Viewing Race in the Comfort Zone

Carter suggests the concept of a “comfort zone” to explain the inability of dramatic African American programs to be successful on television. He argues that a workable formula has been developed for successful African American series, “portray black people in a way that would be acceptable to the millions of potential purchasers (whites) of advertised products. That is, non-threatening and willing to ‘stay in their place.’”. Using a data set constructed from television ratings and shares, this study examines “black-centeredness” within the context of program success and failure. The comfort zone concept argues Black-centered television series are only successful in a comedic genre because White audiences, who have the majority of the ratings power, will only watch Black-centered series with which they are comfortable. The findings suggest that, in general, race, that is Black-centeredness, did not negatively influence program ratings or shares.

article, TV's Black Comfort Zone for Whites, published in 1988, Richard Carter introduced the concept of "comfort zone" to account for the low audience ratings and consequent lack of success of BC dramatic series. He argued that BC television series are only successful in a comedic genre because White audiences, who have the majority of the ratings power, will only watch BC series with which they are comfortable (Carter, 1988). Whether due to bigotry or ignorance-sustained prejudice, White audiences are uncomfortable with serious Black issues in real life, and they are unwilling to learn more about those issues even in the vicarious distance of representations that appear as television entertainment (Carter, 1988). Taking Carter's (1988) supposition of a comfort zone within the context of media choice theories, I argue that viewing is an active process through which audience members decide to accept or reject media content as they perceive its compatibility with their personal preferences for not merely an acceptable world but a comfortable world.

Prior Studies
No prior study known to this author has undertaken the empirical investigation reported in this article. However, various studies have examined conditions and aspects relevant to the present inquiry. For example, several reports have confirmed the obvious-that television network executives, afraid of offending segments of their audience, avoided BC programming (MacDonald, 1992). The exceptions-shows such as Beulah and Amos 'n' Andy, both of the 1950s-were relatively timid situational comedies, which, while featuring Black actors, could as well have been comedies with Black actors in White faces (to use Carter's, 2007, more recent depiction). By the late 1960s, with the country moving toward assimilation, African American actors reappeared in series such as I Spy and Julia, which presented "whitewashed" African American characters. These characters were nearly devoid of references and depiction of issues relevant to African American culture. The next trend in programming for BC series was segregated situation comedies and a revamping of minstrelsy. Series such as Good Times, The Jeffersons, and Sanford and Son focused on Black humor and placed African American characters in a predominantly Black world. In 1984, The Cosby Show broke the mold of these stereotypical representations of African Americans. The Cosby Show presented a view of African American life rarely seen: an upper-middle class, dual-career family with near perfect children, focusing on mainstream values, which appealed to a mass audience. It could be argued that a few other comedy series of this period including Frank's Place, A Different World, and Roc, followed the innovativeness of The Cosby Show by presenting new and non-traditional representations of blackness. Yet, most BC series of this era featured stereotypical characters and plots (e.g., 227 and Amen). The mid-1990s brought a move back toward segregation and "ghettoization" via the concentration of BC series on up-start commercial networks such as United Paramount Network (UPN).
Success in shows is measured primarily by two indices, rating and shares, because both factors tell commercial sponsors how much of an audience their product advertisements are reaching. The two indices differ only by the denominator of a ratio: Rating is the number of households with a television tuned to Show S, in ratio to the total number of households having at least one television set, whereas share is the same numerator in ratio to the total number of households watching television during the time slot of Show S. Thus, the former measure defines its universe as including people who are engaged in an activity other than TV viewing, while the universe of the latter measure is restricted to people who are watching television during the given time slot. The fact that the two measures are highly correlated testifies to the strength of the medium in capturing attention, a fact that advertisers were quick to discern. Either of the measure has been the most influential proximate factor in the success or failure of a series (Webster, Phalen, & Lichty, 2000;Wimmer & Dominick, 1991). Various other factors are significant conditions to that success, however, and the following summary, drawn from results of previous studies, emphasizes three conditions that will be examined in the data analyses below.
The following review of prior studies emphasizes conditions that were extant during the three decades of television shows examined in this article. It was a time of momentous changes in U.S. society, much of it reported in television newscasts and some of it reflected in network decisions about program content. It was also a time of growing interest in the study of television as a window on larger society, and some of that interest resulted in isolation of a number of specific effects relevant to variations in program success-in particular, effects of a show's genre (comedy vs. drama), of the network on which it appeared (ABC, CBS, NBC, and FOX), and of the period during which it debuted.

