Political Correlates of Television Viewing

Transcription

1 Cpyright AU.Rights Reserved. Plitical Crrelates f Televisin Viewing GEORGE GERBNER, LARRY GROSS, MICHAEL MORGAN, AND NANCY SIGNORIELLI THE DECLINING salience f plitical parties and the cncurrent shift t candidate-centered media campaigns and vting patterns have raised fears f grwing manipulatin and instability in American plitics. The mass media have usually been blamed fr the apparently increasing "rtlessness" f the electrate and its abandnment f traditinal vting patterns. Yet the evidence attributing these shifts t the media has been incnsistent. A recent review f research n "vter vlatility" (Bybee, et ai., 1981) cncluded that "the vlatility literature suggests that the emerging pattern f televisin as the dminant surce f plitical infrmatin has cntributed t grwing electral instability" (p. 81). But the authrs' wn study fund n supprt fr the hypthesis. If anything, televisin viewing was assciated with lwer rather than higher levels f vlatility. Abstract This study examines relatinships between amunt f daily televisin viewing and plitical self designatins. It is based n the assumptin that televisin's impact n the plitical frnt is nt limited t news and specifically "plitical" prgramming, but that the entire dramatic structure f netwrk televisin may cntribute t viewers' underlying plitical rientatins. The analysis is based n nine independent data bases, yielding a cmbined N f respndents. The findings shw that thse wh watch mre televisin are significantly mre likely t call themselves "mderate:' and t avid the labels f "liberal" r "cnservative." The implicatins f the effects f televisin viewing n the plitical prcess are discussed. Gerge Gerbner is prfessr f cmmunicatins and dean f The Annenberg Schl f Cmmunicatins at the University f Pennsylvania, as well as editr f the Jurnal (~l Cmmunicatin. Larry Grss is prfessr f cmmunicatins at The Annenberg Schl f Cmmunicatins and c-editr f Studies in Visual Cmmunicatin. Michael Mrgan is assistant prfessr at the Department f Cmmunicatin Studies, University f Massachusetts. Nancy Signrielli is research crdinatr at The Annenberg Schl f Cmmunicatins. Public Opinin Quarterly Vl. 48:2K by the Trustees f Clumbia University Published by Elsevier Science Publishing C., Inc X/84/0048-2fO/$2.50

2 ... Cpyright 2001.AII Rights Reserved," 284 GEORGE GERBNER, ET AI.. A study f "candidate vting" (Wattenberg, 1982) fund that while party salience is declining, and televisin is particularly indifferent t party labels, strng lcal party rganizatins can reverse the decline. Research n the rle f media in children's plitical scializatin (Cnway, et al., 1981) fund that children's media use and level f plitical knwledge cntributed mre t plitical behavir than did gender, grade in schl, instructinal methd, r parental media use, but knwledge had a negative relatinship t supprting the party system. The authrs called fr "greater attentin t the creative effects f bservatin and vicarius experience made pssible thrugh the news media" (p. 176). But what kind f attentin? Our lng-standing prject n televisin's cntributins t viewers' cnceptins f reality (e.g., Gerbner, et al., 1980a) pints t significant gaps in previus research attentin t the media's rle in the plitical prcess. First, the effects f televisin may be fundamentally different frm, if nt cunter t, thse f ther media. Unlike ther media, televisin is used relatively nns electively and in massive dses. Therefre, it has the ptential f expsing viewers t stable cntent cnfiguratins that they d nt get frm ther media. Secndly, the plitically significant cntent f televisin is nt limited t, and may nt even primarily reside in, news and public affairs. The "news and public affairs viewer" that mst studies f plitical influence purprt t study is a rare breed. Few turn n nly news and public affairs. Many f thse wh respnd that they "frequently" view news and public affairs are simply heavy viewers f televisin. Thus whatever they get ut f their expsure can als be attributed t dramatic fare, which makes up the bulk f viewing, and which, with its prtrayal f crime, curtrms, and cnflict-ridden urban life, may well be a principal cntributr t basic plitical rientatins. Finally, the declining salience f and supprt fr plitical parties per se des nt necessarily mean the disappearance f plitically relevant labels. On the cntrary, ne can frequently bserve the characterizatin f plitical-and ther-persnalities as "liberal," "radical," "cnservative," and endwed with ther designatins that carry plitical meanings. The thery we are testing and develping is based n these cnsideratins and n the previusly reprted results f ur nging research int plitical rientatins (Gerbner, et ai., 1982). It pstulates relative standardizatin, hmgeneity, and stability 3S the mst pervasive assciatins with televisin viewing. It views televisin as different frm ther media (althugh mre inclusive f the cultural

