New Film Website Promotion and Box Office Performance
University of Southern California
Today, numerous companies have a presence on the web, with a host of reasons for seeking one. These include a company’s wish to promote its image, provide product information to its visitors, promote awareness about the company’s current and/or new products, or sell its products directly on the web.
In particular, the trend has been growing for movie studios to dedicate websites to the promotion of their forthcoming films. The creation of a website potentially serves as a source of information about the film (e.g., plot, stars, trailers, etc.). In addition, it may serve to induce awareness of a new film, to enhance the site visitors’ intentions to see a film, and consequently increases the likelihood of ticket purchase upon opening. Moreover, after a film’s release, the website may serve to maintain the awareness and interest of filmgoers during the film’s box office run.
Often, film websites succeed in generating considerable hype in anticipation of a film’s release (e.g., Star Wars, Austin Powers, etc.). These sites can be very elaborate and may provide various attractions, including games, pictures and trailers, to enhance site traffic, attract viewer attention, and peak viewers’ interest in a film.
Given the current increase in the emphasis on web-related promotion for new films, this paper focuses on the following critical question: Can website activity measures be related to the performance of a new film? What is the nature of the relationship between the pattern of website activity, in conjunction with other relevant explanatory variables, and a film’s box office performance over its life cycle?
The challenging questions raised above are of potentially critical importance to movie studios, film distributors, and exhibitors who wish to enhance the planning and marketing process for a new film. However, there have not been any reported studies in the literature that have attempted to evaluate and link website activity to a movie’s box office performance.
Likewise, there has been little commercial emphasis directed to linking website activity to the performance of new films to date. One exception is the Hollywood Stock Exchange website (www.hsx.com), which provides an online game, based on the concept of the stock market exchange, to derive measures of values for new films prior to and following their release. Here, the online game results have been studied as a means for providing information for the evaluation of a film’s potential market performance.
However, there is a growing academic research interest in the study of a film’s performance over the stages of its life cycle. For example, Shugan (1998) found that screenwriters and directors may best be used to predict a new film’s box office revenues during a film’s development stage. In another study, Eliashberg and Sawhney (1994) focused on the evaluation of audience enjoyment of key scenes of a film integrated within a preliminary 'rough cut' during the film’s production stage. In preparation for a film’s launch, Krider and Weinberg (1998) have evaluated competitive aspects in relation to the marketing of new films, while Radas and Shugan (1997) examined seasonality aspects relative to the timing of a film’s introduction.
Various model approaches have been proposed to predict the market performance of films during their box office run. For example, Sochay (1994) proposed a multivariate linear regression approach to predict film performance as a function of explanatory variables within three categories (creative sphere, scheduling and release pattern, and marketing effort). This study found independent variables such as genre, the presence of a 'bankable' star, the timing of the film’s introduction, and the screens allocated to the film, to be statistically significant. However, the overall goodness-of-fit of the proposed models were relatively weak (i.e., R-square and adjusted R-square values in the study ranged from .304 to .380).
Among other related studies, Sawhney and Eliashberg (1996) developed a model to help film exhibitors (e.g., theatre owners such as United Artists Theatres) evaluate a film’s distribution effort (i.e., number of screens to allocate to a film). Sawhney and Eliashberg (1995) also developed a model to predict cumulative box office revenues as a function of awareness and word-of-mouth. Another study, by Zufryden (1996), focused primarily on the level of advertising dollars as well as the distribution (number of screens) to be allocated to a new film in relation to their impact on a film’s anticipated market performance. In a recent study, Eliashberg and Shugan (1997) found movie critics’ reviews to correlate with cumulative rather than early box office revenues. The latter result suggests that critic reviews may be of limited value in forecasting the early performance of new films.