Genre Effects
The situation comedy is the foundation of prime-time programming. During the years of the present study, the average rating was consistently higher for comedic shows than for other program types (Mintz, 1985). The other primary genre in episodic television programming is the 1-hr drama, of which there have been several sub-genres including police procedurals, detective stories, action-adventure dramas, fantasy, science fiction, family drama, the Western, medical drama, and prime-time soap opera. The sub-genres tend to be cyclical in popularity, initially debuting as a result of innovation, then evolving via imitation, and ultimately declining as a result of saturation. Situation comedies were more popular than dramatic series among network executives and advertisers, as well as (and partly because) more popular among viewing audiences. Situation comedies provided a form of escapism from the tensions of daily life, whereas dramatic shows made more demands on the viewer's imagination, created suspense, and could put viewers in a frame of mind that would counteract sponsors' hopes for their advertising investments. Advertisers agreed that it was easier to sell products when viewers were in a relaxed frame of mind, and comedy prepared viewers for consumption (Kuhns, 1970). For network executives, economics figured in yet another way: Production costs were lower for situation comedies than for dramas (Eastman, Head, & Klein, 1985). Thus, when examining the possibility of a racial factor in program success, it will be important to control difference of genre.

Network Effects
The broadcast histories and programming preferences of networks probably influenced the types of BC programs aired as well as their ratings, shares, and rates of survival. Gitlin (1983) and Atkin (1992) discussed differing programming trends of each network. NBC, the oldest commercial network that began broadcasting in 1926, was considered the most innovative of the major three and was known for introducing forms such as the "desk and sofa talk show" (e.g., Today Show, Tonight Show) as well as the free-form comedy variety format (e.g., Saturday Night Live and Laugh-In), the 1-hr soap opera, and the miniseries (Brown, 1982). CBS began broadcasting 1 year later and was considered the most conservative commercial network, in part, because during its earliest years of existence, CBS was able to secure prominent broadcast signals for its owned stations and for many of its affiliates, which reached further than NBC signals. CBS had more rural affiliates and greater penetration of rural markets than the competition, and it was ratings leader into the mid-1970s (Brown, 1982). ABC began in 1943 when NBC was forced by the Federal Communications Commission to sell its Blue Network. The youngest of the Big 3 commercial networks with fewer affiliates in small markets and more in big cities, ABC developed programming for young urban audiences (e.g., Happy Days, Laverne and Shirley, Charlie's Angels, and The Love Boat). In 1986, the FOX network debuted and initially focused on young urban audiences. This strategy was partly targeted toward a niche audience often ignored by the older networks-young African Americans-with programs such as Martin, Living Single, and New York Undercover. By the mid-1990s, however, as FOX gained greater presence in the general market, its program content moved away from minority-centered shows and toward White-centered programs that appealed to younger audiences (e.g., Melrose Place, X-Files, and Beverly Hills 90210).
In an analysis of series from 1950 to 1991, Atkin (1992) offered two economically based arguments to explain network trends in minority-lead series programming. 1 He suggested that during periods of greater internal competition among the networks, minority-lead programming should increase, as the networks become more willing to try innovative programming to attract audiences. According to Atkin, network competition peaked between 1976 and1979. In addition, he suggested that periods of external competition (mainly from cable television and videocassette recorders) should correlate positively with minority-lead series. Atkin found partial support for both arguments, noting that increases in minority-lead programming peaked during the late 1970s and again after 1985. Furthermore, he found statistically significant differences among the three major networks in the number of minority-lead series broadcast: NBC had 45% of all minority-lead series; ABC, 31%; CBS, 24%. This is consistent with MacDonald's (1992) report that in the early years of television, NBC, in an effort to improve the network's image with African Americans, published guidelines for the equitable portrayal of minorities on television. The network noted that "henceforth all programs treating aspects of race, creed, color and national origin would do so with dignity and objectivity" (MacDonald, 1992, p. 4).
Those reports support the claim that CBS was the most conservative of the commercial networks and NBC the most progressive. Consequently, the network on which a BC program aired might have affected its success. The more innovative networks might have been willing not only to experiment with unconventional programming but also to be more patient while it tried to build its audience.