3 ... Cpyright 2001, All Rights Reserv.ed... POLITICAL CORltELATES OF TV VTEWING 285 mainstream) in that mst viewing is an essentially time-gverned ritual. Fr mst viewers wh watch several hurs a day, the different types f prgram selectins make relatively little difference. We als cnsider televisin a pervasive cultivatr f significant labels and self-designatins, and thus a pwerful new frce n the plitical scene. The wrld prtrayed n netwrk televisin-the mediated wrld mst viewers attend t mst ften-is fashined under the imperatives f mass market appeal. Cntent analyses f televisin (Gerbner and Grss, 1976; Gerbner, et al., 1978, 1979) and studies f the medium's institutinal prcesses (Barnuw, 1975; Cantr, 1980; Tuchman, 1974) suggest that thse imperatives tend t favr the prtrayal f cnventinal, "balanced," middle-f-the-rad images and values. A viding extremes has lng been televisin's preferred strategy fr survival, as netwrks and advertisers have cme t expect attacks frm vcal grups n the right and the left. Faced with such cnflicting pressures, the televisin industry takes the bvius way utnavigating between the extremes, safely in the mainstream, in the cmfrtable and suppsedly nnidelgical middle grund that hlds the largest pssible audience. This suggests that while liberals find televisin t "cnservative," and cnservatives find it t "liberal," the truth f the matter prbably lies-literally-smewhere in between. But where "in between" des televisin steer its curse? Our analyses (Gerbner, et ai., 1982) suggest that televisin's mainstream runs well t the right f center n plitical issues and clser t traditinal liberal psitins n demands fr gvernment services t a cnsumer-riented ecnmy. At the same time, the fficial media idelgy f balance, bjectivity, and mderatin leads us t expect that heavy expsure t the relatively hmgenized presentatins f mst televisin fare shuld cultivate generally "mderate" self-perceptins amng viewers. Cmpetitin fr the largest pssible audience at least cst means striving fr the bradest appeals; the underlying demgraphy, actin structure, values, and idelgy are stable and lng-lasting. In news, "balanced bjectivity" means, in practice, that the "mderate" center f the visible plitical spectrum is prtrayed as the nrmal, nn idelgical slutin t cntrversial and sensitive issues. In general, divergent r deviant psitins are presented as mstly t be shunned r suppressed-their prpnents are readily dismissed as extremists wh are generally canceled ut by their balancing ppnents n the ther side f the mderate "center." Our research, called Cultural Indicatrs, has been mapping the

4 Cpyright All Rights Reserved. 286 GFAlRGt: GIIRBNER, EI' AL. stable and recurrent features f the wrld f televisin and relating them t viewers' cnceptins f scial reality since We have cnsistently fund that thse wh spend mre time watching televisin are mre likely t express views, beliefs, and assumptins that are cngment with televisin's prtrayals f life and sciety. Amllnt f televisin viewing has been fund t make an independent cntributin t a wide range f cnceptins f reality within and acrss relatively hmgenells subgrups. These cnceptins include images f vilence and mistrust (Gerbner, et ai., 1979, 1980a), aging and age rles (Gerbner, Grss, Signrielli, and Mrgan, 1980), sex rle steretypes (SignrielJi, 1979; Mrgan, 1982), racial and sexual minrities (Grss, 1982), science and scientists (Gerbner, et ai., 1981a), health-related beliefs and practices (Gerbner, et ai., 198Ib), and ther issues. Sme f this wrk has been critiqued by Hughes (1980) and Hirsch (1980, 1981a, 1981b); fr ur respnses, see Gerbner, et a1. (1980b, 1981c, 1981d). (Fr an extended discussin f sme f these pints, including independent cnfirnmtins, see Hawkins and Pingree, 1982). We have recently begun t investigate relatinships between televisin viewing and plitical rientatins (Gerbner, et ai., 1982). This wrk, based n fur years f the NORC General Scial Surveys, suggests that thse with greater immersin in the wrld f televisin are indeed mre likely t present themselves as plitically mderate. In this paper, we replicate, extend, and refine that study, using data frm the 1982 General Scial Survey and fur additinal data bases, including three large natinal surveys. Mainstreaming Televisin's assciatins with utlks and behavirs need nt be unidirectinal r f equal intensity acrss the bard. Rather, the patterns depend n the different scial, cultural, and plitical cntexts in which peple live. In many cases, televisin seems t diminish r verride the influence f these ther scial frces. This is what we call "mainstreaming:" a cnvergence f cnceptins and attitudes held by the heavy viewers f different grups wh share little besides televisin. If the televisin mainstream can be thught f as the aggregate cmmn patterns f images and values that pervade televisin, then mainstreaming is the expressin f that shared cnsciusness by heavy viewers in thse grups whse light viewers hld divergent views. Mainstreaming thus reflects a specific kind f systematic interactin between televisin and backgrund factrs, ne