An objective of this study is to propose a model approach to help a manager more effectively plan and track the performance of new films during the box office run of their life cycle. As such our approach relates to previous models that have been developed to assist in the new product planning process. For example, models such as TRACKER (Blattberg and Golanty, 1978), ASSESSOR (Silk and Urban, 1978), NEWS (Pringle et al., 1982), LITMUS (Blackburn and Clancy, 1982), and PROD II (Zufryden, 1985, 1989) provide comprehensive approaches for the evaluation of new and frequently purchased consumer products. Other diffusion models have considered marketing variables but have been applied primarily to durable goods. For example, Kalish and Sen (1986), Jones and Mason (1990, 1991), and Jain et al. (1991) have focused on the impact of distribution strategy on the innovation of the adoption process. In addition, several diffusion models have focused on price and advertising (see Kalish and Sen, 1986, for a review of related studies).
In this study, we seek to develop a parsimonious model approach that can be used for planning purposes, prior to the launch of a film. Furthermore, we attempt to consider the unique pattern of diffusion of new film releases in the market over time. In particular, the life cycle for films differs from that of typical consumer products (see Zufryden, 1996). This is because the film diffusion pattern is typically characterized by a peak in box office receipts at the time of initial film release that is followed by a pattern of exponential decay over time. This is in contrast to the typical 'bell shaped' lifecycle diffusion curve pattern that has been described for durable products in past studies (e.g., Bass, 1969; Mahajan and Wind, 1986).
An important consideration, in the model proposed here, is ease of implementation. This includes providing for tractable model estimation on the basis of available data. In addition, an important feature of the proposed model is its inherent structural flexibility with respect to the consideration of relevant causal variables and future extensions. In this study, we attempt to consider box office performance as a function of website activity as well as other relevant variables, including film distribution (e.g., screens), seasonality, time in release, and film characteristics (e.g., production budget and overall film attractiveness).
A unique aspect of this study is its integration of relevant data from a number of distinct data sources.
Website activity data
The primary source of website activity in this study was from website log file data. The basic raw data consisted of detailed daily visitor activity information for 21 film websites. For each film, in order to match the weekly performance data from other sources, the data was reduced by weekly periods, starting with the inception of the film website on the internet and continuing over the box office run of each film. The measures available from the log to file information included gross number of page requests, DPOs (i.e., total requests during each period from Distinct Points of Origin), and new DPOs (i.e., new requests during each period from Distinct Points of Origin). Thus the data covered a time span fill that included the life cycle of the film’s website as well as that of the film’s box office run.
EDI Nielsen data
Box office performance data was obtained from EDI Nielsen. This database included information about each film’s opening day, genre, MPAA rating, production budget, box office total, as well as weekly box office and screens during each weekend of the box office run period of the film’s life cycle.
Studio system box office data
Additional box office performance data was obtained from The Studio System. This data provided screen allocation figures as well as gross box office performance information over each week of the life cycle, for each of the 21 films in our website database. In addition, the gross box office of each film was divided into weekend and non weekend performance components during each weekly period of each film’s life cycle.
In order to measure the quality and potential audience appeal of the films in this study, we obtained data from Cinemascore. This data was based on audience-exit interviews for new films that were administered during their first week of screening. The data included the letter 'grade' given by the audience in evaluating the film, the percent of audience that 'Could not Wait to See' a film, audience-demographic information, as well as potential reasons that attracted the audience to the film (e.g., because of genre, director, subject matter, etc.).
We used a film’s TVGen score to reflect film critics’ evaluations and the quality of a film. These scores were obtained from the TV Guide website (see www.tvguide. com) and were based on a 5-point scale, with 5 reflecting the highest level of a film’s attractiveness.
Star Power Index data
In order to measure the potential impact of the stars in various films, we utilised data obtained from the 1998 Star Power Study in the Hollywood Reporter. This raw data provided information on some 400 film actors and their related Star Power Index. This index was a value, scaled from 0 to 100, reflecting each actor’s attractiveness to movie going audiences. These raw values were used to construct index values for each film i in our study by summing up the corresponding Star Power index values of the major stars featured in each film.
By merging the above datasets, an integrated database was available for our study that covered not only information concerning the box office performance of each film in our study but also of website activity and various other potential causal variables. Thus, upon merging our various datasets, we obtained complete data information on 19 films over the major portion of their box office run life cycles. In all, 264 weekly observations were available for our study that covered the films of interest, as well as the variables that were analyzed in our study.