Period Effects
Another factor that could affect program ratings and shares is historical period-that is, the year a program began. In general terms, period effects are often manifest in the production of a cultural commodity; for, as Peterson (1982) put it, "the nature and content of symbolic products are shaped by the social, legal and economic milieus in which they are produced" (p. 145). An example specific to the present investigation is the effect of the Kerner Commission's citation of the media as having played a role in the "creation of a schism between black and white America" (Stroman, Merritt, & Matabane, 1989, p. 44). 2 Whereas even the most progressive executives of the television industry had been reluctant to place African Americans in prominent roles, for fear of offending large segments of the viewing audience, thus reducing ad revenues, the stance of the Kerner Commission opened a safer space by recommending that "television should develop programming which integrates Negroes into all aspects of televised presentations . . . In addition to newsrelated programming we think that Negroes should appear more frequently in dramatic and comedy series" (Stroman et al., 1989, p. 45). In response, networks developed BC programming that had a better chance of being palatable to White audiences by representing of Black life in comedic genres or in supporting roles. At the time of the Commission's report, network television offered only one Black-featured drama: I Spy (1965)(1966)(1967)(1968), with Bill Cosby as the first African American in a leading role in a dramatic series. In the years following the Commission's recommendations, three BC programs debuted: two situation comedies, Julia (1968)(1969)(1970)(1971) and The Bill Cosby Show (1969Show ( -1971, and a Western, The Outcasts (1968)(1969).
There has also been evidence of a technology-related period effect. While numerous observers had been noting a downward drift in average ratings and shares, Lewine, Eastman, and Adams (1989) found that the threshold level of ratings or shares at which a program would be canceled had declined about 10% during the 1980s. Whereas prior to the 1980s, a rating "below 20 (or an audience share of less than 30) almost always resulted in cancellation on any network," the threshold then dropped to a rating of 17 (or a share of 27). Network decision makers were being more patient with "slow builders"-"programs that acquire a loyal audience only after months of patient nurturing" (e.g., Cheers). This shift coincided with the introduction and increased use of cable television as well as satellite systems, which cut into the major network's share of the viewing audience. 3 In sum, the period during which a series was telecastand more especially, the year of its debut-probably had an impact on its success. Programs that aired during the first two decades of the interval of this study had to survive within an environment of relative high average ratings and shares, but they could do so within a more restricted (by comparison with later years) number of market competitors. That situation changed during the 1980s, for a variety of reasons. Thus, the present analyses must be alert to period effects. This especially includes careful attention to series that were broadcast in the years immediately following the Kerner Commission's report.