5 ... Cpyright AIL Rights Reserved... I'OLfl'ICAL C'ORR~;LATES OF TV VIEWING 287 which generally results in a narrwing f differences amng subgrups. The key t the ntin f mainstreaming rests n the fact that relatively lighter viewers in different grups d nt share cmmn perspectives. While thse wh watch mre televisin are mre likely t perceive the wrld as it is presented n televisin, the beliefs f thse wh watch less televisin are mre diversified and are mre affected by ther scial and cultural influences. In ther wrds, while televisin may cultivate ntin Y amng grups whse light viewers are mre likely t believe X, it may als cultivate Y amng grups whse light viewers are mre likely t believe Z. We are nt suggesting that televisin cntent can r need be directly traced t a specific issue r phenmenn in rder t study the pssibility f plitical mainstreaming. Rather, we are asking whether the crrelates f heavy viewing include the tendency t see the plitical wrld in a manner generally cnsistent with televisin's repetitive cnventinal prtrayals. It shuld be stressed that ur use f the term "crrelates" in this discussin has deliberate implicatins. First and fremst, ur data are frm crss-sectinal surveys, thereby preventing the pssibility f drawing cnclusins abut tempral causality, as in befre-and-after experiments. But perhaps mre fundamental1y, we wish t avid the classic cncern with "causality" as a unidirectinal and unidimensinal ntin. (In strict time-rder terms, televisin is the factr which "cmes first," since children are nw brn int a televisin envirnment and begin t watch well befre they can read r even speak-and certainly well befre they adpt plitical labels.) The critical questin is whether televisin's assciatins with these labels are independent f ther factrs which are plausible surces f spuriusness. We believe that bth televisin viewing and plitical selfdesignatin ccur within a dynamic, nging system. Heavy viewing is an integral aspect f certain styles f life, characterized by specific ways f lking at the wrld. Ritualistic expsure t televisin's repetitive fare can be expected t sustain, nurish, cnfirm, and strengthen-i.e., cultivate-these rientatins, perspectives, and behavirs. In sum, we are arguing that heavy televisin viewers shuld be mre likely t apprach the plitical wrld in mre hmgeneus and cnventinal terms than light viewers living under similar cnditins. If these assumptins are valid, then we shuld be able t find systematic relatinships between amunt f televisin viewing and the tendency t designate neself as "mderate;" just as we have fund

6 ... Cpyright All Rights Reserved, 288 GEORGE GERBNt:R, ET AL. such relatinships between televisin viewing and ther relatively hmgeneus, mainstream perceptins f the wrld (Gerbner, et al., 1980a). Methds SAMPLES OF RESPONDENTS The fllwing analyses are based n nine separate data bases, representing 14,067 respndents (unweighted). These are all the datasets in ur archives which cntain apprpriate questins n bth televisin viewing and plitical self-designatins. Five f these are the Natinal Opinin Research Center's General Scial Surveys (NORC/GSS) fr the years which included a questin n amunt f televisin viewing (1975, 1977, 1978, 1980, and 1982). Abut 1,500 respndents tk part in hur-lng persnal interviews each year (between February and April), fr a cmbined N f 7,526. The 1975 sample was drawn thrugh a cmbinatin f blck qutas and prbability sampling; the ther fur years are full prbability. Three f the ther datasets are als large natinal surveys. One was cnducted fr the Cnnecticut Mutual Life Insurance Cmpany by Research and Frecasts, Inc., between September and Nvember, This sample cnsists f 1,610 respndents selected thrugh randm-digit-dialing prcedures, representative f Americans aged 14 and lder, and an versample f 408 teenagers, lder peple, and blacks. The general public and the versampled respnses were weighted t match census figures (n age, ethnic rigin, gender, educatin, and family incme) t prduce a weighted N f 1,610. Anther was cnducted by The Rper Organizatin fr Virginia Slims, Inc., in Octber This sample represents a multistage stratified prbability design at the blck level with qutas fr sex; persnal interviews were carried ut in respndents' hmes with 2,960 wmen and 984 men aged 18 and ver. Anther natinal persnal interview survey was cnducted in This survey, with 1,635 respndents, was cncerned with public attitudes tward science and technlgy (see Miller, et ai., 1980; Miller and Prewitt, 1983). Finally, fr variety, we include data frm a small sample f adlescents (eighth graders) frm rural Minnesta. Self-administered questinnaires were filled ut in class by 294 students, in grups f abut at a time, under teacher supervisin. The data were cllected in Nvember 1981.