EVALUATION OF BOX OFFICE PERFORMANCE
We now consider the development of a model framework for forecasting a new film’s box office performance during its box office run life cycle. Because of the initial promotion dedicated to a new film and its typically extensive initial distribution, the first week usually results in a peak in terms of box office sales. Subsequently, the sales pattern will typically decline. Hence, most films’ lifecycle curves are characterized by an initial spike and a subsequent exponential decay over successive weeks (e.g., see Zufryden, 1996). In Figure 1, we illustrate typical lifecycle curve patterns for selected films in our study. Only a few films deviate from this pattern. For example, 'sleepers,' such as The Blair Witch Project and There’s Sornething About Mary, may build interest some weeks after initial introduction, as a result of extremely positive word of mouth. For such films, subsequent concurrent increases in promotion to support the film may lead to lifecycle curves that more closely resemble the more typical bell shaped diffusion curve that we generally find for other products (e.g., see Bass, 1969).
In our study, we found that website activity has a pattern similar to that of box office. Thus, activity generally increases prior to the release of a film. It tends to peak around the first week introduction of a film and decays after that period. As an example, Figure 2 illustrates typical lifecycle curve patterns for film website activity measured by tracking the number of new DPOs (new Distinct Points of Origin) over time.
Log-linear model to predict film box office performance
Zufryden (1996) showed that a log-linear regression model form was able to characterize the box office performance patterns of new films very well. The model may be shown to conform well to the typical film lifecycle curve pattern and, moreover, is easy to estimate by means of standard regression analysis methods. In this study, we extend this model form by considering additional variables, including a measure of past website activity.
The following model form, suggested by previous research and our empirical analysis, was used to explain BOXi,t, a film i’s box office performance (in terms of dollar ticket sales) over weekly time periods, t:
1n (BOXi,t) = ln (a0) + a1ln(NewDPOi,t-1)
+ a2ln(Screensi,t) + a3ln(ti)
+ a4ln(Gradei) + a5ln(Seasont)
+ a6ln(ProdBudgeti) + eit
In our study, the above model was estimated based upon the consideration of the following variables.
An essential marketing variable that affects a film i’s box office performance is the level of Screens (i.e., theatres) allocated to the film over time t. This variable has been found to be a significant predictor of box office performance in prior studies (e.g., Sochay, 1994; Zufryden, 1996).
Aside from Screens we were unable to use any additional marketing variables in our model. For example, relevant advertising data was unavailable for the majority of the films in our sample. Therefore, it was not possible to examine the impact of advertising expenditures on box office performance over a film’s life cycle. However, it should be noted that Zufryden (1996) found advertising effort to have a statistically significant impact on a film’s box office performance and life cycle. Thus, it is expected that the incorporation of advertising effects in our model would have further enhanced our study results.
We defined a film’s website activity by obtaining measures of new Distinct Points of Origin for each film i over time (NewDPOi,t). Here, we expected that website activity would influence a visitor’s intention to see a film. Hence, in order to establish the appropriate causality pattern, we used the lagged value of the new Distinct Points of Origin (i.e., NewDPOi,t-1) as an independent variable in our model.
We hypothesized that the more interest generated from new visitors over time for a film’s website, the more awareness of the film, and consequently the greater a visitor’s intention to see the film. Based on the notion that a film’s website serves as a potential promotional vehicle, we used the NewDPOi,t behavioral variable to reflect the awareness level and intention to see a film over time. In an attempt to validate this measure, we correlated this value with a measure from Cinemascore ('Can’t Wait to See Film'). Here, we obtained a significant positive correlation between these variables. This supported the notion that the web related variable NewDPOi,t is a potentially valid measure of awareness of a film as well as of the intention to see a film. In our study we concentrated on NewDPOi,t as the most appropriate variable to reflect website activity. Thus, we did not consider page requests over time in our analysis. Because page requests were not distinguishable across visitors, this did not represent a good measure of website activity of individual visitors. Indeed, page requests were not found to be a significant independent variable when considered in lieu of NewDPOi,t in our model.
Time since introduction (ti)
In view of the empirically observable typical decay in box office performance that generally occurs after the first week of a film’s introduction, we considered ti the time (in weeks) since release of each film i as an independent variable.