Theoretical Guidance and Hypotheses
Media choice theories seek to explain why audience members select specific media content. 4 For purposes of the present study, selection of content is treated within the context of theoretical perspectives that seek to explain media choice as an active rather than passive process. Given the finite options from which to choose (e.g., radio, magazines, televisions, newspapers, books) and the finite options of content available within each of those options, audience members not only consciously accept or reject program content but are also attentive to the content selected. 5 According to the "selective exposure" thesis, audience members are more likely to choose content that is consistent with their prior opinions and beliefs and to avoid content they assume to be, or discover is, incompatible with those beliefs and opinions (Hartman, 2009;Webster et al., 2000;Wright, 1986). Experimental research in political campaigns, for example, has shown that preferences in program content tend to determine who listens or watches which messages (Wright, 1986). During the period of the present study, Zillman and Bryant (1985) found that selection of entertainment programming could provide comfort and/or reduce discomfort via pacifying information. Individuals anxious about crime, for instance, were more likely than others to select program content that made ideals of justice readily apparent and victorious and avoided content that featured unanswered victimization (Severin & Tankard, 1992). Other factors that affect media choice include demographic and cognitive-emotive characteristics of potential and actual audience members, as well as factors of opportunity and availability (Hartman, 2009;McQuail, 1997;Webster et al., 2000;Wright, 1986). The present study focuses on characteristics of available program content and on responses by potential and actual audience members as measured by their collective voice in ratings and shares data. Consequences of choice of content to view are for present purposes of the survival of a given show. Consequences of choice extend far beyond that, of course. In a longitudinal panel study of the differential effects of television exposure, for example, Martins and Harrison (2012) found that, in general, the level of self-esteem was adversely affected among Black children and among White girls. White boys, however, displayed improved selfesteem. This and other consequences are undeniably important. They are beyond the scope of the present investigation.
This study looks behind the "comfort zone" thesis to examine detailed evidence about audience reactions to hundreds of shows that were telecast between 1963 and 1994 on the major commercial networks. To what extent do representations of blackness in fictionalized episodic television programs affect success as measured by ratings and shares of audience? Mean ratings and means shares, calculated for the lifetime of each of the television shows, are compared within a 2 × 2 grid defined by BC versus not and drama versus comedy. Other variables that affected program success will be introduced as controls, to gauge sensitivity of success to the two primary variables, race and genre. Accordingly, and in view of the fact that the large majority of the television audience was not Black (much less BC), my main hypotheses are as follows: Hypothesis 1: BC programs received significantly lower average ratings and shares than programs that were not BC. Hypothesis 2: Dramas received significantly lower average ratings and shares than comedies. Hypothesis 3: BC dramatic programs received significantly lower average ratings and shares than other programs (dramas and comedies not BC as well as BC comedies). Hypothesis 4: Each of the expected differences was robust enough to withstand controls for the effects of network and period.

Method
The data analysis is based on a working sample drawn from a full sample of 410 prime-time television programs-87 BC and 323 non-BC-telecasted on the four major commercial networks between 1963 and 1994 (Brooks & Marsh, 1995). BC programs (hereafter, BC vs. non-BC) are operationally defined as (a) the cast predominately consisted of regularly appearing African American characters (e.g., Good Times, The Cosby Show) or (b) the cast included at least one major regularly appearing African American character and the program sometimes featured plots that focused on African American issues (e.g., I'll Fly Away). To check the validity of my own identifications, I consulted three sources in addition to Brooks and Marsh (1995), MacDonald (1983), Woll and Miller (1987), and Jackson (1982).
The 1963-1964 television season is the starting point because it marks the beginning of network attempts to broadcast BC dramatic series. The 1994-1995 television season is the end point because of the predominance of BC programming on Warner Brothers (WB) and UPN, two new networks that initially featured content designed to appeal to African American audiences. 6 All of the 410 prime-time programs offered fictionalized content, whether comedic or dramatic. News, sports, movies of the week, variety programs, "talk shows," and "specials" are excluded. The comfort zone thesis is most relevant to fictionalized and regularly appearing television characters.
The approximate universe of programs among the four networks between 1963 and 1995 was 1,233, almost evenly split between genres (618 comedies and 615 dramas). 7 Given the large disproportion between BC and non-BC shows, sample selection was split: All BC prime-time television programs, comedy or drama, telecast between 1963 and 1994 were selected; a random sample of all other prime-time comedies and dramas of the same time period on the four major networks was drawn, resulting in a selection of 323 non-BC programs, using Brooks and Marsh (1995) as the sampling frame. This produced the full sample of 410 shows.
The sampled observations consist of the ratings and shares for each program from birth to death, extracted from the Nielsen Television Index (NTI). However, 18 of the programs were right-censored on ratings and shares data (i.e., they remained on air at end of observation) and have been excluded. In addition, ratings and shares data were missing for 36 programs. The Library of Congress holds the most comprehensive, publicly available collection of the NTIs. 8 For purposes of this study, I was able to review the NTI journals from September 1963 through December 1977. However, the Library of Congress collection was missing NTIs from January 1978 through August 1985. In addition, data are missing for some programs (e.g., Shaft and Tenafly) that were part of rotating series. 9 Excluding cases because of right censorship and missing data resulted in a working sample of 351 cases. While sample-selection bias cannot be ruled out conclusively, the comparisons displayed in Table 1 suggest that the bias is minimal. The ratio of BC to non-BC shows are nearly identical between the full and the working samples (1 to 4.8 vs. 1 to 4.5), and the distributions by network are very similar. Other comparisons tell the same story: The exclusions very likely do not account for analytic results reported below.

