Discussion of the Findings

information unavailable to outsiders. A receiver evaluates the usefulness of the signal based on the perceived quality and credibility of the Signaler (Connelly et al., 2010), and some have argued that reputation can drive the perceived signal quality (Deephouse, 2000; Coff, 2002).

While previous papers only focused on the relationship between expert reviews and the performance of entertainment goods (e.g., Hirsch, 1982; Shrum, 1991; Hirschman & Pieros, 1995), this paper delved deeper into determinants affecting this interaction. In more detail, this paper investigated whether the reputation of a newspaper outlet moderates the effect of the sentiment of expert reviews published in that newspaper on the performance of a Broadway show. Additionally, it was tested whether the show's content impacts this moderation effect.

Before testing the paper's hypotheses, the general relationship between the expert reviews and performance. The test showed a positive and significant relationship between the sentiment of expert reviews and the performance of theatrical productions - either in the form of the number of performances or net grosses. These results contradict Hirschman and Pieros (1995), who found no apparent correlation between expert reviews of Broadway shows and their performance. Their study showed significant limitations as their data contained only ten Broadway plays. The overall positive correlation between the sentiment of expert reviews and the performance of a show in this paper aligns with the findings of both Reddy et al. (1998) and Simonoff & Ma (2003). Nevertheless, these two works found contradicting results regarding the influence of individual newspaper outlets. While Reddy et al. (1998) discovered The New York Times to yield a more substantial impact than the New York Daily News, Simonoff & Ma (2003) showed the opposite. This research tested the effect of critics' reviews on the number of performances and the net grosses of a show. The findings showed that the New York Daily News

exceeded the impact in both cases and supported the results of Simonoff & Ma (2003). Most newspapers have specific theater critics that review the prevalence of productions. Joe Dziemianowicz, for instance, wrote the great majority, with 88% of the reviews published in the New York Daily News. Compared to this, The New York Times reviews are split between Ben

Brantley (58%) and Charles Isherwood (31%). Readers may associate more credibility and trustworthiness with a critic publishing a higher number of reviews. Linking this to Signaling Theory, it could be argued that with higher credibility of the signaler (critic), the signal itself (review) becomes more relevant and influential to the receiver (reader). However, it must be remarked that the effect of reviews published in the Variety yielded an even more significant impact than the New York Daily News on the number of performances of a show. The critic Marilyn Stasio is responsible for about 61% of the reviews published in this newspaper, which weakens the previously argued explanation. Nevertheless, the fact that the effect of the sentiment of expert reviews differs in strength depending on the newspaper outlet acts as a basis for the hypotheses of this paper.

The first hypothesis (H1) predicted a positive influence of a newspaper outlet's reputation on an expert review's effect on the success of a theatrical show. As for the means of this paper, success was determined by performance - both in terms of financial grosses and the number of performances. This hypothesis was supported based on the significant results of the indirect effect as part of the PROCESS Macro Model 7. While the indirect effect is minor, it increased with a higher reputation, emphasizing that a higher reputation is associated with a more significant impact of the sentiment on the performance, providing sufficient evidence supporting H1.

The influence of a newspaper's reputation regarding the relationship between the sentiment of an expert review and the performance of a theatrical production has never been tested before. However, Kramer (2015) and Verboord (2009) studied the influence of expert reviews on legitimacy and institutionalization, respectively. According to Conelly et al. (2010), receivers perceive signals regarding reliability (also called credibility). Drawing a parallel between these works, it could be argued that a higher level of legitimacy and institutionalization is associated with higher signal reliability. Consumers seek information to fill a knowledge gap, which can help minimize the risk of a failed purchase. This process is especially crucial for experience goods, such as Broadway shows. Therefore, the higher the signal reliability, the more strength the signal yields and, as a result, the more significant influence this signal has on the receiver.

The second hypothesis (H2) argued that the reputation of a newspaper outlet acts as a moderator on the influence a review published by the given outlet has on the performance of a Broadway show. While the analysis provided supporting evidence for H1, it failed to yield significant statistical evidence (p > .050) and, hence, validate the proposed moderation effect.

However, the inclusion of Duration and Previews as covariates led to a moderately significant (p

= .052) but minor moderating effect. While the results show only a moderate significance, they provide supporting evidence in favor of H2.

While many scholars have studied different determinants influencing entertainment goods' success, only a few have focused on combining those factors, such as Desai & Basuroy (2005), who tested for moderating effects of the genre, star quality, and reviews on performance.