7 Cpyright 2001,AIIRightsReserved... POLITICAL CORRELATES OF TV VIEWING 289 MEASURES The independent variable in this research is average amunt f daily televisin expsure, taken frm respnses t survey questins. Questin wrdings and respnse categries varied smewhat amng the nine data bases analyzed here, and we made adjustments t increase cmparability acrss samples. Fr example, the NORC/GSS televisin questin deals with hurs n an average day. The Virginia Slims questin is cded in minutes, s we divided respnses by 60; the Science and Technlgy questin deals with hurs in an average week, s we divided respnses by seven. Mre imprtant, this measure is nt interpreted as prviding abslutely accurate reprts f actual number f viewing hurs, but is used t differentiate viewers. with relatively mre r less expsure t the wrld f televisin. When nt used as a cntinuus variable, amunt f average daily viewing was trichtmized t allw the cmparisn f relatively light, medium, and heavy viewers. The divisins were made at the pints which prvided the clsest apprximatin f an even three-way split within each survey. In general, this translated t ne hur r less (light viewers), tw r three hurs (medium viewers), and fur hurs r mre (heavy viewers). The dependent variable is plitical self-designatin, alng the liberal-mderate-cnservative dimensin. We are assuming that these labels are salient and meaningful t respndents, wing in part t the generally accepted and frequent use f the terms. Items measuring plitical self-designatin als varied acrss surveys. Respndents in the NORClGSS and Science and Technlgy surveys were asked t place themselves n a seven-pint plitical scale (with "plitical" emphasized), ranging frm "extremely liberal"' t "extremely cnservative." The Virginia Slims questin specifically asked respndents abut their general utlks, 'plitically and scially," placed n a five-pint scale. The Cnnecticut Mutual survey simply asked whether respndents were liberal, mderate, r cnservative n mst plitical matters, and a five-pint scale was used with adlescents. Except fr the Cnnecticut Mutual survey (which used nly three respnse categries), we recded respndents as simply either liberal, mderate, r cnservative. Table 1 presents the mean amunt f viewing and the percentage in each plitical self-designatin categry acrss samples. These variables are relatively stable acrss time and surveys, but there are sme interesting variatins. Acrss the five NORC/GSS samples, amunt f viewing shws n linear trend, but significant deviatins frm linearity (p<.ol); mean viewing levels declined frm 1975 t 1978 and have

8 ---.--~-~--- Cpyright All Rights Reserved. 290 GEORGE GERB'iER, ET At. Table 1. Amunt f Viewing and Plitical Self-Designatin by Sample _._ TV Viewing % Wh Are ---_..._ Sample (N) /\1ean ~.d. Liberal Mdem!" _ Cnservtive NORC/OSS 1975 (1,490) % 40.0% 29.9% 1977 (1,530) (1,532) (1,468) (1,506) Science and Technlgy ( 1,635) '~ Virginia Slims (3,944) Cnnecticut Mutual (1,610) Minnesta Adlescents (294) increased since. The percentage wh call themselves mderate has been particularly stable in the NORC/GSS surveys, varying between 38.3 percent and 40.8 percent; neither linear nr nnlinear trends are significant. There is a significant negative linear trend in the percentage wh call themselves liberal (p<.oi), even thugh this prprtin increased a bit in The percentage f cnservatives steadily increased frm 1975 t 1980 but declined slightly in 1982; neither linear nr nnlinear trends are significant. Cnnecticut Mutual respndents and, nt surprisingly, the Minnesta Adlescents, reprt higher amunts f viewing. Virginia Slims respndents were mre likely t say they are cnservative, perhaps because the questin emphasized "plitical and scial" utlks. Cnnecticut Mutual respndents were mre likely than thers t say they are mderate, perhaps because nly three respnse categries were used. Finally, it is interesting t nte the rather high prprtin f adlescents (65.7 percent) wh chse the mderate label, even with a five-pint scale. CONTROLS Ten basic cntrls were used in the analysis: sex, age, educatin, incme, ccupatinal prestige, area f residence (urban/nnurban), race (white/nnwhite), frequency f church attendance, unin membership, and plitical party affiliatin. N ther factrs which appear in mst f the surveys seemed relevant. These backgrund demgraphics als varied in wrding and respnse categries acrss surveys. As a result, different cutff pints had t be used fr subgrup analyses in different surveys, particularly fr such variables as incme and area f residence. Every attempt was made t match

9 . POLITICAL CORRELATES OF TV VIEWING 2<11 categries as clsely as pssible, and we feel they are sufficiently similar t allw meaningful cmparisns. The availability f these measures als differed amng samples. All 10 appear in the Cnnecticut Mutual and NORC/GSS surveys (except fr unin membership in the 1977 and 1982 GSS). Church attendance is missing in bth the Virginia Slims and Science surveys, and the latter has n measure f party affiliatin. Fr Minnesta adlescents, we used nly sex, age, and the educatin and ccupatinal status f the respndent's father. Altgether, we were able t analyze 198 distinct but verlapping subgrups (including verall cmparisns). Results Overall, there is a cnsistent tendency fr thse wh watch mre televisin t call themselves "mderate." Table 2 shws the percentage f light viewers in each plitical self-designatin categry and the percentage spread between them and the heavy viewers in that categry. (Medium viewers are mitted fr purpses f space.) This difference is referred t as the cultivatin differential, r CD. Fr example, the tp rw f Table 2 shws that in the 1975 GSS, 32 percent f the light viewers said they were "mderate"; since the CD = + 12, we see that 44 percent f the heavy viewers said they were "mderate," In all nine surveys, heavy viewers are mre likely than light viewers t chse the "mderate" self-designatin, and the re- Table 2, Simple Relatinships Between Amunt f Televisin Viewing and Plitical Self-Designatin % Wh Say They.4re Liberal Mderate Cnservative %L CD Gamma %L CD Gamma %L CD Gamma NORC/GSS * * J977 3! * * * ' * -.11* * Tech VA Slims * * Cnn. Mut * Minn. Ad * Sci. and * * -- 5 NOTE: %L = % light viewers in categry. CD = Cultivatin Differential: % heavy viewers in categry minus % light viewers in categry. *p<.05.