Dependent Variables
Audience acceptance or rejection is measured by program ratings and shares, as previously described. Ratings and shares are reported on a biweekly or weekly basis in the NTI. 10 Both measures focus on the number of television households viewing a specific program. This number is then relativized to a different base: all households with a television in the case of ratings and only those households viewing any program during the relevant time slot in the case of shares. Advertisers have often been especially interested in ratings, as it indicates the penetration of an advertiser's message into the total potential audience (i.e., the general population of consumers). Shares are better suited for comparing program performance relative to the competition (Webster et al., 2000).
An average rating value and an average share value were calculated for the lifetime of each of the sampled programs, using the NTI data as described above. Although these means are highly correlated (r = .97), each distribution has been used in analyses, in case one of the measures was more sensitive to viewer choices on the BC versus non-BC dimension.

Independent Variables
In addition to the main variable of interest (BC, coded 1, vs. non-BC), four categorical variables comprise the list of conditions available to analysis: genre (drama, coded 1, vs. comedy); network (with CBS as reference category when all four networks are considered); period, which was coded as three partially overlapping dummy variables to explore different possible effects (more below); and season of program debut (with fall as reference category, covering debuts from mid-August to January vs. winter, January through April, and summer, May to mid-August). Season of debut is potentially important as a control, because shows that debut in winter typically had lower chances of survival, as did those debuting during summer, despite the fact that their competition often consisted of "summer reruns." In the present study, the main hypothesized period effect has to do with the report of the Kerner Commission. Period 1 (1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975), coded 1) is designed to tap possible responses in the immediate aftermath of that report. Setting 1975 as the end of that period is arbitrary: If pertinent responses decayed rapidly, 1975 is perhaps too late, but if pertinent responses were slow to occur, 1975 is perhaps too early. However, lengthening the interval increases the probability that other factors, unrelated to the report, will appear as responses to it, while decreasing the interval could work against the fact that program development, including the recruitment of commercial sponsors to pay for the development costs, takes time. (In fact, however, moving the upper terminus did not result in a qualitative change of outcome, and the quantitative difference was very small.)