This variable was used as in Zufryden (1996) to explain the generally exponentially decaying pattern that is experienced over the diffusion of a film. That is, decay in a film’s performance typically results as the initial marketing effort (e.g., promotion and distribution) for the film tapers off, audience interest in the movie dwindles, and market penetration, although it continues to increase, does so at a decreasing rate.
In view of previous studies (e.g., Sochay, 1994; Radas and Shugan, 1998), we expected seasonality factors to influence a film’s box office performance. Hence, we defined a seasonal index variable (Season) corresponding to each weekly period t in the year to reflect the effect of the seasonality of film box office performance.
It is well known in the film industry that considerable seasonality exists in the pattern of box office performance during the year. For example, holiday periods are generally characterized by greater box office sales given the enhanced availability of moviegoers during these periods. In cognizance of this fact, movie studios often time the introduction of their potentially more attractive films to correspond with the higher demand holiday and vacation periods. Indeed, several studies have shown the significance of the release pattern of a film (relative to holiday periods) and seasonality in relation to box office performance (e.g., Sochay, 1994; Radas and Shugan, 1998). In Figure 3, we show the pattern of the Seasonality Index over time that was derived in this study by applying the classical time-series decomposition approach (see Makradakis et al., 1983) to the total weekly box office time series. This series was obtained by decomposing the summed box office figures for the total set of films available in our EDI Nielsen database over time into component influences that included the Seasonality Index. As shown in Figure 3, based on the latter analysis, it is observed that seasonal peaks in the Seasonality Index time series correspond to particular holiday periods.
It was expected that a film’s production budget would tend to reflect the film’s quality and, consequently, would help to explain box office performance. This premise is based on prior empirical evidence (Litman, 1983) where it has been shown that production budget is a significant predictor of movie attendance. Thus, we defined Production Budget for each film i as an explanatory variable to reflect the characteristics of a film.
It should be noted that we also examined other film characteristics with respect to their potential impact on box office performance. Thus, although such characteristics as MMPA rating and genre were available for the films in our sample, they were not used in our study. This is because we found that our relatively small sample of films, for which complete data was available, did not yield sufficient variation on the latter variables to permit the estimation of reliable corresponding parameter estimates. However, it should be noted that several previous studies (e.g., Litman, 1983; Austin, 1984; Zufryden, 1996) have found film genre to be significant predictors of film attendance.
In addition, we also examined the potential impact of the stars in a film by developing a composite Star Power Index value for each film. This index was based on the Star Power Index values for the major stars in each film that were available from 1998 Hollywood Reporter data. Here, we did not find the Star Power Index to be a statistically significant independent variable, and thus it was not included in the final regression model. Our results here appear consistent with results of a recent study by Reddy et al. (1998), which suggests that characteristics of the key talent in theatrical shows do not have a consistently significant influence on the show’s success. However, it should be noted that star characteristics have been found to explain film attendance and revenues in other studies (e.g., see Kindem, 1992; Linton and Petrovich, 1998).
In an effort to obtain a measure of the overall attractiveness of a film, we examined the 'Grade' measure (F to A) from Cinemascore audience exit interviews. The letter grades were then translated into numerically scaled scores and used to define film Gradei, for each film i, as a potential explanatory variable to reflect the relative attractiveness of a film to its prospective audience.
Films are often pre-tested or previewed prior to release. At that time, audience reactions can be measured as an indication of the potential success of the film. In addition, last minute production changes (e.g., changes in ending) or changes in thematic approaches for promoting the movie may also be made in response to the results of the screening test. Hence, the inclusion of this variable was designed to provide insights on the potential usefulness of film screening results for predicting a film’s box office performance.
We also examined TVGen critic ratings (obtained from the TV Guide website at www.toguide.com) as another explanatory variable that might suggest the quality of a film and consequently film attendance. However, we did not find this measure to be statistically significant in our study. Thus, it was not included in the final regression model.
The literature has shown mixed results in relation to the influence of critics on movie attendance. For example, while Cooper-Martin (1992) suggest that critics do not play a significant role in film choice, Litman and Kohn (1989) suggest that critics’ ratings are key predictors of a film’s success.