10 Cpyright All Rights Reserved:.~ GEORGE GERB~ER, ET AL. latinship is significant in all but the 1982 GSS.l While there are variatins acrss samples and years, the simple assciatin appears in these nine independent samples. It is wrth nting at this pint that the tendency fr heavy viewers t call themselves mderate seems t be a specific crrelate f televisin, and nt f general media use. These surveys include varius questins n newspaper reading, radi listening, and s n. As can be seen in Table 3, ther media variables relate t plitical selfdesignatin in strikingly different ways than des televisin. In par- Table 3. Simple Relatinships Between Plitical Self-Designatin and Usc f Other Media % Wh Say They Are Liberal Mderate Cnserl'ative %L Diff Gamma %L DifJ Gamma %L Diff Gamma NORCIGSS Newspaper * * * * * * 41 -I * Radi * '-.09* ' Science Newspapers Radi * VA Slims "Reading" \0* * Cnn. Mutual Newspapers \0* Minn. Ad. Mvies 13 t Bks * 72 -II -.26* Radi II Newspapers \ NOTE: Fr Adult Samples: Radi: lw = 0,1 hurs; medium = 2,3 hurs; high 4 hurs r mre. Newspaper: lw = under nce a week; medium.'~ a few times a week; high,~ every day. Fr Virginia Slims, "Reading" = time spent reading bks, magazines. and newspapers (single questin). Fr Adlescent Sample: All dichtmizatins (median split). % L= % in "Lw Use" Grup in Categry. Diff = Spread between % in "Lw" and "High" Use Crups. *p<.05. I Preliminary analysis f the 1983 GSS, in which the plitical self-designatin questin was addressed t half the sample, shws a resumptin f the pattern seen in all previus years: heavy viewers are smewhat less likely t describe themselves as liberal and much less likely t describe themselves as cnservative.

11 POLITICAL CORRELATES OF TV VIEWING 293 ticular, relatinships with the "mderate" categry are negative r zer; there are three weak psitive assciatins, but nne is significant. Since heavy users f either f these media shw n greater tendency t perceive themselves as mderate, we can cnclude that the bserved relatinship between televisin viewing and selfdesignatin is nt a general media use phenmenn, but a pattern unique t televisin. In general, heavy newspaper readers are mre likely t say they are cnservative, while heavy radi listeners tend t say they are liberal. In any case, these data strengthen ur cntentin that televisin is fundamentally different frm ther media. Hw des the relatinship between televisin viewing and plitical self-designatin vary amng subgrups? Which grups tend t shw the basic pattern and which tend t deviate? Table 4 presents aggregate results fr each subgrup, averaged acrss the nine surveys (r acrss as many surveys as include a given cntrl variable), and Table 4. Average Within-Grup Relatinships Between Televisin Viewing and Plitical Self-Designatins (Weighted by Sample Size) % Wh Say They Are Liberal Mderate Cnservative %L CD Gamma %L CD Gamma %L CD Gamma Overall Male Female Yung Middle Old L Educ Hi Educ II L Incme Md Incme Hi Incme L Occup Md Occup Hi Occup White () Nnwhite Urban Nnurban Unin Nnunin Demcrat Independent Republican L Attend Hi Attend NTE:.%L=% light viewers in categry. CD=Cultivatinal Differential: % heavy viewers In categry minus % light viewers in categry