Basic Analyses
As point of departure, let us revisit the distributions by network of the full sample and of the working sample, both in total and by BC status (Table 1). In addition to the lack of evidence of selection bias, two conclusions stand out. First, the BC shows were distributed across networks in the same proportions as the total of BC and non-BC shows. Second, within network, the proportion of shows that were BC was greatest on FOX; otherwise, the networks were not different. While the FOX difference is in keeping with prior comments about the network's early strategy, it is well short of statistical significance. Also in keeping with prior reports, dramas were scarcer among BC than among non-BC shows (12.6% vs. 30.2%, χ 2 = 15.6, p < .001). Table 2 reports that comparison as well as a comparison of mean ratings and mean shares within the 2 × 2 of race and genre. The main conclusion from the latter comparison is indifference. The main effect of genre could be considered significant (at p < .10, or p < .05 in a one-tailed test), with comedies slightly more successful. But the difference by race (i.e., BC vs. non-BC) was essentially zero (ratings: 11.8 vs. 12.0, p = .54; shares: 20.4 vs. 20.7, p = .77), and the interaction effect (BC × Drama) was also far short of inferential significance (ratings: p = .53, shares: p = .68).
In sum, the analyses so far have demonstrated support for the second hypothesis (the genre effect) but have led to rejection of the first and third hypotheses. Before resting with that conclusion, however, some possible confounding conditions must be examined. For instance, present data confirm that there was a downward trend in ratings and shares (Figure 1). Most of the decline occurred after 1975. During the first several years of observation, ratings and shares fluctuated around a flat (trendless) line, the only notable divergence being an increased volatility during the early 1970s. But from 1975 onward, both ratings and shares drifted downward. 11 Was some aspect of the fact of BC programming implicated in that downward drift? One might ask, for instance, whether a proportionate increase in BC shows, perhaps in network responses to the Kerner Commission report, had an overall negative effect on audience choice of television as an evening entertainment. There is no evidence for it in the present analyses, although this data set is not ideally suited to answer that question. What can be said is that the decline occurred in both BC and non-BC categories, however, without significant difference in rate of decline. Furthermore, if, in line with the comfort zone thesis, White audiences were deflected from television viewing because of programming responses to the Kerner Commission report, that deflection was a long-delayed effect (beginning in 1975, 7 years after the report).
Suppression of a BC effect on ratings and/or shares success is unlikely but should not be ruled out of hand. The question is whether results of the difference-of-means tests have been confounded by other conditions-in particular, network, period, and debut season, as well as genre. Given the relatively small number of BC cases, a multivariate ordinary-least-squares analysis will yield pertinent tests while conserving degrees of freedom. Table 3 reports these tests for the shares measure. Results for the ratings measure are qualitatively identical (and quantitatively different in insignificant degree); thus, they are not shown to conserve page space. 12 Again, there is no evidence in support of any but the second research hypothesis, now including the fourth one. Consistent with results of the bivariate analyses reported above, the genre effect is maintained net of other conditions. Network also made a difference in shares (and ratings), with ABC shows somewhat performing better, and FOX shows performing much worse, than shows on either CBS or NBC. Season of debut was also a significant condition of success, inasmuch as shows debuting in winter (i.e., January through April) performed slightly less well than shows debuting during the fall season. This is not an unexpected result, of course, given the structure of commercial network program development. The regression coefficient for Period 1 is positive and significant, indicating that program inauguration during the years following the Kerner Commission report fared better. However, the weight of this condition was the same for non-BC as for BC programs. Moreover, there is reason to conclude that the coefficient is in fact not describing a period effect so much as it is capturing a chunk of the information shown in Figure 1-that is, the downward trend in shares (as in ratings). When trend is factored out, the coefficient for Period 1 collapses (to .84, with a standard error of .80). The same result holds for each of the other period terms (hence, no point in pursuing them). 13 If a period effect was in fact associated with the Kerner Commission report, it was too subtle to be manifest within the general trend in mean shares and mean ratings.
The chief conclusion from the multivariate analysis is that the empirical observations do not support the main prediction regarding BC programs, as based on Carter's comfort zone thesis. As can be seen in Table 3, the coefficient for BC (vs. non-BC) never reached parity with its standard error. If there was an effect of the status of a program as BC versus not, it was not manifest in these measures of program success.