It was anticipated that competitive films and related marketing activities would have a potentially significant impact on the box office performance of new films. In order to study competitive effects within the constraints of our available data, we examined the impact of the number of new films introduced concurrently with each film i in our sample over time. Here, we expected that the number of competing films, introduced concurrently with each film i’s release, would adversely affect the latter’s box office performance. However, in our analysis, we did not find the number of competing films during each period t to be statistically significant. Consequently, we did not include this independent variable in our final model.
Summary of study results
The empirical results obtained in the estimation of the final model from our data are summarized in Table 1.
TABLE 1: RESULTS OF LOG LINEAR REGRESSION OF WEEKLY BOX OFFICE*
* Dependent variable is 1n(BOXi,t)
Here, six independent variables were determined to be statistically significant. These included screens, time from film release, film grade, lagged value of new Distinct Points of Origin, production budget, and seasonal index. For these independent variables, all of the corresponding coefficients were found to be statistically significant at the p £ .01 level. More specifically, the Seasonal Index variable was significant at the p = .01 level, while the other five independent variables were found to be significant at the p = .001 level. Moreover, we found that these six independent variables explained about 90 percent of the variation in film box office performance in our log-linear multivariate regression model. Figure 4 illustrates the overall goodness-of-fit of our model, for selected films, by contrasting the actual and model-predicted box office performance patterns over time.
From a practical standpoint, it should be noted that the proposed model can be easily implemented in practice. All of the explanatory variables could be estimated prior to the introduction of a film. Consequently, the above framework provides a potentially useful means of predicting box office performance prior to the introduction of a film and therefore for planning a film’s release. Of particular importance in our study is the fact that website activity was found to be a significant predictor of box office performance. In particular, the lagged NewDPOi,t-1 values over time showed that a knowledge of short-term previous website activity can be used to enhance the prediction of box office performance. In fact, it can be shown that the independent variable NewDPOi,t-1 alone explains about 14 percent of the variation in box office performance.
The use of website promotion for new films is now commonplace. However, there has been no information to date about the degree to which website activity relates to the box office performance of a film. To this end, this paper described a parsimonious model approach that relates website activity, and other relevant explanatory variables, to a film’s box office a performance. Our results suggested that website activity is indeed a statistically significant variable in relation to a film’s box office performance. A potential reason for this may be the fact that website activity is instrumental in creating awareness for a film as well as promoting a site visitor’s intention to see a film. Thus, our study suggests that the monitoring of website activity may provide a useful measure to evaluate and predict the performance of new films. As such, it is expected that the proposed model may allow studio executives as well as film exhibitors to enhance their planning and tracking of new film releases.
Despite the encouraging results from our empirical study, a number of limitations inherent in the model should be kept in mind. As previously described, because of limitations in our data, it was not possible to specifically study advertising effects. In addition, data limitations prevented a more thorough look at competitive effects in view of the absence of competitive advertising and promotion information in our data.
A number of films engage in promotional activities designed to complement their advertising media efforts. For example, certain films have promotional liaisons, or tie-in arrangements, with promotional partners such as quick service restaurants (e.g., McDonalds’ and Taco Bell franchises). The promotions carried out by the promotional partners (e.g., providing gifts or premiums that relate to the theme of the film) at the point of purchase may have a significant impact on audience awareness and consequently on the film’s box office ticket sales. However, we could not specifically study these promotional effects, as our data did not include information regarding the related promotional expenditures.
Word-of-mouth effects may have a significant impact on the diffusion of films (Arndt, 1967; Bayus et al., 1986). For example, Zufryden (1996) has examined word-of-mouth effects as a behavioral component of a more complex film box office performance prediction model. However, in the present study, because of the absence of longitudinal awareness data for the films in our study sample, we were unable to consider this phenomenon specifically. In future research, it would be of interest to contrast website activity measures with standard measures such as awareness and intention to see a film. These standard measures are currently tracked commercially in the film industry (e.g., the National Research Group, Inc. collects such data from tracking survey studies for new films). Thus, information about website activity might provide a useful and cost-effective surrogate measure of these customary tracking measures.