This paper only focused on a combination of two aspects, the sentiment of a review and the

reputation of a newspaper outlet. Especially in the entertainment industry, consumers face a significant pre-purchase risk. As a result, the information collection process gains importance, and consumers may look for various types of information. The neglection of additional contributors, such as the star power of the creative team and the cast, the genre, or the pre-opening buzz, may explain the low significance of the results.

Hypothesis three (H3) suggested the basis of a show's plot to influence the predicted moderating effect of a newspaper outlet's reputation, as proposed in H2. While Model 11 provided mostly insignificant results, the findings on the indirect effect (SENT à PERF à GROSS), as shown in Table 10 (p. 54), are noteworthy. The results reveal a positive correlation between the influence of the sentiment of an expert review on a show's performance show and the reputation of a newspaper outlet. However, the analysis only yielded significant evidence of this correlation for shows with an original plot. As for musicals based on an already existing story, theme, or music, the findings were insignificant. Hence, these results only partly support the hypothesis that the moderating effect of reputation carries more importance for musicals based on original content.

The musical Rodgers + Hammerstein's Cinderella is based on the well-known folk tale, which has been the basis for countless movies, books, and more. Similarly, the musical Dames at Sea is loosely based on the film Gold Diggers of 1933 (Teachout, 2015). However, in addition to

only having a loose connection to the movie, the movie itself certainly enjoys less familiarity than others. Moreover, as the musical's creators decided on a different title, audiences may be hindered from forming a connection between the two entertainment goods. Therefore, it is likely that the lack of this differentiation leads to the insignificance in results for not original musicals.

While Model 12 showed a strong, marginally significant effect of the sentiment of a newspaper review on the number of performances, they yielded a significant result for the effect of the sentiment on the net grosses conditional to the moderators. Apart from that, all tested moderating effects on the number of performances as the mediator were insignificant. However, the analysis unveiled the show's content to strongly moderate the effect of an expert review's sentiment on the net grosses (B = 1.102, bse = .487, t = 2.263, p = .024). Moreover, the test discovered that the show's content moderately moderates the influence of the reputation of a newspaper outlet on a production's net grosses. Lastly, the analysis discovered only a slightly positive yet insignificant interaction effect between sentiment and reputation on the net grosses.

However, the three-way interaction between sentiment, reputation, and content negatively influences the net grosses. From this, it can be inferred that the interaction between sentiment and reputation changes depending on the content of a show. It must be noted that with the inclusion of content as the higher-order moderator, the interaction effect of sentiment and reputation on net grosses shifted both in significance and direction. As the test resulted in a negative interaction coefficient between the three variables, combining the three predictor variables has less impact than the total sum of the individual effects. It yielded a significantly strong moderating effect of a show's content on a review's sentiment. Considering all this, the analysis failed to uncover sufficient supporting evidence of H3, which is rejected.

As previously shown, the positive moderation effect of reputation on the sentiment of newspaper reviews was supported only for original musicals. Moreover, the findings of this research shed light on the fact that the sentiment of expert reviews is positively correlated with the originality of a show. This statement is supported by the fact that musicals of original content

received statistically better reviews by critics, which, in turn, positively affects both the longevity of a show as well as its net grosses.

The lack of significant results for non-original shows and the rejection of H3 may be explained by the absence of different levels of familiarity with the content a musical is based on, which has already been discussed previously. Musical productions based on highly unfamiliar content may experience an effect of reputation and sentiment more similarly to entirely original shows rather than musicals based on a vastly well-known story.

Lastly, the fourth hypothesis (H4) expected the moderation effect suggested in H2 to diminish if a production is a revival compared to the new show. The test results revealed a mild significant impact of the interaction between Sentiment, Reputation, and Production Type with the number of performances. Additionally, the number of performances is strongly correlated with the net grosses. Therefore, it can be reasoned that the production type, either a new production or a revival, influences the moderation effect of the reputation of newspaper outlets on the sentiment of reviews and provides sufficient support in favor of H4.

As previously described, many consumers seek expert reviews to gain information about a product to minimize the risk involved in the product's purchase, which stands especially essential if a consumer has no prior knowledge. A production is referred to as a revival if its history includes a previous run on Broadway. As a result, customers had the opportunity to acquaint themselves with the show, and the demand for additional information may become unneeded. This explains the statistical evidence that the influence of the reputation of newspaper outlets and the sentiment of reviews on the performance of a show diminishes for a Broadway revival.