Discussion
The finding on genre effect fits with prior studies and with the sort of gratification theory perspective discussed by Wright (1986). Television viewers, perhaps wanting to escape the daily realities of life, may seek temporary escape through the comic antics of a favorite television character in a situation comedy, which is less tension-filled or anxiety-ridden than one finds in the typical dramatic program that may include a murder mystery, police chase, or courtroom drama. Furthermore, situation comedies are shorter in duration, and  Note. Cell entries are regression coefficient (top), standard error (middle), t ratio (bottom). OLS = ordinary least squares. *p < .05. **p < .01. ***p < .001.
one does not have to be an "active" viewer to follow the plot of a situation comedy. Hour-long dramatic programs require more intellectual engagement to follow the plot. The fact that programs airing on the FOX network tended to have lower mean ratings and shares than programs on the other networks is likely driven, in part, by the date of FOX's debut, which came after the general plateau in ratings and shares and which was followed by reduced scheduling in prime-time hours during its early years. Then, too, of course, a new network would fare less well in the competition for new programs, especially those likely to be winners, and would be less able to poach proven winners from the established networks. So the network effect was no more unexpected than the effect of season of program debut. Programs beginning during Fall months benefited from the long-established fanfare the latest "new TV season," which for television audiences often meant a build-up of anticipation after a long summer of reruns. Such programs had time to build an audience base, whereas replacement programs coming after the holiday season and New Year were less likely to draw viewers away from their established viewing patterns, which are generally more unconventional, Finally, the fact that the multivariate model even in strongest version did not account for a large portion of the variation in ratings or shares is not an unexpected outcome. While this could well be partly because of the proverbial "missing variable problem" (e.g., variables that should have been included but were unobtainable, such as a direct measure of the motivation for viewing a television program), it is probably mainly an indication of a large random component in the process that determines the longevity and popularity of specific television programs. Simply put, as Bielby and Bielby (1994) observed years ago, "All hits are flukes." A mixture of structural factors within the television industry affects the types of television programs that are produced and ultimately their success or failure. Bielby and Bielby found that the success and ultimately the longevity of television programs occur after the fact, so to speak. The industry attempted to minimize risks of failure by using proven formulas and personnel with successful reputations. Programs that were given the "green light"-those that made it into a network's schedule-were likelier to reside within the comfort zone as programming executives had come to know it. For BC programs, this meant stereotypical representations in situation comedies and marginalized representations in dramatic programs. Supported by advertisers, America's commercial television sought to reach as many potential consumers as possible. This interdependent relationship between advertisers, who desire mass audiences, and commercial television, which depends on advertisers for support, resulted in the television industry creating programs that appealed to the largest audiences possible. Therefore, the programming strategies adopted by the industry involved finding common themes and content that appealed to a heterogeneous society. Television content generally reinforced the core norms and values of American society. To challenge these norms and values might offend segments of the mass audience and result in a loss of the large audience needed by advertisers (Wilson & Gutierrez, 1995).
That argument is in many ways compelling. So why was it not manifest in the results of the analysis of mean shares and ratings in terms of the contrast between BC shows, and especially BC dramas, and shows that were not BC?
One possibility is that the argument itself is not subtle enough. It paints too much in the stark contrast of Black and White, not allowing room in the middle, so to speak. By the measures considered here, BC dramas succeeded on average as well as non-BC dramas. Likewise, BC comedies succeeded on average as well as non-BC comedies. There may well have been a substitution effect accounting for part of that lack of difference: Black households, although not a large portion of the total, could increase the odds of a BC show, perhaps especially a BC drama, achieving equivalent success, by viewing that show in much higher proportion than the proportion of non-Black households viewing non-BC fare during that same time slot. But in addition, it could also have been true that a substantial portion of non-BC households also viewed the BC dramas that did succeed. At least judging by the present data on ratings and shares, there must have been a substantial "cross-over" viewership. It was not large enough to give Shaft a run of more than a handful of episodes, but then the program that alternated with it, Hawkins, staring Jimmy Stewart (an image about as different from Richard Roundtree's Shaft image as most viewers at the time could have pictured), also failed to survive more than a handful of outings. This is not to deny the continued presence of stereotypical characterizations. They were rampant in BC comedies (as, differently, in many non-BC comedies). Indeed, Roundtree's Shaft was in its own way stereotypical, with a characterization that for much of the White audience was seen as an "in your face" flaunting, which accounted for part of its appeal to Black audiences. What was discomforting to one segment of viewing audience could be, and was, comforting to another, and the edge of that contrast worked in both directions.