Another study limitation was the relatively small sample of films for which complete data was available. This preempted a thorough study of the impact of movie characteristics such as film genre and MPAA rating within a multivariate context. This is because these variables did not exhibit enough variation to permit the reliable estimation of corresponding parameters.
Despite its limitations, the proposed model provides a parsimonious first step in considering the use of information from a relatively new medium (the internet) for evaluating and predicting the performance of new films. Because of the relatively short life cycle of films and the ready availability of appropriate data, the film industry is amenable to the type of analysis that was conducted in this study. Moreover, we believe that the proposed study approach is not restricted to the study of films alone and may be generalized to other product areas as well. With the growth of the internet as an information source and advertising medium, E-commerce, and E-tailing, it is anticipated that the internet will be used increasingly for product testing and product information dissemination, as well as for conducting online transactions. As in the case of the movie industry, which has been the focus of attention here, it is likely that the measurement of visitor activity on the internet will provide a useful source of data to evaluate and predict the success of new products, in general. This would be highly desirable in view of the high risks involved in the development and marketing of new products (e.g., see Zufryden, 1996). Thus, it is hoped that this study will not only provide impetus for extended applications to films but also to the myriad of other products and services that are featured on the internet.
The author would like to acknowledge Paul Grand, chairman of Word of Net Inc., for providing the film website log file data used in this study. In addition, the author acknowledges the contributions of Robert Wonderlick of LECG, Inc., in providing the EDI Nielsen, Cinemascore and Star Power data for the purposes of this study. Thanks also go to Stephen Lankton of Columbia Tri-star for providing the Studio System box office data used in this study.
ARNDT, JOHAN. 'Role of Product-Related Conversations in the Diffusion of a New Product.' Journal of Marketing Research 4, 3 (1967): 291-95.
AUSTIN, BRUCE A. 'Portrait of an Art Film Audience.' Journal of Communication 34 (1984): 74-87.
BASS, FRANK M. 'A New Product Model for Consumer Durables.' Management Science 15, 1 (1969): 215-27.
BAYUS, BARRY L. 'Word of Mouth: The Indirect Effect of Marketing Efforts.' Journal of Advertising Research 25, 3 (1985): 37-51.
BLACKBURN, JOSEPH D., and KEVIN J. CLANCY. 'Litmus: A New Product Planning Model.' Marketing Planning Models, Studies in the Management Sciences, A. Zoltners, ed. New York: New Holland Publishing, 43-61.
BLATTBERG, ROBERT, and JOHN L. GOLANTY. 'Tracker: An Early Test Market Forecasting and Diagnostic Model for Now Product Planning.' Journal of Marketing Research 15, 2 (1978): 192-202.
COOPER-MARTIN, ELIZABETH. 'Consumers and Movies: Information Sources for Experiential Products.' In Advances in Consumer Research, 19, John F. Sherry Jr. and Brian Sternthal, eds. Provo, UT: Association for Consumer Research, 756-61.
ELIASHBERC, JEHOSHUA, and STEVEN M. SHUCAN. 'Film Critics: Influences or Predictors?' Journal of Marketing 61, 2 (1997): 68-78.
______,and MOHANBIR S. SAWHNEY. 'Modeling Goes to Hollywood: Predicting Individual Differences in Movie Enjoyment.' Management Science 40 (1994): 1151-73.
JAIN, D. C., VIJAY M. MAHAJAN, and EITAN MULLER. 'Innovation Diffusion in the Presence of Supply Restrictions.' Marketing Science 10, 1 (1991): 83-90.
JONES, MORGAN J., and CHARLOTTE H. MASON. 'Incorporating Distribution in New Product Diffusion Models.' International Journal of Research in Marketing 8 (1991): 91-112.
_____,and _____. 'The Role of Distribution in the Diffusion of New Durable Consumer Products.' Report no. 90-110, Marketing Science Institute, Cambridge, MA, 1990.
KALISH, SHLOMO, and SUBRATA K. SEN. 'Diffusion Models and the Marketing Mix for Single Products.' In Innovation Diffusion Models of New Product Acceptance, V. Mahajan and Y. Wind, eds. Cambridge, MA: BallingerPublishing Company, 1986.