12 Cpyright 2001.AII Rights Reserved: 2'14 GEORGE GERBN~:R, ET At. weighted by the number f cases in each sample t adjust fr the wide range f sample N's. The relatinship between amunt f viewing and the tendency t chse the mderate self-designatin (middle clumns f Table 4) reveals very few differences wrth nting. In all 24 subgrups, n the average, heavy viewers are mre likely t say they are mderate. The assciatin is weakest amng nnwhites and Demcrats. But all f the ther average CD's are between + 7 and + II, with gammas ranging frm.09 t.i5-again, the acrss-grup variatins in this pattern are nt particularly prnunced. The data fr the percentage f cnservatives als reveal a highly cnsistent pattern acrss subgrups. All cefficients are negative except amng Demcrats, and the average relatinship is particularly strng amng Republicans and high church attenders (the tw grups in which light viewers are mst likely t be cnservative). Thus, these variatins illustrate the ntin f mainstreaming as a cnvergence amng heavy viewers f cunterpart subgrups. Relatinships in the liberal clumn, n the ther hand, are smaller and less cnsistent. While mst f the average within-grup assciatins are negative, in fur subgrups (thse with less educatin, nnwhites, Republicans, and high church attenders) heavy viewers are slightly mre likely t be liberal Except fr the case f nnwhites, thse three psitive assciatins als retlect a cnvergence amng heavy viewers. The grups whse light viewers are mst likely t be liberal shw negative assciatins, while the grups whse light viewers are least likely t be liberal are the nes shwing psitive assciatins. Fr example, there is a spread f abut 23 pints between light viewing Demcrats and Republicans in the tendency t chse the liberal label; this is reduced by abut half, t 12 pints, between heavy viewing Demcrats and Republicans. Similar patterns hld in terms f age, educatin, and church attendance, but this paradigm clearly des nt fit the differences between whites and nnwhites. In sum, acrss all majr subgrups in each f the nine samples, heavy viewers are cnsistently mre likely t present themselves as mderate. They are als generally less likely t say they are either liberal r cnservative, althugh the relatinships between amunt f viewing and the percentage f liberals tend t be weaker and shw mre exceptins. The tendency fr heavy viewers t chse the mderate label hlds up under mre stringent cntrls. We analyzed partial crrelatins between amunt f viewing and the tendency t chse each f the three plitical self-designatins under single and multiple simulta-

13 ... ".. ".,," Cpyright 2001, AU Rights Reserved. POLITICAL CORRELATES OF TV VIEWING 295 neus cntrls within every available subgrup in each f the nine surveys. This yields 1,628 first-rder partials and 198 nth-rder partials fr each plitical self-designatin categry.. Table 5 summarizes the results. The table shws hw many partial relatinships between televisin viewing and each plitical selfdesignatin categry are psitive r negative and hw many f these are significant. The mst imprtant finding is the paucity f negative relatinships in the "mderate'" clumn. Abut 80 percent f the crrelatins are psitive under either single r simultaneus cntrls. Of these, abut half f the first-rder partials and abut 40 percent f the nth-rder partials are significant-far mre than culd be attributed t chance. In cntrast, nne f the negative crrelatins between amunt f viewing and the tendency t chse the mderate label are significant under simultaneus cntrls. Thus, with cnsiderable cnsistency, heavy televisin viewers are mre likely t say they are mderate. Strictly speaking, f curse, the within-sample tests are nt independent since respndents belng t mre than ne subgrup, but the stability f the basic pattern under cntrls is impressive. This again shws that the tendency fr heavy viewers t see them- Table S. Summary f Within-Grup Partial Crrelatins Between Amunt f Viewing and Plitical Self-Designatin First-Order Partials Psitive (N) % Sig. (N) Negative (N) % Sig. (N) Partials Between Amunt (!{Viewing and % Wh Are: ----~ _--- Liberal Mderate Cnservative % (703) (110) 56.8 (925) 23.1 (214) % 17.9% (1371) (291) (682) (17) (257) (1337) L (3) (632) Nth Order Partials Psitive (N) % Sig. (N) Negative (N) % Sig. (N) 38.4% (76) (4) 61.6 (122) 21.3 (26),, 78.3% 30..3% (155) (60) (61) (8) 2L (43) (138) 34.8 (0).. (4~_)_ _..,, NOTE: First-rder clumns each based n J,628 crrelatins, representing all pssible first-rder partials within all available subgrups in each sample. Nth-rder clumns each based n 198 crrelatins, representing simultaneus implementatin f all available cntrls within all available subgrups in each sample.

14 ... Cpyright 2001, All Rights Reserved. 296 GEORGE GERBNER, ET AL. selves as mderate generally retlects a narrwing n bth ends f the plitical spectrum. The majrity f relatinships between viewing and chsing either the liberal r the cnservative label are negative. (Multiple cntrls increase the percentage f negative assciatins in the liberal clumn but decrease the percentage in the cnservative clumn.) Cmpared t the psitive crrelatins in the mderate clumn, these negative relatinships are less likely t be significant. Still, the trend is clear: heavy viewers are generally less likely t chse either the liberal r the cnservative label. Mrever, relatively few f the psitive assciatins between viewing and chsing either the liberal r cnservative labels are significant. These findings reflect aggregate patterns acrss all nine surveys. There are, hwever, sme imprtant variatins amng the samples. Table 6 summarizes the nth-rder simultaneus partials between Table 6. Nth-Order Within-Grup Partial Crrelatins by Sample (Multiple Simultaneus Cntrls) (N) --- NORClGSS 1975 Ttal (25) Sig Ttal (23) Sig Ttal (25) Sig Ttal (25) Sig Ttal (23) Sig. Science Ttal Sig. VA Slims Ttal Sig. Cnn. Ttal Sig. Minn. Ttal Sig. (20) (23) (25) ========== ".-._-_-_ _ -c= Relatinships Between Amunt f Viewing and % Wh Are: -c Liberals Mderates Cnservatives Ps. Neg. P;;. Neg. Ps. Neg. --_ I I 18 I 7 () (9) () I --=-----._--_ _....-_..._. 13 I I _ () () 20 2 () 2 () 5 () I 23 II