Conclusion
The present study found no support for Carter's comfort zone thesis with regard to BC television programming. One could argue that the definition of "Black-centered," used to support reliability of analysis, was too loose, and that a tighter definition would have yielded evidence in support of Carter's thesis. Indeed, the definition of BC programs included not only those in which all regularly appearing casts members were African American (e.g., The Cosby Show and Amen) but also those with plots that focused on issues relevant to African American issues, though with mixed casts (e.g., I'll Fly Away and In the Heat of the Night).
But a narrower definition of "Black-centered" would have reduced the power of the statistical tests. Not surprisingly, therefore, a replication of the main analysis using the narrower definition did not lead to a different conclusion. A dimension of program success that has not been directly addressed is duration. All television shows fail; some fail sooner than others. While duration is no doubt a function of mean ratings and mean shares, that relationship is far from complete. This study found no race-based distinction related to ratings or shares, but it was not designed to detect differences in duration as such. Future research should explore this dimension of program success.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) received no financial support for the research and/ or authorship of this article.

Notes
1. "Minority-lead" indicated a program in which only the lead or co-lead character, whether adult or child, was Black, Hispanic, or Asian.

The Kerner Commission-formally, the National Advisory
Commission on Civil Disorders-held the media partially responsible for creating a social environment that resulted in the urban summer riots of the mid-1960s. The limited visibility of representations of Blackness in entertainment portrayals reinforced the inequalities of American society. Even in the make-believe world of television, African Americans were on the periphery when not excluded altogether. The commission's report was released February 29, 1968. 3. Note, too, that FOX was becoming a strong competitor to what had been the three major networks. 4. For a recent extensive overview, theoretical and empirical, see Hartman (2009). 5. The context of this study, bear in mind, is the pre-Internet world. Options have greatly expanded, leading some to argue that greater media richness has posed new opportunities and new problems for media choice, though the "richness" claim has been disputed. See, for example, Robert and Dennis (2005) and Trevino, Lengel, and Daft (2012). 6. Both the Warner Brothers (WB) network and United Paramount Network (UPN) began broadcasting in January 1995. 7. That near-equivalence is telling, as most comedies were 30-min long, while many dramas were of hour length: In general, the a priori chance of success was much lower for a dramatic series.

The Nielsen Television Index (NTI) is a publication of Nielsen
Media Research that issues biweekly or weekly national network television ratings and shares. 9. Rather than appearing weekly, Shaft alternated with Hawkins and The New CBS Tuesday Night Movies. Tenafly was one of four rotating series on the NBC Tuesday/Wednesday Mystery Movie (Bogle, 2001;Brooks & Marsh, 1995). The NTI did not identify rating and share with a specific program in the rotating sequences. 10. The NTI was published every 2 weeks until August 1987; subsequent issues were published on a weekly basis. 11. The missing data problem for the 1978-1985 interval might be confounded in the downward drift, but the fact that the decline continued at roughly the same pace after 1985 suggests that the observed pattern is not an artifact. 12. The corresponding table is available from the author. The correlation between measures was the same within the BC category of cases as within the non-BC category. 13. Factoring out the trend effect also reduced the weight of the FOX effect (to −7.25). The genre effect was unchanged, and although the point estimate for BC was higher (.79), it remained far below significance level.