KINDEM, GORHAM. 'Hollywood’s Movie Star System: A Historical Overview.' In the American Movie Industry: The Business of Motion Pictures, Gorham Kindem, ed. Carbondale, IL: Southern Illinois University Press, 1982.
KRIDER, ROBERT F., and CHARLES B. WEINBERG. 'Competitive Dynamics and the Introduction of New Products: The Motion Picture Timing Game.' Journal of Marketing Research 35, 1 (1998): 1-15.
LINTON, JAMES M., and JOSEPH A. PETROVICH. 'The Application of the Consumer Information Acquisition Approach to Movie Selection: An Exploratory Study.' In Current Research in Film: Audiences, Economics and Law, 4, Bruce A. Austin, ed. Norwood, NJ: Ablex Publishing Co., 1988.
LITMAN, BARRY R. 'Predicting Success of Theatrical Movies: An Empirical Study.' Journal of Popular Culture 16 (1983): 159-75.
______, and LINDA S. KOHL. 'Predicting Financial Success of Motion Pictures: The ‘80s Experience.' The Journal of Media Economics 2 (1989): 35-50.
MAHAJAN, VIJAY, M., and YORAM WIND. 'Innovation Diffusion Models of New Product Acceptance: A Reexamination.' In Innovation Diffusion Models of New Product Acceptance, V. Mahajan and Y. Wind, eds. Cambridge, MA: Ballinger Publishing Company, 1986.
MAKRIDAKIS, SPIROS, STEVEN C. WHEELWRIGHT, and VICTOR E. MCGEE. Forecasting Methods and Applications. New York: John Wiley and Sons, Inc., 1983.
PRINGLE, LEWIS G., R. DALE WILSON, and EDWARD I. BRODY. 'News: A Decision-Oriented Model for New Product Analysis and Forecasting.' Marketing Science 1, 4 (1982): 1-29.
RADAS, SONJA, and STEVEN M. SHUGAN. 'Seasonal Marketing and Timing Introductions.' Working paper, University of Florida, Gainesville, 1997.
REDDY, SRINIVAS K., VANITHA SWAMINATHAN, and CAROL M. MOTLEY. 'Exploring the Determinants of Broadway Show Success.' Journal of Marketing Research 25, 3 (1998): 370-83.
SAWHNEY, MOHANBIR S., and JEHOSHUA ELIASHBERG. 'A Parsimonious Model for Forecasting Gross Box Office Revenues of Motion Pictures.' Marketing Science 15, 2 (1996): 113-31.
_____, and _____. 'MOVIEMOD: A Prelaunch Forecasting and Diagnosis System Applied to Motion Pictures.' Working paper, Northwestern University, 1995.
SHUGAN, STEVEN. 'Forecasting Failure and Success of New Films.' Working paper, University of Florida, Gainesville, 1998.
SILK, ALVIN J., and GLEN L. URBAN. 'Pre-test Market Evaluation of New Packaged Goods: A Model and Measurement Methodology.' Journal of Marketing Research 15, 2 (1978): 171-91.
SOCHAY, SCOTT. 'Predicting the Performance of Motion Pictures.' The Journal of Media Economics 7, 4 (1994): 1-20.
ZUFRYDEN, FRED S. 'Linking Advertising to Box Office Performance of New Film Releases-A Marketing Planning Model.' Journal of Advertising Research 36, 4 (1996): 29-41.
_____ . 'How Much Should Be Spent for Advertising A Brand?' Journal of Advertising Research 29, 2 (1989): 24-34.
_____. 'Prod II: A Model for Predicting from Tracking Studies.' Journal of Advertising Research 25, 2 (1985): 45-51.
NOTES & EXHIBITS
Fred S. Zufryden is the Ernest W. Hahn Professor of Marketing at the University of Southern California. He received a Ph.D. in business administration from the University of California at Los Angeles. Dr. Zufryden�s numerous publications have appeared in the Journal of Advertising Research, Management Science, the Journal of the Operational Research Society, the Journal of the Royal Statistical Society, the Journal of Marketing, the Journal of Marketing Research, Marketing Science, and other journals. He has served as consultant to various market research and consumer products firms.