15 Cpyright 200LAIiRightsReserved. POLITICAL CORRELATES OF TV VIEWTh'(; 297 amunt f viewing and plitical self-designatin fr each survey. Within each sample, the bulk f relatinships in the mderate clumn are psitive. with ne striking exceptin: the 1982 GSS. In this, the mst recent survey in this study, 20 ut f 23 partial assciatins between viewing and chsing the mderate label are negative, and the same number f relatinships in the cnservative clumn are psitive. A smewhat similar pattern ccurs in the 1977 GSS, but nly in the cnservative clumn. Relatinships between televisin viewing and the tendency t chse the liberal label als vary ver time and acrss surveys. But the 1982 relatinships are the clearest and mst drastic exceptin t the general rule. Althugh nne f the 1982 negative assciatins between viewing and chsing the mderate label is significant, these data reflect a cmplete reversal f the trends fund in all ther datasets. It is certainly t early t determine whether this signals a real shift in the pattern bserved between 1975 and 1980 (all ur ther adult samples fall between these dates), r a temprary fluctuatin which will disappear in subsequent years. In any case, heavy viewers in ]982 are generally mre likely t call themselves cnservative. Cnclusin The amunt f time peple spend watching televisin relates systematically and cnsistently t hw they psitin themselves alng the liberal-mderate-cnservative plitical cntinuum. These data shw that heavy viewers are mre likely t present (and presumably, perceive) themselves as "mderate, middle-f-the-rad." This cnclusin must be tempered smewhat by the results in the 1982 GSS, where we find an almst cmplete reversal frm the therwise cnsistent pattern f self-prclaimed mderatin. It is pssible that these data merely reflect randm fluctuatins and that subsequent surveys will replicate the dminant patterns. As we nted in ftnte 1, the 1983 GSS data shw the previus pattern f heavy viewers describing themselves as mderate, thus supprting the pssibility that 1982 was a temprary fluctuatin rather than a shift in the plitical climate. In any case, ur primary finding des nt seem t be spurius, at least in terms f the demgraphic subgrups we cnsidered. The questin f the directin f any causal assciatin between televisin viewing and plitical self-designatin wuld be mt if the relatinship were t be explained by any f these variables" Of curse, radical leftists and ultra-cnservatives prbably d nt watch much televisin, and n ne wuld argue that their light viewing "causes" their

16 Cpyright All Rights Reserved. 298 GEORGE GF.RIlI\ER, ET AI>. plitical utlks. But this des nt explain r negate ur interpretatin f these data, unless the apprximately three-fifths f the adult ppulatin that call themselves liberal r cnservative can be cnsidered particularly "extreme." The ntin that heavy viewing is merely symptmatic f sme set f "mderate" predispsitins cannt, f curse, be ruled ut n the basis f crss-sectinal data, but it seems far-fetched, In fact, that explanatin cannt easily accunt fr the exceptins fund in the 1982 GSS, unless ne argues that suddenly cnservatives are watching mre televisin. Whether we are seeing the effects f plitical transfrmatins r subtle changes in televisin cntent-r bth--we prefer t cnclude that televisin is part and parcel f these cnstellatins which it helps t maintain. Except fr the 1982 data, these findings may be part f a mre general phenmenn: the cultivatin f hmgeneus, "average" self-perceptins. Onging analyses (Gerbner, et ai., 1982) reveal that peple whse bjective scial class is lw and wh watch mre televisin are mre likely t call themselves "middle class"; amng high SES respndents, hwever, heavy viewers are mre likely than light viewers t call themselves "wrking class." In additin, heavy viewers are mre likely t say they have "average" incmes, particularly as real incme increases. Again, these patterns may be traced t general prgram plicies r cnventins and they may stem frm lng-term, cumulative expsure t shared, cmmn imagery amng large and therwise hetergeneus publics. In general, televisin cultivates "mderate" self-designatins and a cnvergence upn the plitical mainstream. But that des nt necessarily mean that televisin is a frce fr genuine mderatin. Again, the actual psitins n issues taken by heavy viewers under these labels (Gerbner, et ai., 1982) des nt fit the image f plitical centrality r mderatin. Viewing seems t blur traditinal differences and t blend them int a relatively hmgeneus mainstream that is less vlatile in terms f extremes and penness t alternatives. It is typically perceived as "mderate"; whether that cnceptin is shifting and the label "cnservative" is becming a mre acceptable label fr the mainstream, pssibly replacing "mderate," we d nt yet knw. But we knw frm previusly reprted data that the mainstream bends t the right n issues dealing with minrities and persnal rights, reflects the anxieties and mistrust f televisin's vilent "mean wrld," and tilts t expansive ppulism n sme welfare ecnmic issues. A vlatility f substantive rientatins may hide behind the relatively stable and bland plitical label that televisin tends t cultivate.

17 Cpyright 200LAIiRightsReserved, POLmCAL CORRELATES 01' TV VIEWING References Barnuw, E Tube f Plenty: The Evlutin'f American Televisin. New Yrk: Oxfrd University Press. Bybee, C. R., 1. M. McLed, W. D. Luetcher, and G. Garramne 1981 "Mass cmmunicatin and vter vlatility." Public Opinin Quarterly 45: Cantr, M Prime-Time Televisin: Cntent and Cntrl. Beverly Hills, CA: Sage. Cnway, M. M., M. L. Wyckff, E. Feldbaum, and D. Ahern 1981 "The news media in children's plitical scializatin." Public Opinin Quarterly 45: Gerbner, G., and L. Grss 1976 "Living with televisin: the vilence prfile." Jurnal f Cmmunicatin 26(Spring): Gerbner, G., L. Grss, M. Jacksn-Beeck, S. Jeffries-Fx, and N. Signrielli 1978 "Cultural indicatrs: vilence prfile n. 9," Jurnal f Cmmunicatin 28(Sumrner): Gerbner, G., L. Grss, M. Mrgan, and N. Signrielli 1980a "The 'main streaming' f America: vilence prfile n. 11." Jurnal f Cmmunicatin 30(Summer): b "Sme additinal cmments n cultivatin analysis." Public Opinin Quarterly 44: a "Scientists n the TV screen:' Sciety 18(May/June):4l b "Health and medicine n televisin." New England Jurnal f Medicine 305(Octber): c "A curius jurney int the scary wrld f Paul Hirsch." Cmmunicatin Research 8: d "Final reply t Hirsch." Cmmunicatin Research 8: "Charting the mainstream: televisin's cntributins t plitical rientatins." Jurnal f Cmmunicatin 32(Spring): , Gerbner, G., L. Grss, N. Signrielli, and M. Mrgan 1980 "Aging with televisin: images n televisin drama and cnceptins f scial reality." Jurnal f Cmmunicatin 30(Winter): Gerbner, G., L. Grss, N. Signrielli, M. Mrgan, and M. Jacksn-Beeck 1979 "The demnstratin f pwer: vilence prfile n. 10." Jurnal f Cmmunicatin 29(Summer):177-%. Grss, L "The cultivatin f intlerance: televisin, blacks, and gays." Paper presented t the Sympsium n Cultural Indicatrs fr the Cmparative Study f Culture, Vienna, Austria. Hawkins, R. P., and S. Pingree 1982 "Televisin's influence n scial reality." In D. Pearl, L. Buthilet, and 1. Lazar (eds.), Televisin and Behavir: Ten Years f Scientific Prgress and Implicatins fr the Eighties, Vlume 2. Rckville, Maryland: U.S. Department f Health and Human Services Publicatin N Hirsch, P "The 'scary wrld' f the nnviewer and ther anmlies: a reanalysis f Gerbner et al.'s findings f cultivatin analysis, part 1." Cmmunicatin Research 7: a "On nt learning frm ne's wn mistakes: a reanalysis f Gerbner et a1.'s findings n cultivatin analysis, part II." Cmmunicatin Research 8: b "Distinguishing gd speculatin frm bad thery: rejinder t Gerbner et al." Cmmunicatin Research 8: Hughes, M "The fruits f cultivatin analysis: a reexaminatin f the effects f televisin

18 Cpyright All Rights Reserved. 300 GEORGE GERBNER. ET AL. n fear f victimizatin, alienatin. and the apprval f vilence." Public Opinin Quarterly 44: Miller, J., imd K. Prewitt 1983 The American Peple and Science Plicy. New Yrk: Pergamn. Miller, J., K. Prewitt, and R. Pearsn 1980 Attitudes f U.S. Public Tward Science and Technlgy. Chicag: Natinal Opinin Research Center. Mrgan, M "Televisin and adlescents sex rle steretypes: a lngitudinal study." Jurnal f Persnality and Scia! Psychlgy 43: Signrielli, N "Televisin's cntributin t sex rle scializatin." Paper presented at the Seventh Annual Telecmmunicatins Plicy Research Cnference, Skytp, Pennsylvania. Tuchman, G The TV Establishment: Prgramming fr Pwer and Prfit. Englewd Cliffs, NJ: Prentice Hall. Wattenberg, M. P. 19R2 "Frm parties t candidates: examining the rle f the media." Public Opinin Quarterly 46: