It ... 3.. 3 .33. I333... . 3 . . 3 .3)! 3 3 . ..3 3 3| 3 . , fl 3. 333 3hr... u. 3 ill . H.....3. 33 33 33313 H 3»? LC. ”1 33,303.? .3 >333 3313.3 Ill-t3 AL. 3. A 3.33. 3 3. . ‘1 3.3 .3 3 33.33 3.3 .13. . 3 3 . . . 3 . . "€33. 331.. 3:13dn33.3 3 : . . 33 33.33333 {If} .. “33".3393 31.0.“ I 3 3 3 1.34.3.3 33.311131. 3 3 . t. 3.33.31.3th . 3 - . - 33M...33l.33.3. .3333» 3 3 Clprdifhf” 3333. 3 3 D . I .3..\33 .1 ubqql I’m“ 343 3 II I 3 I (33333....3 31333. . w. 3 - . . .r..3\l.. 3H3! 3.3!.ut... 3 .3: 3 3333. 3333333931335“ .3 a 3. . 33 33 l3. ‘lllnl 33 3 a u A 33.333315 ”3.333.333 . . 3oof|l..rl.l333 3-3 ‘13... 3o 3 33"341..{3333 33F}. 3 3 3 3‘ I 3fl1H3d“-.3 3.3.“. 3 3. 3;. 33.0 u .33 . I3 J 3 3a ll .hvna..3n.3 3W3!) . .Uhvr L d3...” H33. ”3% 3 3le4.jvdr I! .33 33 tun-l3! “4,431.3 '43": . _ o it «IIIFDU . 3. 3.3 1.3!? II 3 .!3 "3’33 '3 I 4| 3’: 3 I . '3')»ng .3 . 33.. {I .3 13 i I3 i 3 2333.33: 333 .33.. 1.3.33 3 31"... 13 3! 3 313 .3 I, 3 3.3 3!. 3 33 I 33 . .3333 1?. L433. Ll lug) 3 U 3311.! .ll'; 1’1 3.3 33. .3-.. .3. .3 .3 :133733313 .1333...||33\ I"... 33 333;)0'3 33.33 In .1333” .(3n3H3333U u 3.. 33.33.333.333. 43333 A $333.35 .3 "3 3.. 3. .3..33333..JI,I.3..3P . ’ Ha . I3. I 33731.»...33.Iuh3v3 .333! b.3303." Ill! 33 .3333 3.33334 3 33333 3. a 33.)..3ufl3rifll33 I 3W3... .J . 33 3. .33 3N3333F3.3 33“.!” 333K333 .3 3 . 33..3I3H 333.36 ”3333““. 3315' 3/ 3 333303 33 [333333 33“). 33I33l 3O: .33 1 3 . 3:34.. {133313 .mQI. 3 3 3 333333333 30333433133 3 ’3 v 3 3 3 f“? 3 .331 A.) 3. .3_.[|3I33 1.3 L:- . 333p: 3333.) (333" .3 .3... 3H *Kfl3. ’1 I: 0‘33”” .3"? I". 33“.” .353 . 3 3 .3 . [“13 . HUN 333v'l".33afl3..3vlcuu 3|...I3 3 3 1.3 I..3W...33Hh3h 47 3. 3 3 3 i 3 .‘lfll « 3 .3: 3|.O 633.333. . 3 3 .3 3. .3 3.3l [.3 3 . \IIV.‘ 3 v H. .041" 34 w l' ‘ Mn“..- 333 M- ":33" W ' O .3333"). ' 3. . 3 I 3 ' "p 331"” ‘ 3. - 3 3“. . ( h “ It; ' :‘3‘3‘. 013 3 J 3“ 333‘ 3333 3 '3 3333 nl‘ "' l a} .3 ' 3‘! J (3 ’3'! 3 3 U o h .3 . .3 .lf. 33" I33 Nun-DI. 33333.. 333333333! ..L .5333." . 3 I 3 3', I... nil! 3r. r 3331.134 u.” ”U runlrm. |.. I I33 0 “J, 3.3! h '3. ll 3 '. 3 1" d...- J.‘ lull Ill! '3’ "a 3 3’ 3’ .I 3.3 3311' II. 3 3” 3.33 33 .|3l " I II. 31!".33 Ill. 3033. 333 1“".me 3 . 3. F333”: u.L3.“3h -3 3 3 l3 330.3 MI 3033.! n 3 H173 blfl'r. .«3 H 133 33.13133! n ..a3...333un.l.nJ ’ lil“ V I I .3: 33. ‘39”. 3 "33133313 .31.! 3"“ ' J ;.'3* '333'3333“ 3 n . ‘h - 1321.3 .3 1mm ' 1 ‘~ . :3 I. ‘ . 3 NM? 3.. 3 33’ 3|. 3 M .m-3L.u..‘3....3W . ‘ " '1 ' ”3‘uwu..hr‘un~l30hfil 3| - I‘. l3 31“.. .WJ' 333. W3 -3_«m..nu...:..3.w .3 3‘ W a’lumu flimut.'|5 .33 . l3 3 “W33. . 333 . .3 3:. . 3 3'33... 3.. ”’7...l riflJ... 3.3313 p1) . ...3..u3. . 3n 3 .b ..I ' I 3 . aqua-"1 j“: u. DJ. 3.5: .73; (5.31:3, K 3 MICHI IIIIIIIIIIIIIIIIIIIIIIIII 3129310668 341 L' “t [9% A %‘! h‘n' A" “I, “It B Q I . . .‘ ‘ a . "- '— 1.-$'--. lunar] g. f " '_ a: I ’ ' O 3 . b‘-—v 4—wqi5a’IGU-s“ W “ O .4 n 1.1:. Quin-“l: :3 In. 'l-au ring 51: This is to certify that the dissertation entitled ANALYZING THE HOUSEHOLD PURCHASING PROCESS FOR CABLE TELEVISION SELECTED VARIABLES AND HYPOTHESIZED RELATIONSHIPS presented by Ronald Paugh has been accepted towards fulfillment of the requirements for Ph.D. degree in Mass Media Z???” 71K Major professor Date April 29, 1986 MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771 MSU RETURNING MATERIALS: PTECe in book drop to remove this checkout from LIBRARIES ”- your record. FINES will be charged if book is returned after the date .? stamped below. mu; ‘5’“;‘izwre - i . LUN o 9 1 3 1: r53 02.11194 r ’“ 'fl .1 “‘5 1.9—}; fi‘. ANALYZING THE HOUSEHOLD PURCHASING PROCESS FOR CABLE TELEVISION SELECTED VARIABLES AND HYPOTHESIZED RELATIONSHIPS By Ronald Paugh A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Mass Media 1986 426- 330; Copyright by RONALD PAUGH 1986 Ronald Paugh was obtained. The data were analyzed using various bivariate and multi- variate statistical techniques. The results indicate that: (l) children, when present, do not significantly alter spousal joint, husband-dominant or wife-dominant cable purchase decisions, (2) a larger proportion of family members tends to be involved in the decision-making process when the respondent holds a more "contemporary" family ideology, (3) the cable purchase decision tends to be joint when the spouses are approximately equal on relative occupational status and (4) cable subscribers engage in more retailer and interpersonal search activities but non-subscribers tend to engage in more media search activities. The model presented in this research presupposes that the theories emerging from those disciplines in regard both to the distribution of power within households and to group decision-making may offer insights into joint purchase behaviors. This research indicates that relative spousal occupational status may be useful as a central construct in a theoretical development of power in the marital dyad. Moreover, house- hold consumer choices involve interpersonal relations and information searches as well as product-centered responses, and a fuller elaboration of the former dimensions may yield a better understanding of the dynamics of household decision-making processes. ACKNOWLEDGEMENTS I would like to thank the Chairman of the Dissertation Committee, Dr. Thomas F. Baldwin (Telecommunication), for his valuable assistance and inspiration in the completion of the dissertation and throughout the doctoral program. Thank you also to the members of the committee, Dr. Bradley S. Greenberg (Telecommunication), Dr. Bruce Vanden Bergh (Advertising), Dr. Gilbert D. Harrell (Marketing and Transportation Administration) and Dr. Donald A. Taylor (Marketing and Transportation Administration), for their insights and criticisms offered throughout this research endeavor. Thanks are due to Wanda Del Toro and Allen Harris for "loaning" me their students for data collection and coding purposes. This stage of the research process would have been infinitely more difficult with- out their assistance. I would like to thank the Department of Telecommunication at Michigan State University for their support in helping me to complete this work. The use of the conference room and the phone bank is very much appreciated. Ann Alchin typed this dissertation for which I am very thankful. Also, Ann has generally kept me in line and out of trouble during my residency. Finally, my thanks and appreciation are extended to those close to me who have offered inspiration, encouragement and support ii (psychological and financial) all along the way: to my parents Richard and Antoinette Paugh for their unending faith and guidance and the good example they have set; and to Jan and Kim Davis for the strength and succor they have given me. This work is dedicated to all of these individuals, for without their assistance, it would never have been completed. Also, this is dedicated to my late brother, Richard Jr., who I'm sure would have been proud of this accomplishment. TABLE OF CONTENTS Page LIST OF TABLES .......................... vi LIST OF FIGURES ......................... IX CHAPTER I. INTRODUCTION ........................ 1 Purpose of This Dissertation ............. 2 Contributions and Implications ............ 6 Summary Overview ................... 7 11. REVIEW OF FAMILY BEHAVIOR LITERATURE ............ 8 Theoretical Model .................... 9 Specification of the Model ................ l4 Component One: Inputs - Resource Variables ........ l5 Discussion and Rationale for Hypothesis l ....... l8 Component One: Inputs - Demand Variables ......... l9 Discussion and Rationale for Hypothesis 2 ....... 2l Component One: Inputs - Demand Variables ......... 22 Discussion and Rationale for Hypothesis 3 ....... 24 Component Two: Throughput - Consumer Search: Discussion and Deliberation ............... 25 Discussion and Rationale for Hypothesis 4 ....... 28 III. RESEARCH METHODOLOGY .................... 30 Introduction ....................... 30 Description of the Study: Final Questionnaire ...... 3O Selection of Sample ................... 34 Administration of the Survey ............... 35 Measurement Instruments: Hypothesis l .......... 36 The Joint Involvement Index ............... 38 Relative Spousal Resources ................ 40 Measurement Instruments: Hypothesis 2 .......... 42 Measurement Instruments: Hypothesis 3 .......... 43 Discussion/Deliberation Indicies ............. 52 Determination of the Discussion/Deliberation Indices. . 53 Analytical Techniques Used in Hypothesis Testing ..... 56 iv TABLE OF CONTENTS (continued) Chapter Page IV. RESULTS OF HYPOTHESIS-TESTING ............... 57 Introduction ...................... 57 V. DISCUSSION AND CONCLUSIONS ................ 85 Introduction ...................... 85 Summary of Key Findings ............... 85 Limitations of the Research ............. 92 Contributions of this Research ............ 93 Suggestions for Future Research ........... 97 Expanding the Consumer Satisfaction Concept ..... 97 APPENDICES ........................... lOO REFERENCES ........................... ll4 Table 4-5. 4-6. LIST OF TABLES Recent Research on Cable Television Audiences ...... Framework for Managerial Research Into The Family Consumption Process for Cable Television ........ Sample Disposition of Subscriber Groups ......... Distribution of Sample Survey Telephone Calls ...... Family Ideology Scale .................. Varimax Rotated Component Analysis Factor Matrix of the Levinson and Huffman Scale ........ Item-to-Total Correlation: Spousal Relationships Scale . Pearson-Correlations: Parent-Child Relationships Scale . Discussion/Deliberation Indices An Illustration ..... Relative Spousal Occupational Status Frequency Distribution ................. T-Test Results of Joint Involvement Index Relative Spousal Occupational Status .......... T-Test Results of Joint Involvement Index Relative Spousal Occupational Status .......... T-Test Results of Joint Involvement Index Relative Spousal Occupational Status .......... Relative Spousal Educational Attainment Frequency Distribution 3 ................ T—Test Results of Joint Involvement Index Relative Spousal Educational Attainment ......... T-Test Results of Joint Involvement Index Relative Spousal Educational Attainment ......... vi Page 3-4 33 37 45 54 6l 62 Table Page 4-8. T-Test Results of Joint Involvement Index Relative Spousal Educational Attainment .......... 62 4-9. Relative Spousal Income Contribution ........... 63 4-lO. T-Test Results of Joint Involvement Index Relative Spousal Percentage Income Contribution ...... 63 4-ll. T-Test Results of Joint Involvement Index Relative Spousal Percentage Income Contribution ...... 64 4-l2. T-Test Results of Joint Involvement Index Relative Spousal Percentage Income Contribution ...... 64 4-l3. Pearson Product-Moment Correlations of Differences in Relative Spousal Resource Variables With The Joint Involvement Index ..................... 66 4-l4. Influence Structure in the Cable Television Decision Prbcess - Original Purchase and Upgrades ..... 67 4-15. Influence Structure in the Cable Television Decision Process - Downgrades ............... 68 4-l6. Influence Structure in the Cable Television Decision Process - Disconnects .............. 69 4-l7. T-Test Results of Joint Involvement Index Family Versus Couple Households .............. 7l 4-18. Difference of Proportions Z-Test Results of Family Versus Couple Households ..... 71 4-19. Difference of Proportions Z-Test Results of Family Versus Couple Households ..... 72 4-20. Difference of Proportions Z-Test Results of Family Versus Couple Hoaseholds ..... 72 4-2l. Pearson Product-Moment Correlations of Child-Related Household Attributes with Cable Influence Structure . . . . 73 4-22. Age Distribution of Children in Cabled Family Decision-Making Units ............ 73 4-23. Summated Unweighted Average Scores: Levinson and Huffman Scale: Spousal Relationships Factor ........... 75 4-24. Difference of Proportions Z-Test Results of Contemporary Versus Traditional Spousal Relationships (Basic Service). . 75 vii Table Page 4-25. Summated Unweighted Average Scores: Levinson and Huffman Scale: Parent-Child Relationships Factor . . . . 76 4-26. Discussion Deliberation Indices Subscriber/Non-Subscriber Comparisons .......... 78 4-27. T-Test Results of Discussion and Deliberation Indices (Subscribers) ...................... 80 4-28. T-Test Results of Discussion and Deliberation Indices (Upgrades) ....................... 80 4-29. T-Test Results of Discussion and Deliberation Indices (Downgrades) ...................... 82 4-30. T-Test Results of Discussion and Deliberation Indices (Disconnects) ...................... 82 4-3l. Point-Biserial Correlations of Subscriber Types With Discussion and Deliberation Indices ........ 83 5-l. Pearson Correlation Matrix Spousal Resource Variables Female Characteristics ................. 87 5-2. Pearson Correlation Matrix Spousal Resource Variables Male Characteristics .................. 87 5-3. T-Test Results of Husband's Influence Structure: Female and Male Perceptions ............... 94 5-4. T-Test Results of Wife's Influence Structure: Female and Male Perceptions ............... 94 5—5. T-Test Results of Children's Influence Structure: Female and Male Perceptions ............... 95 viii LIST OF FIGURES Figure Page 2-l. The Deacon and Firebaugh Systems Model of Home Management ................... ll 2-2. The Managerial System Adapted from Deacon and Firebaugh: Cable Television ............ l2 ix CHAPTER I INTRODUCTION Traditionally, buyer behavior has been studied from the perspective of the individual consumer. One result of this orientation has been that most of our knowledge about consumer decision-making processes is limited to the behavior of individuals. Only during the last twenty years has a different unit of analysis received any attention. Marketers and researchers of consumer behavior have since come to recognize the importance of the family unit in buyer behavior research. Drawing upon concepts from sociology, psychology and economics, researchers have gradually gained a better understanding of the family as a consuming unit. Although the models and most of the consumer behavior concepts which have been used to explore the household setting recognize family influence, the primary focus has been the individual. The emphasis of family buying behavior research has been upon the decision-making process by which a household acquires goods and services. Overwhelmingly, the focus of family decision-making research has been upon: (l) determinants of husbands' and wives' involvement in the decision process and (2) the final outcome of the decision process, namely, "Who decides" (husband or wife). The emphasis on the deter- minants and outcomes of family decision-making behavior can be attri- buted to three major factors: l) The focus by researchers on the act of making a decision rather than on the process by which a decision is made, 2) Conceptual and methodological difficulties involved in developing and testing multiperson/family (husband/wife/ children) level variables, and 3) Problems associated with the collection and analysis of multiperson data (time, expense, techniques). Even with these problems, the family unit provides an important oppor- tunity to observe both individual and group buying behavior in action. The present study deals with an expanded exploration of the family decision-making process. The basic premise of the study is that, by focusing upon the family as the primary unit of analysis, the under- standing of buyer behavior will be enhanced. Purpose of This Dissertation Typically, cable television audience research has been quite limited. For example, a common focus of audience research has been on the description of viewership patterns (Jackson-Beeck, 1977; Jackson- Beeck and Robinson, 1981), and the specific behavioral tradeoffs either between alternative media (Hornik and Schlinger, 1981) or between alter- nate cable systems (Agostino, 1980). Other types of cable television audience research have attempted to discern differences among premium- pay subscribers, basic subscribers and non-subscribers of cable tele- vision (see Table 1-1 for a brief review of this research). This body of research exists as a small host of mainly empirical studies, most of which do not have a rigorous theoretical foundation, precise statements of testable hypotheses, nor an organizing framework or mid-range theoretical structure. mcaa23Ecv mac; .m.zu> .meoo ceuaaeou \omu.> we mcoczo .o xuo_n .m N p wououacu ppm: mmw_ . uaamsumcu ».c~_cg.uaagu soc: .ua_ttas . m? unoucou nurses ”on cu mcmovcumnam mmmummo— no.85om -co,m_uou xpasoe -coc cage a_mx.— egos “cw—way noummu «.1 uacwumam oemummop upozmmzo; xu,_wnunocn Rama—V use “an "oz was mcoaacomazm m_ncu -mcov “may-“ 0: ans .mco_uauumoxm mocupmo. co use: mumcomau «was» uoo.m uca Luxumm .mcma,tumn:m mpooo uwmoc cagu maucum owsocouo cozow; a co vco .cocc__zu once gov: Lamaze» agave: cw .ucoapbco aces acoacm-.gu ouoaa Ammmpv osvua.cumuv «oz one atom: opoou as; .ummu-u acoz mm..p u_ogem:og o>.uoacomocaoc gumcxuu co use: sauce: van cosaacx mcoa_cumn=m-ccc cozy mew—WEom ccocn— ecu Awmm_v agave: :. oeouc' cmgowz m>mg muwumwuoum . vcosmmo o>waawcumou "oz mconwcumoam o—aou u.cuwscceo-coz mcoz Noe mm>mzmmaoz m—asmm Eoucom uco pcp-mm mcmnwcumoaw -coc van mcmnwcumnam cowmw>o_ma apnea co mopxum sauces Amumpv ocauoc a? mmucmcwcepc umwu : -ocvp on» c_ oucmcoccvc uwuosmumxm muuocoevm uco m>vqucummu no: acouvacawm oz Aucuvzzuccoz acou*»pcmvm o: mv wcwc» omm czocxcs pocovucoaocm .Locacm .ucoxcpm ocvvpvam mmc.uc—u cane: o:c_czum» vuuwu» Amovmwmoguonxz z acmocoamoz wax» ~_aEcm Acooxv Acough\_uoo: mvmxpoc< xcme.ca Amvcoguaa pup upon» mmucuwua< coamv>upmp m—aou co zucoomom ucoumm I..." T covuuxnpmc use “coacaoucmucm Low >p mm: mcwnvcumaam o.n~u.m mpaou o» we.cumn=m cu »_mg.. mmm— ecu o_ume ucvca yo anew: x>owz.c wawcumoam ou xpmgwp egos ac: «to o.uoe ammo -uooca co «com: a>omz.n >h mpaau cu on_cumn:m o» xpmx_— a:_u_v:a awe. oco spams .ocm Ponce on use o>vu -coo co meow: >>omx.~ -umamcug ocamwo— >p opaou Co co,» mcwm: “soap cog» onvcumnam cu »_mxwp mmmp uco eve usage co meow: x>oox " neon: ugmwp cog» wn_cuma:m cu apex.p egos etc even: ammo -voocn mo mama: x>uox u : .mcmma usm_p cog» >h apnou o» ma.cumnam o» xpoxvp ages ago ave pocucou mo atom: Asuox " >p apnou o» mn.cuma:m uco .nco.uuc=» -aocu as“ upa.::, Ho: mama—mac ou wc.mou ozu u.n.zc' Amwmpv soc» 8686.8» mmoc wee; as“ ouwmuao co.un_~ccou .p': 85oz use ouvmuao upogmmaog o.aEnm «munguo: -mu mmwmzuoaxz muv>wuuo po.ucoumaam._ .ocoaam-_;u >uw>muoo ”otacz.vn=m up: muw co tam: Eoucou use uuoscoou: mupnuvco> :o_a naesmcou >» c, oucwcm» -cvo o: "co'ua.cumn=m >p-»ma uco oucoovmmc we guaca— cmozuon 8.;mco.u nope; mmco>cv nmcmnvcuw -aam son E:.E~ca as ou apox*_ mace ace mow>oe ugo30u cowuoucmwco cmuooca a wee coco Ammopv mung»: c. -Pvgu ocoe gu.3 mGPog mamxpmco upozmmaoz mpasom xu.cxum oco wsaugwcgmuo do: -mmao: «sous. cmgmmt acoc.sacom.: was: coo._ .0 one: xup—vnoaoca coeoacg .xuuao a:.u_.=m macvvc.m Loan: 0:3.c5um» woumaw Amevm.mozuonxz z unaccoamoa was» uchom Acomxv xcooz»\pocox m.nxpuc< accevca Amvcozua< Aa~,=.s=°u. .-. ..s.» According to many in the cable industry, more attention must be paid to the perceptions and needs of subscribers if success is to be achieved in an increasingly competitive video marketplace (Broadcasting, 1983; Advertising Age, 1984; Multichannel News, 1984). Therefore, the purpose of this research is to provide a unifying framework as a basis for organizing and identifying factors which influence the purchase and retention of cable television. Without such a structure, generation of hypotheses and empirical verifications occur in a vacuum without light. The bulk of the published research findings suggests that the cable subscribers have larger families than non-subscribers (Table l-l). It is, therefore, proposed that the unit of analysis shift from the indi- vidual to the family unit and that specific hypotheses be tested within this framework. These hypotheses will be gleaned primarily from the consumer behavior literature, which has its roots in psychological, sociological and economic theory. Moreover, the following four ques- tions are used to guide the direction of the present research: 1) To what extent do relative spousal contributions influence the joint decision-making process in the cable purchase decision(s)? 2) To what extent do children alter the influence structure of both spouses in the cable purchase decision(s)? 3) How does "traditional" versus "contemporary" family ideology, i.e., spousal relationships and parent-child relationships, affect the distribution of family decision influence among all family members in the cable purchase/retention decisions(s)? 4) To what extent do prepurchase planning and search activities impact on the family's decision to subscribe or not subscribe to the cable television service(s)? The answers to these questions outlined above will serve as the objec- tives of this research. If these objectives are accomplished, a better understanding of the process of family decision-making will be achieved and increased knowledge about the individual consumer will be obtained. Contributions and Implications Now that the technology is pretty much in place, the cable industry has become more market driven. A high "churn" or disconnect rate has plagued the industry (Paul Kagan Associates, 1981), and myriad attempts to rectify the situation have been undertaken. Maintenance marketing (Baldwin and McVoy, 1983) or retention marketing programs (Broadcasting, 1983) attempt to reduce churn. Moreover, a distinction is now being made between "former“ and "never" non-subscribers of cable television. It, therefore, becomes evident that research should be undertaken which simultaneously advances the study of consumer behavior and allows marketers to better meet the needs of their current and potential customer franchises. The potential contributions of this research should be of interest to both researchers and marketing practitioners. Researchers must be- come more knOwledgeable about the ways in which family role structures affect decision-making processes within the family. By studying the family unit, researchers may discover: (1) better ways of conceptual- izing and operationalizing multiperson consumer studies, (2) new con- sumer constructs applicable to group buying behavior and (3) consumer variables which in turn can be used to study individual buying behavior. Frequently, practitioners attempt to determine the principal decision-maker for their particular product or service and use this information as a basis for determining market segments, promotional messages, product development plans and pricing strategies. The results of this research should be of particular importance to cable marketers. Specifically, knowledge of family members' preferences, measured by influence levels, and the extent of prepurchase activitiesirlevaluating cable television should aid this group in developing appropriate marketing and advertising strategies. Summary Overview A review of the major research in the area of family decision- making, drawing upon conceptual and empirical studies conducted in psychology, sociology, economics and marketing, is presented in Chapter 11. Knowledge from past research is used to aid in the develop- ment of a model which can be used to investigate hypothesized processes of family decision-making. The relevant family variables and their relationships are described and hypotheses are presented which are used as a basis for testing the model's relationships. Chapter III provides an overview and description of the research methodology employed in conducting the field study and the analytical techniques applied to the data collected. In Chapter IV, the data from the field study are analyzed. Conclusions, implications and suggestions for future research are presentedierhapter V. CHAPTER II REVIEW OF FAMILY BEHAVIOR LITERATURE Modern consumer behavior research is about three decades old, commencing with the motivation and personality research of the early 1950's. During the intervening years a substantial portion of research efforts in marketing and advertising have been focused on consumer behavior. Despite the relatively large amounts of effort focused on consumer behavior research, the results have been somewhat discouraging. In a 1971 review article (Kassarjian, 1971) the state of consumer re- search was termed “equivocal.” Kassarjian, in his review, went on to comment that researchers too often ignore the many interrelated influ- ences in the consumer decision process. Wind has argued that the following changes, among others, are required if the study of consumer behavior is to advance (Wind, 1977): 1. The unit of analysis: shift to the buying center. The concept of the buying center refers to all those members of an organization (household) who become involved in the buying process for a particular product or service. While the major buying roles (i.e., initiator, influencer) remain constant over all purchases, the participants can and do change over purchase types and categories. 2. The domain of interest: shift to purchase, usage, mainte- nance and product disposal. In other words, the research involved with purchase activities and roles needs to be expanded to include how the product/service is utilized, the amount of satisfaction/dissatisfaction derived from the use of the product/service and brand switching behavior associated with the product/service. There has been an increasing interest in research focusing on the household as the unit of analysis. In a comprehensive review article, Davis (1976) cites over one hundred bibliographic references, most of which relate to household decision—making behavior. Many other scholars have concluded that more research on household buyer behavior is needed (Burns and Granbois, 1980; Grashof and Dixon, 1980). Also, most recent texts in consumer behavior include more discussion of household behavior than do earlier versions (Assael, 1981; Hawkins, et al., 1983; Zaltman and Wallendorf, 1983). Theoretical Model The choice of the household as the relevant unit of analysis requires the construction of an appropriate household model. Such a model was suggested by Cairncross (1958), who visualized the household as a small factory in which various productive services are utilized to produce the desired output. Subsequent work which has viewed the house- hold as an organization engaged in production include Becker (1965) and Lancaster (1966). Contemporary marketing theorists who have viewed the household as a social system consisting of a set of individuals organized into defined roles and responsibilities include Sheth (1974), Beutler and Sahlberg (1980), Grashof and Dixon (1980), and Buss and Schaninger (1983). The Deacon and Firebaugh (1975) systems model of home management provides the theoretical context for the empirical analysis of this paper. The systems model furnishes a useful framework for identifying and analyzing the relationships between/among pertinent managerial variables in the consumption process. The family is viewed as managing tasks that meet both external requirements of society and the internal needs and demands of family members by planning the use 10 of resources and implementing the plans to meet demands. The basic model is comprised of inputs, throughputs, outputs, and feedback, as shown in Figure 2-1. Selected aspects of the systems model relating directly to cable marketing are shown in Figure 2-2. Alternatively, Figure 2-2 can be transposed into tabular form to facilitate concept- ualization of the model (Table 2-1). The use of a five phase decision process differs somewhat from the classic conceptualization (Engel, Blackwell, Kollat, 1978) which in- cludes a phase of alternative evaluation. Alternative evaluation is concerned primarily with the factors that are used by consumers to evaluate alternative products and the relative importance of each factor. Within this process of alternative evaluation are evaluative criteria, or the factors and standards utilized by the consumer in evaluating the products or brands available. Also, there are beliefs, which are the pieces of information that relate a given product or brand to evaluative criteria. The alternative evaluation phase is eliminated because it is so intimately related to the search process. The search process relates to the sources of information that are utilized by the consumer to reach a decision and the influence of each source on the consumer. The first step in this process is an internal search within the individual's conscious, sometimes referred to as the central processing unit, to determine if the problem can be solved given current information and experience. If internal search fails to yield the required information, then external search will likely occur. The consumer may consult personal sources, such as friends or mass communications that are general in nature and are not controlled by marketers. The consumer may also actively seek or be more open to ll mmULJOmmL flmmD xuanumma Amnmpv gmamnmcwm use coummo mpcmscocm>cm I I I l I I I I I I I I I I I I l I I I I I mucmemu pm: Ami muaapzo pawsmmmcmz use: mo pmuoz msmpmam gmamnmcwd can coummo mg» mmucaomwm Amvcomcwa psnzmzoczp Acmucaom .P-N mesmst ”muczom mmuczomom mucmEmo muzacH 12 aumccoumwo. e cmsuppcu mo mum ocm consac. smopomuw «Po; xmm .mmmzoam. awe: mcwxme mccmEmo mmumcmczoo. mmumcmgs. xuwuum wmmzuczgmcav cowmwumu AFFEmw mamcm> m—aaou. cowgocmnvpmu\cowmm=umwu. mccmpumg mcwxmsncopmvumo Amlllll. mzosocoaam msmcw> pcwon. ATIIL mmucmapwcw »_wsmm. cmnvcumnam awn E=wEmca mmmmmooca mcwxmzucowmwumo cwawcumpam mpnmu uwmmg. mmmgucam accusaom megmmmcmz mpzapao I. hsaxwzomzh m acmcogsou N “cocoaeou momsogm Co weoucw w>wpmpmc. mmmaoam mo magnum pmcowpmasuoo m>wpmpmc. momsonm mo covpmuscm m>Pumpmc. mmuczommm whaazH F pcmcogsou cowmw>mpmk mpnmu “smsmnmcwu use coummo sect cmuamu< Emumxm mecmmmcmz och .~-N mesmwd 13 Table 2-1. Framework for Managerial Research Into the Family Consumption Process for Cable Television Cable Television A B C D E Stage in the Process Members Variables Method of Variables Involved Affecting B Conflict Affecting 0 Resolution Initial Mention - (Problem Recognition)a Source(s) of Information (Search) Discu5510n/De11beration bsee below Purchase (Choice) .basic cable subscriber .premium pay subscriber Evaluation (Outcomes) .upgrade .downgrade .disconnect aTerms in parentheses are taken from EHQEI: BIaCkWEII and Kollat, 1978’ p. 21. bIn an attempt to simplify a complex process, it is assumed that potential conflict, and hence the use of a conflict resolution strategy occurs only during the discussion/deliberation phase of the decision process. m denotes areas not specific to the present research l4 marketer-controlled sources such as advertising and personal selling. This search process will continue until the consumer is satisfied that enough information has been gathered. Several researchers have sug- gested that consumers actually evaluate information simultaneously with search (Claxton, Fry and Portis, 1974; Davis and Rigaux, 1974). The discussion/deliberation phase is added because it views behavior as a process rather than as a discrete act and is concerned with how a deci- sion is reached as it is with the decision itself (Engle, Blackwell, Kollat, 1978). Whereas most consumer behavior models have attempted to specify the exact network of buying decisions, their interrelationships and determinants, the proposed family decision-making model (Figure 2-2) focuses on the identification of the major sets of variables that could affect family purchase decisions, specifically, the cable purchase decision. Recognizing that no deterministic model can explain and predict accurately all_family buying decisions, the model provides a framework for the identification of variables and specification of hypotheses, reflecting the current state of knowledge, on the relation- ship among the key variables. As such, it serves only as a guideline for the selection of variables for one specific empirical study on family buying behavior with no attempt to specify general cause and effect type relationships. Specification of the Model The marketing and sociology literature presents a number of vari- ables that are hypothesized to affect the family decision-making process. Without a theory to indicate which variables should affect the family decision-making process, the researcher risks accepting spurious 15 results. In reality, no adequate theory has been developed to explain all aspects of the family decision-making process. Yet some generaliza- tions have been advanced to explain fragmentarily, selected aspects of this process. The purpose of this section is to introduce the primary variables of the proposed model (Figure 2-2), offer some hypotheses to be tested, and rationalize their presence. Component One: Inputs Resource Variables Blood and Wolfe's (1960) resource theory explains the distribution of power within the marital dyad by examining the comparative resources of the spouses. A resource is defined as anything (income and education for example) that one partner may make available to the other, helping the latter satisfy his or her needs or to satisfy his or her goals. Based on this theory, "the balance of power (in decision-making) will be on the side of the partner who contributes the greatest resources to the marriage" (Blood and Wolfe, 1960, p. 12). Blood and Wolfe's Detroit study assessed many different product/service purchase decisions such as food, insurance and the type of house or apartment to take. It was found that influence of either spouse varied from product to product. Hence, it is dangerous to generalize from studies that construct role structures for general decision-making behavior. It is, therefore, necessary to measure the role and influence of family members for the marketer's specific product or service. Several authors, including Heer (1963), Rodman (1967), Safilios-Rothschild (1970), Burr (1973), and Cromwell and Olson (1975), have questioned the resource theory as being a sufficient explanation for decision-making within the household. However, in spite of the questions which have been raised relative to 16 the resource theory, no author has completely dismissed resources as having no influence on decision-making within the marital dyad. Resources also include such specific factors as time available for decision-making. For example, wives have been found to lose influence vis a vis their husbands during the child-rearing stage of the family life cycle (Blood and Wolfe, 1960; Sheth, 1974). The reason appears to be the combined effects of having to care for children and of having to give up activities outside the home. Because wives have less time to be involved in decision-making and less to contribute in terms of financial resources and information, they become more dependent on their husbands during this period. Support for this resource theory is mixed. A positive association between education and participation in decision-making has been found by several researchers (Zaltman and Wallendorf, 1983). In examining the wife's contribution to family income, her occupational status, and the number of hours per week she is employed, one study found that all of these variables had a positive association with the wife's participa- tion in decision-making (Hempel, 1975). In a similar analysis, however, another study found that there was not a statistically significant difference between the influence levels of employed and nonemployed women (Zaltman and Wallendorf, 1983). However, this study did not take into account the wife's position relative to her husband's or the employed wife's type of job. Both of these have been shown to be important determinants of influence and consumption patterns (Schaninger and Allen, 1981). The only conclusion one can make from this research is that whether a woman is employed does not seem to explain differences in l7 household decision-making structures. Rather, the relative positions of the husband and wife appear to be a determining factor. Thus, current research does not disconfirm the resource theory. Several studies have shown that the greater the importance of a specific buying decision to the household, the more likely it is to be jointly made by all members of the household. Moreover, it has been demonstrated that joint participation by husband and wife will vary directly with the degree to which household members directly engage in the use of the product. These research studies have focused more on durable goods purchases than frequently purchased items. At one ex- treme in terms of length of the decision process, amount of delibera- tion and financial importance is the housing purchase. A number of studies have shown, not surprisingly, a high degree of joint decision- making in buying homes (Davis and Rigaux, 1974; Hempel, 1975; Munsinger, et al., 1975). The automobile purchase has been another popular arena in which to study joint decision-making (Wilkes, 1975; Shuptrine and Samuelson, 1976) as well as home furnishings (Davis, 1970; Green and Cunningham, 1975; Woodside, 1975). Other product categories for which similar product data are available include small appliances, home entertainment (i.e., TV and stereo), cameras, life insurance, vacation travel and watches (Davis and Rigaux, 1974; Green and Cunningham, 1975). There have been no published studies of joint decision-making related specifically to the purchase of cable television. It has been documented, however, that cable households tend to be larger and watch significantly more television than non-cable households (Cabletele- vision Advertising Bureau, 1985; Table l-l). It is obvious that 18 cable television is an important household purchase decision, given the degree to which all members of the household view and the fact that very few households are willing to give up a month's television viewing for pay (Jackson-Beeck and Robinson, 1981). Given this information, and the current status of Blood and Wolfe's "relative contributions" theory, the following hypotheses need to be tested: Hla: A purchase decision is more likely to be joint when the relative position of the husband and wife on occupational status is approximately equal. Hlb: A purchase decision is more likely to be joint when the relative position of the husband and wife on educational attainment is approximately equal. ch: A purchase decision is more likely to be joint when the relative position of the husband and wife on dollar contribution to total family income is approximately equal. Discussion and Rationale for Hypothesis 1 This hypothesis assumes that it is the spouse(s) (if there is/are chi1d(ren) present in the family) who make(s) the ultimate purchase, upgrade, or downgrade decision, when cable television is the purchase being analyzed. This assumption is tenable given various research findings (Filiatrault and Ritchie, 1980). The hypothesis tests Blood and Wolfe's (1960) resource theory which asserts that the distribution of power within the marital dyad can be explained by examining the relative resources of the spouses. Based on this theory, the balance of power in decision-making will be on the side of the partner who contributes the greatest resources to the marriage. This hypothesis will be tested by the construction of (1) a joint involvement index and (2) measures of both spouses' occupational status, educational attainment and dollar contribution to household income. Each will be operationalized more fully in the Methodology chapter. 19 The joint involvement index assumes that it is the spouse(s) who make(s) the ultimate purchase, upgrade, or downgrade decision and that the purchase of cable television is a process rather than a discrete act. A question will measure the structure of influence across the over- all cable choice decision process, utilizing the constant sum scale (Tull and Hawkins, 1980). Different questions will be asked, dependent upon the subscriber type, i.e., basic subscriber, upgrade or downgrade. The influence structure will be measured by asking the respondent to estimate the relative influence of the husband, the wife and the children as a group (if present) across each of the designated decision elements. Component One: Inputs Demand Variables The last decade has witnessed a substantial change in the family role structure and family decision-making which is, in turn, reflected in the marketplace. Role structure describes the manner in which in- fluence across elements of the decision process is distributed among members of the family decision-making unit. These role structure changes affect existing products, services and marketing practices. Marketers must remain abreast of the impact of these family role changes and their concomitant consumer purchasing patterns in order to modify marketing and advertising strategies accordingly. As discussed earlier, practically all research has centered upon the husband-wife dyad (Davis, 1970, 1971, 1976; Cunningham and Green, 1974; Davis and Rigaux, 1974; Hempel, 1974; 1975; Cox, 1975; Munsinger, Weber, and Hansen, 1975; Wilkes, 1975; Woodside, 1975; Shuptrine and Samuelson, 20 1976; Hempel and Tucker, 1980; Wortzel, 1980) and has omitted the impact and involvement of children in the family decision-making process. This apparent void is slowly being filled, however. Researchers have studied parent-child interaction in a field setting (Atkin, 1978), and the influencing role of children in the decision to purchase various products (Berey and Pollay, 1968; Ward and Wackman, 1972; Szybillo and Sosanie, 1977; Jenkins, 1979; Nelson, 1979). One study yielded inter- esting results on this topic. In a study of vacation and lodging decisions, differences were found in the decision-making patterns of families (couples with children) and couples (no children). Husbands tended to dominate decision-making more in the family decision-making unit (DMU) than in those where no children were present; joint decision- making was more prevalent in couple DMUs. This effect of children on the decision-making structure is particularly interesting given a second finding. Within family DMUs children exerted relatively little influence on the overall decision process, though the extent of their influence varied substantially across different elements of the decision process. It appears that children may have the potential to influence family decisions by forming alliances with either the husband or wife to produce a "majority" position (Filiatrault and Ritchie, 1980). One other study is worthy of mention at this point. Jenkins (1979) collected data from 105 husband—wife couples which focused on the perceived role of children in family decision-making in various product/service areas. Some of the research findings were that children are perceived as being instrumental in some product subdecisions, to the point of being "child dominant," and that older children, especially teenagers, are perceived to have more influence. 21 It is apparent that there is a relative dearth of research which incorporates children into the family decision-making process in general and into the cable purchase decision in particular. It is, therefore, suggested that the following hypotheses be tested. H2a: The presence of child/ren will shift the influence pattern of the adults in a purchase decision to an influence pattern of less dominance by husband or wife. H2b: The older the child/ren, the more perceived influence (on the part of the adult respondent) that the child/ren will have in a purchase decision. Discussion and Rationale for Hypothesis 2 Based on the research conducted in the family decision-making area, three conclusions can tentatively be put forth: (1) family member in- volvement studies need to be product specific to be of value to the planning and implementation of marketing and advertising strategies, (2) older children tend to exert more influence in purchase decisions, that is, there is a positive relationship between age and parental yielding, and (3) children can and do exert an influential role in salient, product-specific purchase decisions. It is possible that the child(ren) may exert influence in the purchase of basic cable in order to view MTV or Nickelodeon, or certain premium services like Disney, given that the average child views over 30 hours a weeks (Banks, 1980). It is, therefore, hypothesized that children will exert influ- ence and possibly even dominate in some of the purchase and upgrade decisions. The primary purpose of this hypothesis is to add knowledge of children's influence in the purchase of cable television, which will aid in the ability to generalize across product/service categories. Further, this will contribute to the relative dearth of studies which 22 explore the child(ren) as an influencing agent in the purchase of products and services. Component One: Inputs Demand Variables The nature of household decision—making has traditionally been characterized in terms of distinct role perceptions, role definitions, and role performances by family members. Each family decision has been analyzed and classified on the basis of specific decision-role responsibility for both husbands and wives. During the seventies, significant changes occurred in the role perceptions and role structure of the family. Later marriages, decreasing family size, changes in wives' employment status and reasons for employment, and changing attitudes of both men and women toward proper roles for husbands and wives all challenge predictions based upon earlier research (Roberts, 1981). These changes, and particularly shifts in sex-role perceptions and the roles portrayed by husbands and wives within the household, suggest a transition away from male-dominated and female-dominated family decisions and a move toward shared roles and joint decision making (Haas, 1980). It is held, therefore, that the concept of sex role is central to the process of modeling family decision-making. Sociologists have discussed men's and women's normative sex-role attitudes, and scales to measure these attitudes have been developed and administered in various empirical situations (Osmond and Martin, 1975; Brogan and Kutner, 1976). Unlike earlier conceptions of masculinity-femininity, which tended to confuse sex-role preference, sex-role adoption, and sex-role identity (Brogan and Kutner, 1976), 23 these normative attitudes refer to what people consider appropriate male and female behaviors. Early developers of normative scales emphasized a type of equality-inequality continuum, but such scales now tend to emphasize roles of both sexes in terms of a "traditional" (sharply dichotomous role for males and females) versus a "modern" (role sharing between the sexes) continuum (Osmond and Martin, 1975; Rosen and Granbois, 1983). The precise meaning of the term sex-role is often left undefined but generally it is based upon one of three commonly employed concep- tualizations: (1) sexual gender (Holter, 1970; Nielsen, 1978), (2) equality-inequality of rights (Meier, 1972), and (3) preferences, norms and attitudes (Araji, 1977; Scanzoni, 1977). Sex role orientation for this study will be used as an all-encompassing term that subsumes such terms as sex-role attitudes, values, opinions, behavioral standards, and cultural norms and constraints. Regardless of which approach or definition is utilized to measure the sex-role construct, the result has been a sex-role typology conceptualized as a continuum, the extreme ends of which are sex-role traditionals and sex-role moderns (Tomeh, 1978). Sex-role traditionalists exhibit attitudes consistent with past conceptions of the male-provider and female-housekeeper roles. Decision-making within such a family is dominated by the husband, while the wife and other household members take a more subservient role in the family decision-making process. Alternatively, sex-role modern family members are identified by their more equalitarian role percep- tions. Household members in this group tend to share decision respon- sibility on a more egalitarian basis. One of the earliest studies examining the effect of sex-role per- ceptions on family purchase behavior was conducted by Green and 24 Cunningham (1975), who found that sex-role perceptions were related to the number of decisions that were husband-dominant, wife dominated and joint. They concluded that there is a trend toward more joint decision making. More recent studies also suggest that sex-role perceptions have an influence on decisions related to "who prepares dinner" (Roberts and Wortzel, 1979), nondurable purchases (Schaninger, Buss and Groves, 1982), durable purchases (Qualls, 1981, 1984), and money management (Ferber and Lee, 1974; Rosen and Granbois, 1983). These studies support the contention that traditional theoretical paradigms of family decision- making may no longer be sufficient to explain and predict household decision and purchase behavior. Further, while the majority of these studies have investigated the impact of sex-roles on family decision outcomes, very few studies have attempted to explore their effect on the process of family decision-making (Davis, 1976; Qualls, 1984). The present study attempts to fill this void by examining the impact of spousal sex-role orientation on the joint decision process for the cable purchase/retention decision by husbands, wives and children. The fol- lowing hypothesis will, therefore, be tested. H3: Spouses (adult household members) who are more sex-role "contemporary" will involve a larger proportion of house- hold members in a purchase decision than those spouses (adult household members) who are more sex-role "traditional." Discussion and Rationale for Hypothesis 3 There is considerable support in the consumer behavior literature that sex-role orientation has an impact upon the family decision-making process. Scanzoni (1977) suggests that changing sex roles and how they are perceived by family members have a tremendous impact upon family decision-making processes. The precise nature of these shifts 25 in role perceptions and role behavior and the extent of their effects on family decision-making processes has remained unclear. A perusal of the relevant literature indicates, however, that "contemporary" sex- role norms are found in couples characterized as younger, with more education and couples with higher social status (Davis, 1976; Ericksen, Yancey and Ericksen, 1979; Filiatrault and Ritchie, 1980) and the ten- dency is for sex-role "contemporary" family members to perceive a more egalitarian distribution of family decision influence and thus increased interaction between family members (Scanzoni, 1977; Spiro, 1983; Qualls, 1981, 1984). These changes and findings in sex-role orientation would suggest the need to add this variable to models of family decision- making. Component Two: Throughput Consumer Search: Discussion and Deliberation Studies of family decision-making have generally focused on the outcomes of the decision-making process such as the decision-making structure in the family and/or bases for role differentiation, rather than considering the process that has led to these outcomes. The result is that very little is known about how families reach decisions. As noted by Davis (1976) in his comprehensive review of the household decision-making literature, research is needed to explain how families make decisions rather than simply who is involved. It should be noted, however, that from a marketing management standpoint, it is still important to determine family influences at each stage of the decision process. Consumer search has a strong research tradition in marketing theory, influencing virtually all variables in the marketing mix. 26 Past studies have included the number of informational sources from which information was sought (retailers, various kinds of advertising, and interpersonal sources), the amount and types of information sought, the time dimension over which information was sought, the number of brands for which information was sought and deliberation occurred, and the manner in which the information was sought. Many studies have been undertaken on the consumer's use of differ- ent sources of information. In particular, the nature and extent of retailer search have been studied by numerous researchers including Dommermuth (1965), Bucklin (1966), Udell (1966), Thorelli (1971), Newman and Staelin (1972), Claxton, Fry and Portis (1974), Newman and Lockeman (1975) and Kiel and Layton (1981). These studies range across numerous product categories and most indicate that the majority of consumers visit three or fewer stores during search (Newman, 1977). Fewer pub- lished reports exist on the use of advertising as a source of informa- tion seeking. Bucklin (1965), Udell (1966) and Thorelli (1971) found that the percentage of respondents reporting use of advertisements ranges between 15 and 30 percent. Interpersonal communication has been widely studied in many research projects and is held to be a major source of market information. In most previous studies in which com- parisons were made of the use of different sources of information, interpersonal sources were named more frequently than were mass media sources (Lazer and Bell, 1966; Udell, 1966; O'Brien, 1971; Thorelli, 1971). Another dimension of the search process is the time interval during which search occurs. Two consumers may be identical in terms of retailers visited, advertisements seen or heard, etc. but be 27 different in aggregate search behavior if one consumer undertakes this search activity in a short period of time (a week) and the other takes a much longer period of time (a year). Studies by Kleimenhagen (1966- 67) and Newman and Staelin (1971) show that consumers differ consid- erably in decision and search times and that the time intervals differ according to the product studied. A third dimension of search behavior is the number of brands about which the consumer sought information. Dommermuth (1965), and Newman and Staelin (1972) included some measure of brand deliberation among other measures of external information seeking. The rationale for doing so is that the number of brands for which information seeking occurs is an essential element of search not covered in the other two dimensions of external search behavior. Findings from the above studies consisted of tabulations of buyers' responses to individual questions which were of interest not only for their own sake but also because they frequently were regarded as proxies for total prepurchase information search. To reflect more fully the amount of search, several researchers constructed more comprehensive measures. Katona and Mueller (1955) awarded from zero to two points for answers to each of three questions to form an index of "extent of information seeking" for purchasers of major household appliances. More recently, Newman and Staelin (1972) and Newman and Lockeman (1975) employed indices for out-of-store and in-store information seeking for automobiles and shoes and combined them into an overall index. In summary, three dimensions of information seeking have been proposed: (1) a sources of information dimension, (2) a brand 28 dimension, and (3) a time dimension (Kiel and Layton, 1981). For purposes of this research, the sources of information dimension is deemed most critical and has been delineated into a retailer search (number of cable employees talked to, visit to cable office to inquire about services, etc.), media search (number of commercials and/or advertisements heard or seen), and interpersonal search (discussions with friends, family members, etc.). This tripartite classification is not meant to cover totally all possible dimensions of consumer behavior, but owing to the nature of cable television (one supplier in a franchise), the most pertinent. Given this somewhat abbreviated review of the consumer search literature, the following general hypotheses will be tested. H4a: The greater the family discussion and deliberation (search activities) in evaluating a product/service, the more likely a favorable decision to purchase the product/service. H4b: The greater the family discussion and deliberation (search activities), the greater the likelihood of a repeat purchase of a product/service. Discussion and Rationale for Hypothesis 4 These two general hypotheses will be tested within the specific context of the cable purchase/retention decision. Given this specifi- city, two sub-hypotheses within each will be empirically verified. In hypothesis 4a, the two sub-hypotheses with respect to the purchase are: (l) the decision related to the purchase of the basic cable . service and (2) the decision related to the upgrading, or the addition of a premium service(s). In regards to hypothesis 4b, the two sub- hypotheses with respect to the repeat purchase are: (1) the decision related to the downgrading, or the discontinuance of a premium service(s) and (2) the decision related to the discontinuing, or 29 complete elimination of the cable television service. All four of these hypotheses will be specified in the Methodology chapter. Intuitively, these hypotheses are logical. When family members become more aware of each others' interests and desires for different programming, the latent demand for a wider variety of programming should manifest itself. This hypothesis has support from theory too. The marketing and social psychology literatures suggest that consumers are willing to take greater risks after informal group discussions than before such discussions (Beal and Rogers, 1957; Cox and Rich, 1964; Wallach, Kagan and Bem, 1964; Bem, Wallach and Kogan, 1965; Woodside, 1972). A consumer's peer groups, reference groups, and signi- ficant others offer support and legitimation to one's purchase choices (Menzel and Katz, 1955; Venkatesan, 1966), implying that group inter- action enhances individual risk-taking dispositions. This "risky- shift" effect is subject to a variety of alternative theoretical inter- pretations, but the one that seems to have taken hold in the litera- ture is the diffusion of responsibility hypothesis (Wallach and Kogan, 1965; Kogan and Wallach, 1967). This hypothesis states that individuals are willing to assume greater risks in a group context because responsibility for failure cf a risky course can be shared with others. Moreover, the origin of the risky-shift phenomenon appears to lie in "emotionally tinged interpersonal connections and attempts at influence which inhere in face-to-face discussion" (Wallach and Kogan, 1965, p. 17-18). The family, therefore, seems to be the natural setting in which this is likely to occur. CHAPTER III RESEARCH METHODOLOGY Introduction This chapter provides an overview of the research and analytical procedures employed in the present study. It is divided into two major sections: (1) a description of the research methodology utilized and (2) a discussion of the operationalization of the hypotheses to be tested and analytical techniques used to interpret the data. Briefly stated, a telephone sample survey design yielding 488 completed inter- views was used in this cross-sectional study. All interviewing was conducted from a central supervised location by trained interviewers. The survey data were analyzed using a statistical package on the CDC Cyber 750 mainframe computer. A series of analytical techniques, including t-tests, factor analysis and Pearson correlation was utilized to interpret the data. Description of the Study: Final Questionnaire The overall objective of this research is to test selected hypotheses which can be placed in two basic categories: (1) prepurchase search activities and their relationship to the purchase, upgrade, down- grade, disconnect or nonpurchase of cable television and (2) specified familial factors related to joint decision-making within the decision- making unit. The rationale for these hypotheses has been articulated in chapter II of this report. Moreover, it is important to note that 30 31 all the published cable research has been directed toward activities and behaviors gftgr the purchase decision has been made. This research is unique in that it addresses prepurchase activities and behaviors specific to the cable decision, albeit in a post hoc manner. The majority of the items on the final questionnaire were designed to collect the necessary information to test the hypotheses. Some questions, designed to measure the price sensitivity of selected premium services, were included. This was done in exchange for the subscriber list, and has no direct bearing on the primary purpose of the research. The first set of questions included in the final instrument was the screening protocol. Only the male or female head of the household in couple or family decision-making units was interviewed. "Couple" in this situation is defined as a man and a woman, eighteen years of age or older, living in the same household. No attempt was made to deter- mine if the couple was married or not married. "Family" is defined in the same manner, except that children, aged one year through 17 years of age or older were present. Although it is not likely that a one-year old child will have substantial influence potential, this is more appropriate than deciding, somewhat arbitrarily, the age at which the child(ren) should be included. Moreover, while collecting the data for this study, several respondents stated that the Disney pay service was purchased specifically for children as young as one and two years of age. It is feasible, therefore, that although a direct influence from the child is not present, an indirect influence may be operating. One further screening device was utilized to determine if the household had definitely decided not to subscribe to cable, if not already a subscriber. If the decision to subscribe was still "pending," 32 the household was only asked the pricing questions and not the questions related to any specific hypotheses. In total, there were 20 such house- holds excluded from the analysis, yielding a total sample size of 468. A final screening question relating to length of subscribership was also implemented. Because many of the search behaviors and influence struc- tures may be considered somewhat trivial and easily forgotten by the respondents, the interviews in the study were conducted shortly after the cable purchase was completed. The average time between purchase and interview was four months and no respondent was interviewed if they were a cable subscriber for more than six months. The second set of questions was designed with the purpose of categorizing the respondents into two primary groups, those of cable subscribers and confirmed non- subscribers. The interviewers were then routed to specific questions, dependent upon the respondent type. These classifications are de- scribed in Table 3-1. The "pending" non-subscribers were excluded in the analytic treatment of the hypotheses tested. The third set of questions was designed to measure how the nature of influence structure in the decision-making process differs across decision-making units of varying composition (Dunsing and Hafstrom, 1975; Ferber, 1975). The decision-making units in question were families (husband, wife, and child(ren))and couples (adult male and females or husbands and wives only). The substantive area of interest concerned the influence structure of the basic cable service and all premium services that were applicable to each household. The responses were obtained from only one of the male or female head of household, and every attempt was made to generate a response ratio of fifty per- cent men and fifty percent women. This was very nearly achieved, as 52 percent of the respondents were women and 48 percent were men. 33 Table 3-1. Sample Disposition of Subscriber Groups Subscriber Group Group Size Percenta 1. Premium subscribers (those who initially and currently subscribe to the basic service plus at least one premium service, like HBO or Disney) 107 22.9 2. Upgrades (those who added a premium service after initially subscribing to the basic service) 55 11.8 3. Downgrades (those who are still cable subscribers but who dropped a premium service after initially subscribing to the basic service) 60 12.8 4. Basic subscribers (those who currently and always have subscribed to the basic service only) 99 21.2 5. Disconnected premium subscribers(those who are no longer a subscriber but did at one time subscribe to the basic service plus at least one premium service) 21 4.5 6. Disconnected basic subscribers(those who are no longer a subscriber but at one time subscribed to the basic service only) 22 4.7 I. Non-subscribers (those who have definitely decided not to subscribe) 104 22.2 8. "Pending" non-subscribers (those who have not definitely decided about the cable purchase decision) 20 -- TOTAL 468 100.1b aPercentages are based on 468 cases, as the "pending" non-subscribers group was excluded from any hypothesis testing. bDoes not add to 100% due to rounding error. 34 The fourth set of questions was designed to ascertain prepurchase search activities of cable subscribers and non-subscribers. This sec- tion was designed so as to stimulate the respondents' memory. Many details of prepurchase behavior may be forgotten or overlooked, so this section was somewhat detailed, with a mixture of dichotomous and open- ended questions. A second set of questions is also included in this section. It is a scale which measures family orientations on a tradi- tional-contemporary continuum. These questions are a Likert scale format, that isaffive-point scale, with the following alternatives, ranging from "strongly agree," "agree," "neutral," "disagree" and "strongly disagree." The final set of questions pertains to the classification or demographic information. The primary questions were number and ages of children, spouses' educational attainment, occupational status and income contribution to the family. These are open-ended questions that were coded by the researcher and trained interviewers. A pretest was administered according to the guidelines offered by Hunt, et. a1. (1982) and a few ambiguous questions were rewritten. The entire finalized version of the survey instrument is reproduced in Appendix B. Selection of the Sample A subscriber list was secured through United Cable Television of Mid-Michigan, Incorporated. This list contained the addresses of all subscribers and non-subscribers in their franchise area (East Lansing and Meridian Township). Telephone numbers for cable subscribers were listed but not for disconnected cable subscribers and non-subscribers. 35 These telephone numbers were secured through Bresser's Directory. Also included on the subscriber list was the installation and disconnect date. Only those who had been subscribers for six months or less were considered in the sampling frame. This was done to better ensure accuracy of recall of prepurchase behavior and activities. A listing of 1,585 qualified households was in the sample after screening out those who were subscribers for more than six months. A total of 114 unlisted telephone numbers occurred among cable subscribers. These numbers were not included among the total sampling frame and no attempt was made to reach these households. Several general conclusions have emerged from the various studies which have examined the differ- ences in the characteristics of listed and unlisted households (Glasser and Metzger, 1975; Blankenship, 1977). One conclusion is that single parent households are more likely to have unlisted phones relative to households with both parents (Tyebjee, 1979). Therefore, no attempt was made to reach these households, since the proportion of eligible house- holds would probably not be large in relation to the total sample. Administration of the Survey Telephone interviews were conducted by trained student interviewers over a six-week time interval (March 15, 1985 - April 30, 1985). These students were participating in the study for credit as part of a class project in a research methodology course and a consumer behavior course. All interviewers received a written set of instructions in addition to an oral briefing on telephone interviewing techniques (see Appendix A). The interviewing occurred in a supervised, central location. Calling sessions were scheduled from 6:00 p.m. to 10:00 p.m., Monday through Thursday. Some calling took place on the weekends and the weekdays 36 from 10:00 a.m. to 5:00 p.m. This was done to maximize the at-homes and minimize the refusals (Kerin and Peterson, 1983). Interviewers were instructed to let the telephone ring at least six times before coding the attempt as "no answer." All telephone numbers yielding results of "no answer" or "busy" were called back at least five times. The distribution of the sample survey telephone calls is presented in Table 3-2. The noncontact rate and refusal rate are important measures of data quality. The noncontact rate of 8.4 percent is sub- stantially below the median noncontact rate of 39.1 percent for tele- phone surveys. Moreover, the refusal rate of 24.6 percent is somewhat below the median refusal rate of 28.0 percent (Wiseman and McDonald, 1979). Finally, the non-response rate of 39.1 percent is somewhat high, since it is desirable to achieve a non-response rate of from 30 to 35 percent. Overall, however, it would seem that the administrative attempts to control data quality were achieved. Measurement Instruments: ,Hypothesis 1 The hypotheses to be tested were stated in Chapter II. For the readers' convenience, there are restated: Hla: A purchase decision is more likely to be joint when the relative position of the husband and wife on occupational status is approximately equal. Hlb: A purchase decision is more likely to be joint when the relative position of the husband and wife on educational attainment is approximately equal. chz A purchase decision is more likely to be joint when the relative position of the husband and wife on dollar contribution to total family income is approximately equal. This hypothesis will be tested by the construction of: (l) a joint involvement index and (2) measures of both spouses' occupational status, 37 o .Ao1mom .a .mmm— .sseEsmu mmmv mowues owes» ms.a=ssou s, umxopaem use: msowuvspmmu uwmz xpusmzcms» umoe use .Aomm_ .upesoouz use seemmszv msmsoeupuuesa msOEe mesa» omega so» sovuvsrmmu umuamuue apsoseoo use as we uses» .apuswssaue ap.mm u .hpm AmmrnvmrPmsfi+mmmswm=m+ms_xsozsozv-osess msppaEem _euoh mpm u msoeuesmssmh + mpemzmmm + uwuueusou uoz u ouem mmsoammmusoz . w_m_ msonezz mmmswmam-mswxsozsoz-msess mswbaEem Peach fie Pm u som— u mzmv>smus~ umumpasou + msoeuespesm» + mm—nrmepmsn + mpemasmm www u mpemswmm + mzww>smus~ umumbusou seem uueusou no.o~ u amp mpemammm u even Femawwm fies - £3. - use: 3:33 38 . BNP .. £533.35 umumpmsou + 2235.53 + mmpfflpmsm + 33.3mm u emuem mmsofimm e.oe was Anemonsag mvmapese eueu so» umm: use mumsosmms xm>sam owes: musmusonmmsv mzm+>smus. umumpssou so smasaz Aev o.m me Amwmoasaa m_mx_ese eueu so» use: on posseu zme>smuse sews“ .upamms e m< .mueswaoou smmso— as ou muwumu zm_> -smus? msu osvsau asses mace we use .msomumoao x~>s=m mo muses so msovummac uses so moss» smzmse as: musmusosmms opa_aepmv msorues.ssmu so sensaz Amv a.me Rpm Amsomummaa xm>szm use so» mpuwmpposw as on uszos use: msorummzo us'smmsum cu msmzmse s_esu mo upzmus e we on: musmusoamws peeusmuoav muposmmaos e_awmepms_ so sense: Amy «.mp amp Axm>s=m es» s, mueavu.usea o» sesame uposmmaos smsuo xse uwzoppemmu use muesvuvuses on use—pets: me: useusosmmsv mpemawms uposmmsos mo smaeaz Apv ~.ms ~o~.p umuueusou senses Peach .m o.~ Fpp Amusemuue zme>swuse m>wm «weep we "xeu msouau soeuumppou eueu use o» sovss uwsuems «as uposmmaosv museusousos so sensaz Amy m.m em msmnsas msosampwu uposmmaossos\mmmspm=o es smaszz Amy e.Pp omp Aumsmvmme xpusmssau nos mp umpe_u cessasv senses wsosam—ma asexsoxsos e we omaeuma uueusou cu wpnes: smpe=z APV m.m~ mum uueusou o» apnesa senses Fence .< o.oop mam.P Aumsmvmme use umuumpmm msmnE=s msosqmpmuv uses; asepaEem peach usmusms Nosezmmss uppeu msogawpmh am>sam mpssem mo sovuanesumwo .mum o—neh 38 educational attainment and dollar contribution, as a percentage, to total family income. The joint involvement index methodology is based on the assumption that it is the adult members of the household who make the ultimate purchase, upgrade or downgrade decision and that the pur- chase of cable television is a process rather than a discrete act. The constant sum scale methodology will be utilized to measure the structure of influence of all family members involved in the cable choice decision process (Tull and Hawkins, 1980). This technique re- quires the respondent to divide a constant sum, in this research 100, among all household members in proportion' to their perceived influence or importance in the cable choice decision process. For example, if the wife (or husband) was totally responsible for the disposition of a service, they would get the entire 100 constant sum attributed to them. If, on the other hand, they were equally influential in the disposition of a cable service, each would get 50 of the constant sum. When children are involved, the perceived constant sum would be allocated to them. This device extends the type of analysis possible, and it pro- vides a more appropriate measure for studies involving more complex decision-making units for which categorical ratings of dominance become unwieldy. All previous and current cable subscribers will be asked the familial influence structures for the basic cable service, and all the applicable premium services. The Joint Involvement Index The degree of spousal involvement will be measured by the JII (Churchill and Hanson, 1974). The proportion of the various decisions that were made jointly by the husband and the wife on the purchase, 39 upgrade and downgrade decision will be summed together to form the index. The formula for this index follows: JII = P8 + (PH/s . PW/S) where: JII = joint involvement index P8 = proportion of the decisions made jointly by both the husband and the wife PH/S = proportion of all the decisions that were made that were husband dominant PW/S = PFOPOrtion of all the decisions that were made that were wife dominant The JII possesses two desirable properties. It is zero if either the husband or wife made all the decisions because then PB = O and either P = O. The index will equal one if all decisions were H/S °r Pws made by both spouses since then PB = 8/8 = 1 (when there are eight possible decisions). Simply stated, the closer the JII is to one, the more joint the decisions. Three types of "dominance" in the decision are defined in this methodology: (1) husband dominant: the husband's influence is greater than the wife's and both spouses' influence is greater than zero. (2) wife dominant: the wife's influence is greater than the husband's and both spouses' influence is greater than zero. (3) joint decision: the influence of both spouses is equal and both spouses' influence is greater than zero. This sort of treatment may be perceived as a methodological weakness, as a 90-10 split is defined "dominant," as a 60-40 split would be. 40 A joint decision is a 50-50 split, whereas it is just as reasonable to define a 60-40 split as a joint decision. Having acknowledged this flaw, it was felt that this technique was more justified than arbi- trarily deciding where the cut—off point for "joint" and "dominance" would be. Relative Spousal Resources The final aspect of this hypothesis will be tested by determining spousal occupational status, educational attainment and percentage of total annual household income. Each will be discussed separately. Numerous devices have been used in the past to define social class or status. These techniques typically employ some weighted composite index composed of occupation, income and education (Miller, 1977). This study is not concerned directly with social status per se but with occupational status of both spouses. Occupational status or prestige has been defined in many ways. One of the definitive texts on this subject is by Reiss, et. a1. (1961). It appears that occupational prestige ratings are relatively stable (Hodge, Siegel and Rossi, 1964) and that the correspondence between different occupational status methodologies is quite high (Miller, 1977). Therefore, the measure of occupational status used should be the one that is the most current and the most comprehensive. The measure utilized in this study fits these criteria. The occupational classification codes and prestige scores for this study were taken from the 1975 Basic Background Items for U.S. Household Surveys (Social Science Research Council, 1975). The prestige scores assigned to occupations were taken from a rating system developed at 41 the National Opinion Research Center in 1963 in a project on occupa- tional prestige, directed by Robert W. Hodge, Paul S. Siegel and Peter H. Rossi (1964). This concept of prestige is defined as the respondents' estimation of the social standing of occupations. Specifically, occupational ratings were elicited by asking respondents to judge an occupation as having "excellent," "good," "average," "somewhat below average," or "poor" standing (along with a "don't know" option) in response to the item: "For each job mentioned, please pick out the statement that best gives your own personal opinion of the general standing that such a job has," (Hodge, Siegel and Rossi, 1964, pp. 288). This methodology of occupation prestige determination was adapted to the 1970 U.S. Census occupational codes by the Social Science Research Council. It is, therefore, a current and comprehensive listing of occupational status codes. This booklet has hundreds of occupations and their respective status codes, categorized into various occupational classifications, along with two additional codes -- "Graduate/Research Teaching Assistant," and "Homemaker/Housewife/Unemployed," coded 51 and 0, respectively. The other status codes ascribed ranged from a high of 82 for "Physicians, Dentists, and Related Practitioners" to a low of 14 for "Baggage Porters and Bell Hops." The final measurement phase of this hypothesis includes measures of spousal educational attainment and percentage income contribution to total family income. These are open-ended questions and the income question was verified by making certain that both percentages summed to 100 percent. Three segments will then be determined to test the hypothesis: (1) husband's occupational status equals wife's occupa- tional status, (2) husband's occupational status is greater than the 42 wife's occupational status and (3) husband's occupational status is less than the wife's occupational status. The same procedure will also be followed for relative educational attainment and relative percentage contribution to total family income. A series of t-tests will be administered to determine if any significant differences exist. Measurement Instruments: ,Hypothesis 2 H2a: The presence of child(ren) will shift the influence pattern of the adults in a purchase decision to an influence pattern of less dominance by husband or wife. H2b: The older the child(ren), the more perceived influence (on the part of the adult respondent) that the child(ren) will have in a purchase decision. This hypothesis will be tested by analyzing three primary variables: (1) the joint involvamnt index, (2) the total number of purchase decisions made within the decision-making unit, and (3) various child related variables. Each will be discussed in more detail. The joint involvement index (JII) has been operationalized under Hypothesis 1. A t-test will be administered to determine if the JII changes in any significant manner when the family respondents are compared with the couple respondents. The second set of variables represents the total number of pur- chase decisions made within the decision-making unit. The percentage of husband dominant, wife dominant and joint decisions within the family and couple groups will then be analyzed via a t-test to determine if any significant changes occur. Also, the average influence level of all decisions across all family members will be analyzed to determine where these changes,if any, occur. The third set of variables deals with child-related characteristics. Specifically, the number of children within the decision-making unit, 43 the average age of the children, the age of the oldest child present, the total number of purchase decisions made when a child was involved and the average influence structure of the child(ren) will be measured. The Pearson product—moment correlation between these child variables and the average purchase influence of the children will then be administered to determine if any relationship exists between the two variables. Some of these data can be taken directly from the question- naire and some must be created through the use of computer statements. Tables and charts illustrating these data will be discussed in more detail in the next chapter. Measurement Instruments: Hypothesis 3 H3: Spouses (adult household members) who are more sex-role "contemporary" will involve a larger proportion of house- hold members in a purchase decision than those spouses (adult household members) who are more sex-role "traditional." This hypothesis will be tested by: (l) measuring the extent to which the adult household member believes in a "contemporary" versus a "traditional" family ideology and (2) determining the proportion of people in the decision-making unit who were involved in the purchase, upgrade and downgrade cable purchase decisions. The scale developed by Levinson and Huffman (1954) was used to test fer "contemporary" versus "traditional" family ideology. Specifi- cally, the scale is designed to measure ideological orientations re- garding family structure and functioning. These orientations are placed on an autocratic-democratic continuum. The autocratic extreme is represented by various forms of "traditional family ideology" - viewpoints which inVolve an hierarchical conception of familial rela- tionships, emphasis on discipline in child-rearing, sharp 44 dichotomization of sex roles and the like. The democratic orientations tend to decentralize authority within the family, to seek greater equality in husband-wife and parent-child relationships, and to maxi- mize individual self-determination. The scale presents four primary areas of family life: (1) parent-child relationships; child-rearing techniques; (2) husband and wife roles and relationships, (3) general male-female relationships; concepts of masculinity and femininity and (4) general values and aims. The scale is in a Likert-type format, consisting of the following forced-choice alternatives: "strongly agree," "agree," "neutral," "disagree" and "strongly disagree." The scale has demonstrated sufficient internal consistency, and owing to its length (11 items), was deemed an ideal instrument. The scale is reproduced in Table 3-3. To provide a parsimonious description of "traditional" versus "contemporary" family ideology, a principal components multiple factor analysis, using the varimax criterion for factor solution, was per- formed on the 11-item Levinson and Huffman (1954) family ideology scale. The initial analysis of these items incorporated an eigenvalue-one criterion for determining the number of factors to extract. Factor analysis was chosen as the appropriate technique because of its expressed purpose of data reduction and summarization. The method allows the researcher to analyze the interrelationships among variables (i.e., scale items) in terms of their common underlying dimensions (factors). These factors may be considered the essential determining constructs representing a new set of variables which are defined solely in terms of the original dimensions. 45 Table 3-3. Family Ideology Scale Area I: Parent-Child Relationships; Child-Rearing Techniques .If children are told much about sex, they are likely to go too far in experimenting with it .A child should never be allowed to talk back to his or her parents, or else they will lose respect for them .A woman whose children are messy or rowdy has failed in her duties as a mother .If a child's unusual in any way, their parents should get them to act more like other children .There is hardly anything lower than a person who does not feel a great love, gratitude, and respect for their parents .The facts on crime and sexual immorality show that we will have to crack down harder on young people if we are going to save our moral standards Area II: Husband and Wife Roles and Relationships .Some equality in marriage is a good thing, but by and large the husband ought to have the main say-so in family matters .Women who want to remove the word obey from the marriage service don't understand what it means to be a wife Area III: General Male-Female Relationships; Concepts of Masculinity and Femininity .The most important qualities of a real man are determination and driving ambition .I would feel uncomfortable if my immediate supervisor at work were a woman Area IV: General Values and Aims .The family is a sacred institution, divinely ordained Source: Levinson and Huffman, 1954. 46 The varimax criterion for orthogonal rotation was selected because it maximizes the number of very high and very low factor loadings, thus providing the simplest factor structure solution. As such, varimax is generally accepted as the best analytic orthogonal rotation technique (Hair, et. al., 1979). Factors were labeled according to factor loading patterns of the scale items. The highest loadings were considered most important for the purpose of deriving labels. These factors, including their item loadings are presented in Table 3-4. Since there are two independent factors, two sub-hypotheses will therefore be tested in the next chapter. Once these factors were determined,the individual item scores (coded O for "strongly disagree", 1 for "disagreeJ'Z for "neutral," 3 for "agree" and 4 for "strongly agree") were summed and then divided by the number of items comprising each of the two factors, yielding an unweighted average factor score. The first factor, "spousal relationships," has three variables and the second factor, "parent-child relationships," has two variables. To assess the degree to which the items measure the same underlying con- cept, the test data for the first factor were subjected to a measure of internal consistency using Cronbach's (1951) coefficient alpha. The second factor consists of only two variables and could not be analyzed by this methodology. Instead, a series of Pearson correlations was conducted. Table 3-5 and Table 3-6 illustrate these data for the "spousal relationships" and "parent-child relationships," respectively. As indicated in Table 3-5, the item-to-total correlation coefficients for the first factor ranged from .45 to .57. Nunnalky(l967) indicates that coefficients of .50 to .60 are sufficient in basic research of 47 Table 3-4. Varimax Rotated Component Analysis Factor Matrix of the Levinson and Huffman Scale Factor 1 Factor 2 Spousal Parent-Child Variables Relationships Relationships . Some equality in marriage is a good .61024 .17131 thing, but by and large the husband ought to have the main say-so in family matters. . Women who want to remove the word .62423 .24728 obe from the marriage service don't un erstand what it means to be a wife. . The most important qualities of a .20238 .28760 real man are determination and driving ambition. . A child should never be allowed to .19243 .13325 talk back to his or her parents, or else they will lose respect for them. . I would feel uncomfortable if my .16956 .09860 immediate supervisor at work were a woman. . The family is a sacred institution, .56270 .04435 divinely ordained. . A woman whose children are messy or .04117 .63370 rowdy has failed in her duties as a mother. . If a child is unusual in any way, .18971 .60621 their parents should get them to act more like other children. '. There is hardly anything lower than .31314 .39269 a person who does not feel a great love, gratitude, and respect for their parents. '. The facts on crime and sexual .35302 .30466 immorality show that we will have to crack down harder on young people if we are going to save our moral standards. 48 Table 3-4. (continued) Factor 1 Factor 2 Spousal Parent-Child Variables Relationships Relationships k. If children are told much about sex, .17889 .36161 they are likely to go too far in experimenting with it. Sum of Squares . (eigenvalue) 3.93 1.16 Percent of Variance 35.7 10.6 Note: Underlined factor score coefficients in each column are used to determine "traditional" versus "contemporary" family ideology. 49 Table 3-5. Item-to-Total Correlation: Spousal Relationships Scale Item-Total Correlation Variable Coefficient a. Some equality in marriage is a good thing, but by and large the husband ought to have the main say-so in family matters. .57 b. Women who want to remove the word obey from the marriage service don't understand what it means to be a wife. .57 f. The family is a sacred institution, divinely ordained. .45 Alpha = .71 50 Table 3-6. Pearson-Correlations: Parent-Child Relationships Scale Parent-Child Relationships a b Scale Variable 9 Variable h Parent-Child Relationships Scale 1.00 r=.86 r=.81 n=312 n=312 p=.OOl p=.OOl Variable 9a 1.00 r=.39 n=3lO p=.OOl Variable hb 1.00 aA woman whose children are messy or rowdy has failed in her duties as a mother. be a child is unusual in any way, their parents should get them to act more like other children. 51 this nature. Consequently, the coefficients generated in this study appear acceptable for this scale, with the one possible exception being item "f" in Table 3-5. An analysis of Table 3-6, which utilizes the Pearson-correlation methodology to assess the scale, indicates that both of the items vary with each other, but not to the extent that they both vary with the entire scale. In other words, each item does contribute uniquely to the overall factor. It should be noted, however, that more items should be developed in future research. Finally, those households whose average score on each factor ranges from 0.0 to 2.0 are classified as "contemporary" and those households whose average score ranges from 2.01 to 4.0 are classified as "traditional." These data will be presented in the next chapter. The second part of this hypothesis will be tested by determining the proportion of household members in the family decision-making unit who were influential in the purchase, upgrade and downgrade of the basic service, Cinemax, Disney and H80. Each of these services will be analyzed independently. If children were involved in the decision, the respondent was then asked how many children were involved. A z-test of proportions will then be administered to the "contemporary" and "traditional" groups to determine if any significant differences exist. H4a: The greater the family discussion and deliberation (search activities) in evaluating cable television, the more likely a favorable decision to subscribe to cable television. H4b: The greater the family discussion and deliberation (search activities) on an upgrade decision, the more likely an upgrade decision will be made. H4c: The greater the family discussion and deliberation (search activities) on a downgrade decision, the less likely a downgrade decision will be made. 52 H4d: The greater the family discussion and deliberation (search activities) on a disconnect decision, the less likely a disconnect decision will be made. This hypothesis will be tested by (l) the construction of a dis- cussion/deliberation index and (2) administering t-tests to those groups who have had a chance to upgrade, but have not, downgrades and those who have had a chance hadowngrade, but have not, and voluntary disconnects and those who have had a chance to disconnect, but have not. Discussion/Deliberation Indicies These indicies will measure the interaction with three fundamental sources of information internal and external to the decision-making unit: (1) a retailer search, (2) a media search, and (3) an inter- personal search. The validity of composite survey-based measures of discussion/deliberation depends on the fidelity of respondent reports, the appropriateness of discussion/deliberation source weights, and the ability of the researcher to strucutre questions which reduce double- counting and address conceptual ambiguities such as active versus passive discussion and deliberation. As a result, the general level of the index scores may be understated because of the insensitivity of the discussion/deliberation measures and the inability of respondents to recall all prepurchase discussions and deliberation details. Moreover, other sources of bias possible in retrospective accounts are largely uncontrolled. It should be pointed out, however, that this study is concerned with relative variation in discussion/deliberation behavior rather than absolute level of discussion/deliberation behavior. Further, this methodology presupposes that the same kinds of discussion and deliberation occur for original purchasers and for those who decide to S3 upgrade, downgrade or disconnect at a later point in time. This may be perceived as a methodological weakness, but it seems to be a reasonable assumption. Determination of the Discussion/Deliberation Indices Three separate factors comprise the discussion/deliberation indices: (1) retailer search factor, (2) media search and deliberation factor and (3) interpersonal search factor. These three factors have four, six and five component variables, respectively. These component vari- ables are illustrated in Tablel3-7. All of the component variables within each factor are dichotomies and will be given a score of one for each activity that the respondent participated in prior to the decision to subscribe or not subscribe to cable television. Thus, the total possible maximum score for each factor is four, six and five, respec- tively. The average score for each factor will be determined and t-tests will then be administered to the four different sets of cable subscribers to determine if any significam:differences in prepurchase activities exist. Limits were imposed on the number of contacts with which respondents will be credited in each of the component variables. For example, no more than 10 advertisements in each of the four major print and broad- cast media will be recorded on behalf of the respondents. This is consistent with various advertising wearout studies wherein it is stated that only a small percentage of buyers claim participation above these levels (Greenberg and Sutton, 1973). Also, length of the household's discussion/deliberation period will not be included in the index be- cause of the weak association between time and effort. It is apparent 54 Table 3-7. Discussion/Deliberation Indices An Illustration Possible Response Maximum Source Yes No Score Retailer Search Factor 1. Household member(s) talked with salesperson l 0 l or other employee(s) of cable company 2. Household member(s) initiated the contact 1 0 l with the cable company 3. Had discussion(s) with household member(s) l O 1 after contact with cable employees(s) 4. Household member(s) visited cable office 1 O 1 to inquire about service(s) .___ TOTAL 4 Media Search and Deliberation Factor 1. Household member(s) has/have read information 1 O 1 articles about cable TV in consumer oriented magazines like Changing Times, Money or Consumer Reports 2. Household member(s) has/have compared the l O 1 programming of cable and non-cable channels in daily TV listings 3. Household member(s) read magazine ad(s) about 1 O 1 cable TV or cable program(s) in TV Guide 4. Household member(s) has/have seen TV 1 0 1 commercial(s) about cable TV 5. Household member(s) has/have heard radio 1 0 l commercial(s) about cable TV 6. Household member(s) has/have seen newspaper 1 O 1 ad(s) about cable TV TOTAL 6 Interpersonal Search Factor 1. Household member(s) talked with relative(s) l 0 1 or friend(s) about cable TV before deciding to subscribe/not subscribe 2. Household member(s) discussed and/or 1 O 1 expressed interest of cable TV programs 55 Table 3-7 (continued) Possible Response Maximum Source Yes No Score 3. Household member(s) has/have discussed 1 0 1 information seen or heard in ad(s) about cable TV 4. Have discussed the cable TV program 1 O l preferences of spouse 5. Have discussed the cable TV program 1 0 l preference of children ____ TOTAL 5 56 that the same amount of effort may be expended in a short and intensive deliberation period and in a long and casual one. Analytical Techniques Used in Hypothesis Testing Four inferential statistical procedures were used for the testing of research hypotheses: the t-test of significance or "test against the null hypothesis” (Sawyer and Peter, 1983, p. 126) for independent samples; the z-test of proportions; Pearson product-moment correlation; and factor analysis. The alpha level was set to .05 for all tests and although the hypotheses are directional, they will be tested by more conservative two-tailed statistical tests. The SPSS (Version 9.0) T-TEST, PEARSON CORR, FACTOR and RELIABILITY were used for the various statistical tests. The tabled results, with discussion follow in the next chapter. CHAPTER IV RESULTS OF HYPOTHESIS-TESTING Introduction The purpose of this chapter is to present the results of the statistical analyses used in hypothesis-testing. Each hypothesis is stated and then the data are interpreted in light of the statistical model used for the analysis. An alpha level of .05 was specified prior to data analysis. Hla: A purchase decision is more likely to be joint when the relative position of the husband and wife on occupational status is approximately equal. To test this hypothesis, a series of two-tailed t-tests was administered. The four subscriber groups in the analysis are the premium subscribers, basic subscribers, upgrades and downgrades, for a total potential sample size of 321. A smaller sample was obtained, however, since both sets of data for the husband and wife had to be present to test the hypothesis. Respondent refusal and the inability to classify the occupation explains this smaller sample size. Table 4-1 presents the frequency distribution of the relative occupational status of both spouses. One of the three hypotheses is supported by the data (Tables 4-2 to 4-4). When the occupational status of both spouses is approximately equal, the decision tends to be joint. Table 4-2 and Table 4-4 results are in the expected direction, but are not statistically significant. 57 58 Table 4-1. Relative Spousal Occupational Status Frequency Distribution Number Percentage Wife's Status Higher than Husband's 47 22 Status Approximately Equala 71 33 Husband's Status Higher than Wife's 96 45 TOTAL 214b 100 315 status "points" from equal. The measurement device utilized is technically an ordinal scale. It is possible to have very similar jobs but not the same status number. For purposes of data analysis, however, the scale is assumed to be interval. bNon-subscribers and disconnects excluded. Table 4-2. T-Test Results of Joint Involvement Index Relative Spousal Occupational Status Joint Involvement Index Standard a Independent Variable N X Deviation t Significance Job status not equal 143 .56 .46 -1.68 .088 Job status equal 71 .67 .43 Note: the larger the JII, the more joint the decision a2-tai1ed test of significance 59 Table 4-3. T-Test Results of Joint Involvement Index Relative Spousal Occupational Status Joint Involvement Index Standard a Independent Variable N X Deviation t Significance Wife's status greater than husband's status 47 .49 .47 Wife's status equals '2'13 '039 husband's status 71 .67 .43 Note: the larger the JII, the more joint the decision a2-tailed test of significance Table 4-4. T-Test Results of Joint Involvement Index Relative Spousal Occupational Status Joint Involvement Index Standard a Independent Variable N X Deviation t Significance Wife's status less than husband's status 96 .60 .45 Wife's status equals 1'09 “272 husband's status 71 .67 .43 Note: the larger the JII, the more joint the decision a2-tailed test of significance 6O Hlb: A purchase decision is more likely to be joint when the relative position of the husband and wife on educational attainment is approximately equal. To test this hypothesis, the same procedures were applied as in hypothesis la. Educational attainment was measured on an interval scale, asking how many years of formal education each spouse has com- pleted, rounded to the closest whole number. Therefore, a one year difference in educational attainment would constitute an unequal status on this variable. Table 4-5 presents the frequency breakdown for this variable. Overall, this hypothesis is not supported. The joint involvement index does not increase when educational attainment is equal, but in- stead tends to decrease. The results in Table 4-7 are most troubling because the joint involvement index is significantly lower when spouses are equal on educational attainment. These results are not easily explained. ch: A purchase decision is more likely to be joint when the relative position of the husband and wife on dollar contribution to total family income is approximately equal. This hypothesis was tested in the same manner as hypotheses 1a and lb. The percentage of total household income was measured on an inter- val scale. Again, a one-percentage difference constitutes an unequal contribution of spouses. Table 4-9 illustrates the relative income contribution of both spouses. This hypothesis is not supported by the data. The joint involve- ment index tends to be lower when spousal income is equal, contrary to what was expected. None of these differences are statistically signifi- cant, however (Table 4-10 through 4-12). 61 Table 4-5. Relative Spousal Educational Attainment Frequency Distribution Number Percentage Wife's Attainment Higher than Husband's 60 19.0 Wife's Attainment Equals Husband's 112 35.4 Husband's Attainment Higher than Wife's 144 45.6 TOTAL 316a 100.0 aNon-subscribers and disconnects excluded. Table 4-6. T-Test Results of Joint Involvement Index Relative Spousal Educational Attainment Joint Involvement Index Standard a Independent Variable N X Deviation t Significance Educational attainment not equal 204 .64 .43 Educational attainment 0.88 .384 equal 112 .59 .45 Note: the larger the JII, the more joint the decision a2-tailed test of significance 62 Table 4-7. T-Test Results of Joint Involvement Index Relative Spousal Educational Attainment Joint Involvement Index Standard a Independent Variable N X Deviation t Significance Wife's attainment greater than husband's 60 .72 .41 Wife's attainment 1'90 ‘050 equals husband's 112 .59 .45 Note: the larger the JII, the more joint the decision a2-tailed test of significance Table 4-8. T-Test Results of Joint Involvement Index Relative Spousal Educational Attainment Joint Involvement Index Standard a Independent Variable N X Deviation t Significance Wife's attainment less than husband's 114 .60 .44 Wife's attainment “0°17 ~355 equals husband's 112 .59 .45 Note: the larger the JII, the more joint the decision a2-tai1ed test of significance 63 Table 4-9. Relative Spousal Income Contribution Frequency Distribution Number Percentage Wife's Percentage Income Higher than Husband's 75 36.1 Wife's Percentage Income Equals Husband's 54 26.0 Husband's Percentage Income Higher than Wife's 79 38.0 TOTAL 208a 100.1 aNon-subscribers and disconnects excluded. Table 4-10. T-Test Results of Joint Involvement Index Relative Spousal Percentage Income Contribution Joint Involvement Index Standard a Independent Variable N X Deviation t Significance Income contribution not equal 154 .62 .43 0.41 .692 Income contribution equal 54 .59 .47 Note: the larger the JII, the more joint the decision a2-tai1ed test of significance 64 Table 4—11. T-Test Results of Joint Involvement Index Relative Spousal Percentage Income Contribution Joint Involvement Index Standard Independent Variable N X Deviation t Significancea Wife's income contri- bution greater than husband's 75 .60 .44 0.13 .901 Wife's income contri- bution equals husband's 54 .59 .47 Note: the larger the JII, the more joint the decision a2-tailed test of significance Table 4-12. T-Test Results of Joint Involvement Index Relative Spousal Percentage Income Contribution Joint Involvement Index Standard a Independent Variable N X Deviation t Significance Wife's income contri- bution less than husband's 79 .64 .42 -0.60 .552 Wife's income contri- bution equals husband's 54 .59 .47 Note: the larger the JII, the more joint the decision a2-tailed test of significance 65 Overall, this set of hypotheses is not supported by the data. Four possible explanations may account for these results, which will be discussed in the next chapter. Finally, in an attempt to assess the overall relationship between these three spousal resource variables and their impact on the joint involvement index, a Pearson correlation statistic was computed. These results are presented in Table 4-13. The coefficients for all of the differences in each of these relative resources variables are negative and thus in the hypothesized direction. These findings imply that a higher joint involvement index is associated with a low spousal dif- ference on these three variables. In other words, the more equal the spouses are on these variables, the more joint the purchase decision. None of these results are statistically significant, however, indicating that other factors are operative. H2a: The presence of child(ren) will shift the influence pattern of the adults in a purchase decision to an influence pattern of less dominance by husband or wife. To test this hypothesis, a series of two-tailed z-tests of propor- tions was administered to the four subscriber groups (premium, basic, upgrades and downgrades) for a total sample size of 321. A total of 159 family respondents and 162 couple respondents was obtained. Dif- ferent sample sizes were obtained for each of the particular pay ser- vices. If the sample size for any pay service was less than 20, it was excluded from any statistical test. Tables 4-14 to 4-16 portray the influence structure of both the family and couple decision-making units for the original purchase/ upgrade, downgrade and disconnect decisions, respectively. It can be seen in Table 4-14 that the husband is the dominant influence in the 66 Table 4-13. Pearson Product-Moment Correlations of Differences in Relative Spousal Resource Variables With The Joint Involvenent Index Difference Difference Difference Joint in Relative in Relative in Relative Involvement Occupational Educational Income Index Status Attainment Contribution Joint Involvement Index 1.0 r=-.O6 r=-.05 r=-.03 n= 321 n= 316 n= 208 p= .14 p= .19 p= .34 67' .umou museu.e.se.n peu.um.ueum owesosoo o» Ao~,sv ppesa sou o—ssem .o~.m s. opaesessou “as use mopszem as» use we ue.—sse as: umeu peu.um.ueum ..o.. .o—neu.—sse use ".<.zn .mo. 5 n .useu.».sm.u as: ".m.ze Ac . s. --- --- --- --- --- --- --- --- me e me .~ . s. .ossesu u..o: as»... A“. A. . s. --- --- --- e. we --. --. --- s s~ oe Am . s. s33s... Re. AA . sv --- --- --- N. we --- --- 0 --- me o~ Am Am 1 cs mass... Ame ANe . ss.m.z one. ems. e. em coo. coo. see. e, am Re Res 1 s. oesmpw A. . sv .<.z s.<.z .m.z om om .m.z .m.z .m.z an en Fe Ace . s. sasm.smmW As. . s. .m.z .m.z a.m.: an we see. Nos. Npo. e. on em Roe . s. eases.u~mw . . . . . . . . e we we Amm— . sv see, c. m 2 soo use me Am coo coo m.o F muessem 0.3ee use... ._s Am-~w Ae-_ml Am-ev Am. Me Ae-~. fie... A~-es and AN. A.“ . Axv As. Ax. :W fies Ass A1. .03 eessuagsm use: eseemas 3-: use: esemems, 8-3 8-: 30:1. susepesu see: esenaaz so seemsuoe 3.xpesec we so ousoapes_ eo «usuapes, so sac» ousoapes, susupesu so oussapwsp musou «usespws. s. s? noususuweeu mo eusoapus~ \uerzxuseauas as» usonmoc opasou .m» musousoa -mes xpwEe» es meusosoeeru .so e moususo»».u mo museupe+co+m reu.umeueum mpzuqzosmus “assoc ous.6,e.so.m peu.um_ueuw mhzwozonmmz >4_xopo» o—aeu es» s. osauuasum ouseapps_ cpae ape-h 68 .mc.xe .eseeweeseem ees ”.m.ze .emee eeseeeeesmem _eewemeeeem eeesesse ee Aom-sv __eEm ecu opseem ”--e Amuse -- -- -- em em a -- -- -- e eop e A. u av possesu ee>ez as»... _ s Acusv -- -- -- -- -- a” -- -- -- o om em a_ u sv oeeuxesm... m_ “my 3.5 -- -- -- em 8 _ 1. .1 I. -- - -- 8 u 5 323:... _ 3 5.5.9: .3. a... 3 mm : 08. So. a.m.z o 8 8 SN - 5 8:3 2. A 2-5 -- .1 8 8 m. o z 8 A... u 5 seems... : E Aeusv -- -- -- em we ”w -- -- e -- o em ee Am u sv eros_u... H“ A». 3-2 9-3 .3 Gd 3 J AZ 2; 3-: m N _ sz Axe A v sz are H. v v Mew Mzw Mzw ”seswuseemeu one: ueenmez 134:, we?) usenmem “ Tunn- u-z 3am. smsqusur- ewe: useemon- ea se_mrucu m.>—¢Ee» we we ouseepws_ N we eeseepes_ - we see» museepus_ sesu—esu we eusoapusp musmu mesmepmsr s, " cw mouswsmeewu ee mesmepus~ \ou'3\useem:z on» -seemes opeeeu .m> museusee -mws a—esem we mousocovau weseewv_so,m we museuee'smmm peewumeueum .eumemwueem mmeseseeepu ee mhzmonQmum meezou ouseu.e.seem —eupem*ueum uwu sevuseeese veg: s~ "seempuoueem mhzmonemmm >4—:opoh opeeu use s, oseeuesum mesmepms_ m—-u upper 69 .mo-x e .aseu_mvsm_m «es ".m.ze hep-cw .m.z .m.z e.m.z me mm ooo. coo. ope. e Fe mm flaw n sv eup>som opeee one «emssoem_u... eye 3-: 3-: 3-: 3% 3 2-3 3-: 3-: E E 3 ”3V Axe sz Axe Rev Axe Axe m»_3 useemzz 3-: we.) usemmem u-n v-2 ”Hep swsmpwmu esp: usenme: “on sorm_uuu m_»FVEe~ - 1 so ee muses—~s_ we ouseepvse we seox muses—es. swLuF.;u we eusoepws_ mesou mesw:_es_ :9 se muesesneeru ee asse:—»s~ \mew:\useemez use useemws upeeou .w> wusausoe mweswseem—u we euzeummwsmrm reoruuwueum uwu :evuseeece «ex: ea -mws x—WEee we meusmsee._u museure_sa—m ”sovmruuueem so eeseuewpsewm .eu_umpeeem _eepumeeeem mhzmonQmmx ue820u mhzuezOAmuz >e_:(u muuosmmmMMm mmouess sewmruue seem+>o—eh ope-u use s. useuuesum mucus—us— w—-u w—neh 70 original purchase and upgrade decision. The husband's influence in- creases significantly for the couple respondents in the basic and H80 purchase decision. The wife and children tend to have smaller influ- ences in the cable purchase decision, with the exception Of the Disney channel. The downgrade and disconnect decisions tend to be more evenly distributed among the husbands and wives. The children tend to have small or no influence in these decisions. Tables 4-17 to 4-20 test the hypothesis. These tables show the change in the joint involvement index, the percentage of husband- dominant decisions, wife-dominant decisions and joint decisions made in the family and couple decision-making units. Overall, this hypoth- esis is not supported by the data. H2b: The older the child(ren), the more perceived influence (on the part of the adult respondent) that the child(ren) will have in a purchase decision. This hypothesis was tested by means of a Pearson product-moment correlation coefficient. Various child attributes are correlated with the perceived adult influence level across all the services. Again, the four subgroups of basic, premium, upgrades and downgrades were used to test the hypothesis. The results indicated in Table 4-21 support the hypothesis. The average age of the child, the oldest child present in the household and the number of children in the household are all positively associated with the average child influence. The age distribution of the children in cabled families is presented in Table 4-22. In terms of the number of decisions in which the child was involved, 22.6 percent were involved in one decision, 15.7 percent were involved in two decisions and 1.6 percent were involved in three decisions. 71 Table 4-17. T-Test Results of Joint Involvement Index Family Versus Couple Households Joint Involvement Index Standard Independent Variable N X Deviation t Significancea Couple Households (no children) 161 .64 .45 0.71 .576 Family Households (children) 159 .60 .44 a2-tai1ed test of significance Table 4-18. Difference of Proportions Z-Test Results of Family Versus Couple Households Husband Dominant Decisions a Independent Variable N (Percent) 2 Significance Couple households (no children) 161 28.2 -0.02 .842 Family households (children) 159 28.3 a2-tailed test of significance 72 Table 4-19. Difference of Proportions Z-Test Results of Family Versus Couple Households Wife- Dominant Decisions a Independent Variable N (Percent) t Significance Couple households (no children) 161 8.7 -l.12 .272 Family households (children) 159 12.5 a2-tailed test of significance Table 4-20. Difference of Proportions Z-Test Results of Family Versus Couple Households Joint Independent Variable N Decisions a (Percent) 2 Significance Couple households (no children) 161 63.2 0.74 .441 Family households (children) 159 59.2 a2-tailed test of significance 73 Table 4-21. Pearson Product-Moment Correlations of Child-Related Household Attributes with Cable Influence Structure (Number = 159; Correlation c0efficients(r) and probabilities(p) are shown) Average Influence Child Attributes of Childa Average age of child r = + .33 (X = 8.43 yrs.) p = .001 Oldest child in household r = + .40 (X = 10.0 yrs.) p = .001 Number of children in household r = + .34 (X = 2.20 children) p = .001 aaverage child influence, when present in household, is 18.13. Table 4-22. Age Distribution of Children in Cabled Family Decision-Making Units (Number = 159) Age Category Number Percent 1-5 years 114 32.6 6-12 years 112 32.1 13-19 years 87 24.9 over 19 years 37 10.4 TOTAL 350 100.0 74 Approximately 60 percent of the purchase, upgrade and downgrade deci- sions were made without any influence from the children. As was noted in Tables 4-14 to 4-16, children tended not to be influential in down- grade and disconnect decisions. H3a: Spouses (adult household members) who are more sex-role "contemporary”in their spousal relationships will involve a larger proportion of household members in a purchase decision than those spouses (adult household members) who are more sex-role "traditional." H3b: Spouses (adult household members) who are more sex-role 'tontemporary'in their parent-child relationships will involve a larger proportion of household members in a purchase decision than those spouses (adult household members) who are more sex-role "traditional." To test these two hypotheses, a two-tailed difference of proportions z-test was administered to the premium subscribers, basic subscribers, upgrades and downgrades. Only the households with children were selected in the testing of the hypotheses. Also, it may be possible that "traditional" households have a larger number of children on the average than "contemporary" households. As seen in Tables 4-23 and 4-25, however, the two household typologies are similar on this charac- teristic. Nevertheless, the proportion of household members involved in the basic cable purchase/retention decision will be analyzed to account for this difference in household size. The difference of proportions z-test was administered to the basic service subscribers only. The z-test was only performed on the spousal relationships factor, because the sample size for "traditional“ house- holds was too small on the parent-child relationships factor (Table 4-25). Moreover, tests on the other pay-services could not be administered because of small sample size, making inferences impossible. The results of the difference of proportions z-test for the spousal relationships factor (Table 4-24) are not statistically significant, but 75 Table 4-23. Summated Unweighted Average Scores: Levinson and Huffman Scale Spousal Relationships Factor Average Number of Numberof Household Range Households Percent Children Typology 0.0 to 2.0 (strongly disagree to disagree) 106 67.5 1.94 "Contemporary" 2.01 to 4.0 (somewhat agree to strongly agree) 51 32.5 2.22 "Traditional" Table 4-24. Difference of Proportions Z-Test Results of Contemporary Versus Traditional Spousal Relationships (Basic Service) Proportion of House- hold Members a Independent Variable N Involved 2 Significance "Contemporary" spousal relationships ideology 106 68.1 1.22 .115 "Traditional" spousal relationships ideology 51 58.2 a2-tailed test of significance Table 4-25. 76 Summated Unweighted Average Scores: Levinson and Huffman Scale Parent-Child Relationships Factor Average Number of Number of Household Range Households Percent Children Typology 0.0 to 2.0 (strongly disagree to disagree) 140 89.7 2.04 "Contemporary" 2.01 to 4.0 (somewhat agree to strongly agree) 16 10.3 1.94 "Traditional" (0318' Sir? I 1- 77 the proportion of household members involved in the basic cable purchase decision for the "contemporary" households is in the hypothesized direc- tion. It does appear, therefore, at least from the results Of this small sample, that this construct may be a useful addition to other studies conducted in the household decision-making context. H4a: The greater the family discussion and deliberation (search activities) in evaluating cable television, the more likely a favorable decision to subscribe to cable television. TO test this hypothesis and the following three hypotheses, a series of two-tailed t-tests was administered to six groups of current and former subscribers. The current subscriber group consists Of the premium subscribers, upgrades, downgrades and basic subscribers; the former subscriber group consists of the disconnected premium-pay sub- scribers and the disconnected basic subscribers. This entire group constitutes a sample size of 364. The confirmed non-subscriber group sample size is 104. Before discussing the results Of the hypothesis test, it is useful to discuss the findings presented in Table 4-26. This table illustrates the mean responses of the subscribers and non-subscribers for the three search factors. It is evident that significant differences emerge in the retailer search factor and the interpersonal search factor-. Sub- scribers appear to be more heavily engaged in these activities than non-subscribers. It may be hypothesized that this group may have higher than average cable information needs, because of an above average prod- uct interest, and a positive attitude toward gathering information in the product area which consequently lead to greater search during the Prepurchase decision period. In other words, the subscriber group appears to be more favorably predisposed to the product of cable 78 ouseu.e.se.n .8 any“ uo_.eu- .uouepus. use usoeecemeem uoueosseumeu ..o.. .msoe_suwe:m guise; use acua.sunaan ucussau nous—us.“- .m... 8. 7 3... 8.. 3 99:3 €533.52: es be}. .m pee. ~_.~ u~o we“ Ame» av Assess: u—egomee; asoEe «seguese opeeu es gaseous. es se.uuauu.e A. .m.z cm.e umm ea Ame» a. sosu—.gu sup: nousosopose >» .0 seanuauu.a an eoe. mo.m «on roe Ame» a. unseen cu.) neusosouese >» »e so.»m:uu.e .~ spo- om.~ ue~ wen Ame» av usages: up.see usage mu. .8 se.mmeua_e .— Leueeu :useom pecansoeuuus_ .m.z m..—- ~m.m am.v ueos nu. sequent»: eo cones: Ac .3. 3s 3.. :2 2.2 3:28.83 22: so secs: 3 .3. Se :5 New so: 2:828 t .0 sues: .e .m.z oo..- om.¢ mo.n ueoc oueeu >» s. mu. yo Lucia: An .m... 3.... a; a... 3...» a. 3.5:. >» mpveu s. ossssesaoce o—eee-ses\epeeu useeseu A~ .m.z eo.o- pn.n m~.~ sopm.>o_oe o_eeu eeoee ueos mms.~emee ueusoese soeemseu ee sense: A, seeded se.uesee,.ee use seseom e_uo: coo. o~.m aw um— Ame» rv moup>sem ueeee oceeac. ea ou.~»e o—aeu u.u.> A. 08. 2e- so A: 7% 3 3&2?» e23 5:. eeeesoe Levee «sees»: u—ogemeos su.: sewnmeum.e Am 8o. 8.? Se “2 E 33...... 29.88.. 3 83:5 33:8 3 e8. a.m- a: o: 8 8:3 3:29; .23 S :22. : segue; seseem sop.euo¢ :5 a: Sen-5 esoseu_e.se.m e ”see.sme:m-soz emsoaflmwmeem msemeseeeeu Leeecumeem-sezxsoepsemeem noe.uc_ se.uesoe..oe ceeuueumve mN-v open» 79 television. It appears that non-subscribers do attend to the mass media, particularly in terms of the non-marketer controlled, consumer oriented magazines like Consumer Reports. Table 4-27 shows the results of the search activities for the subscriber and non-subscriber groups. The search activities have been delineated into three primary categories; retailer search, media search and interpersonal search. It appears that subscribers tend to engage in more retailer search and interpersonal search and non- subscribers engage in more media search activities. These findings tend to corroborate the findings in Table 4-26, but the hypothesis is not totally supported. The finding that non—subscribers engage in more media search is both troubling and perplexing. It is troubling in that it implies a relative ineffectiveness of the various promo- tional campaigns. It is perplexing in that, if, as previously asserted, cable subscribers are more favorably predisposed to cable television, an understanding of how this predisposition is engendered is crucial. Perhaps more resources should be allocated to free-trial offers, as is so common in the personal care products industry, in an attempt to stimulate this favorable predisposition. Further, it may be that promotional campaigns are more useful as a maintenance marketing or retention marketing tactic. H4b: The greater the family discussions and deliberation (search activities) on an upgrade decision, the more likely an upgrade decision will be made. The t-test results for this hypothesis indicate that upgrades do not engage in more discussion and deliberation than non-upgrades. Specifically, Table 4-28 shows that upgrades engage in significantly more media search, but not retailer and interpersonal search activities. The hypothesis is therefore not generally supported. 80 Table 4-27. T-Test Results of Discussion and Deliberation Indices (Subscribers) Retailer Media Interpersonal Search Search Search Subscriber Type N X X X Subscriber 364 2.54 2.21 2.41 Non-subscriber 104 2.28 2.87 2.18 t-value 3.02 -4.88 1.71 (alpha level) (.003) (.000) (.088) Note: 2-tailed test of significance Table 4-28. T-Test Results of Discussion and Deliberation Indices (Upgrades) Retailer Media Interpersonal Subscriber Type N Search Search Search X Upgrade 55 2.53 2.47 2.55 Non-upgrade 206 2.54 2.01 2.31 t-value -0.10 2.03 1.30 (alpha level) (.920) (.043) (.195) Note: 2-tai1ed test of significance Note: Non-upgrade group consists of those groups who have had a chance to upgrade, but who have not. The non-upgrade group thus consists of premium subscribers (n=107) and basic subscribers (n=99). 81 It may be that more media and interpersonal search contribute to experimentation with premium services. The theoretical explanation for these findings, i.e. the diffusion of responsibility phenomenon and/or latent demand among family members being stimulated, appears not to be supported. H4c: The greater the family discussion and deliberation (search activities) on a downgrade decision, the less likely a downgrade decision will be made. The results of this hypothesis in Table 4-29 are not supportive. Downgrades tend to engage in more search and deliberation activities, specifically media search activities, contrary to expectations. This finding may be explained by the methodology employed. It is implicitly assumed that the same types of search and deliberation activities occur for the basic and premium services alike. Although this may be a reasonable assumption, it is a limitation which must be acknowledged. H4d: The greater the family discussion and deliberation (search activities) on a disconnect decision, the less likely a disconnect decision will be made. The results in Table 4-30 do not lend support to the hypothesis. It appears, however, that disconnects do engage in discussion and deliberation activities. Again, there may be a weakness in the methodology employed. Although respondents were asked if they volun- tarily discontinued the service, it may be that they were in fact dis- connected from the service for non-payment and/or illegal hook-up. Also, it may be that this subgroup disconnected because of a move, and have not yet subscribed to the service because of time and/or financial constraints. It appears that the disconnect subgroup is actively engaged in discussion and deliberation activities prior to subscribing to cable 82 Table 4-29. T-Test Results of Discussion and Deliberation Indices (Downgrades) Retailer Media Interpersonal Search Search Search Subscriberglype N Downgrade 60 2.52 2.51 2.65 Non-downgrade 107 2.48 1.97 2.34 t-value 0.25 2.22 1.52 (alpha level) (.803) (.028) (.130) Note: 2-tailed test of significance Note: Non-downgrade group consists of those groups who have had a chance to downgrade, but who have not. The non-downgrade group thus consists of premium subscribers (n=107) Table 4-30. T-Test Results of Discussion and Deliberation Indices (Disconnects) Retailer Media Interpersonal Search Search Search Subscriber Type N X X Disconnect 43 2.53 2.39 2.36 Non-disconnect 321 2.54 2.18 2.41 t-value -0.01 0.82 -0.32 (alpha level) (.995) (.413) (.752) Note: 2-tailed test of significance Note: Non-disconnect group consists of those groups who have had a chance to disconnect, but who have not. The non-disconnect group, therefore, consists of premium subscribers (n=107), upgrades (n=55), downgrades (n=60) and basic subscribers (n=99). Table 4-31. 83 Point-Biserial Correlations of Subscriber Types With Discussion and Deliberation Indices Subscriber Types Subscribers Upgrades Downgrades Disconnects Search Indices (n=468) (n=261) (n-l67) (n-364) Retailer Factor .104a .040 .011 .032 Media Factor -.110a .038 .175a .087a Interpersonal Factor .094 .081 .155a -.042 asignificant at the .05 level 84 television. Assuming that all 43 households in the disconnect subgroup were voluntary disconnects, the reason(s) for disconnecting need(s) to be ascertained. Once these reasons are determined, appropriate market- ing and/or service strategies can be implemented. As a summary measure and a test of the overall strength of the relationship between the four subscriber types and the search indices, point-biserial correlations were conducted. Point-biserial correlation is appropriate when one of the variables has two "x" values, each accompanied by an array of "y" values. In this specific case the two "x" values are subscriber/non-subscriber, upgrade/non-upgrade, downgrade/non-downgrade and disconnect/non-disconnect. The "y" values are the ranges for the individual search indices, as discussed in the previous chapter. These data are presented in Table 4-31. The four conclusions that can be drawn from this table are: (1) subscribers make more use of the retailer factor but less use of the media factor, (2) there is no significant relationship between upgrades and the three search indices, (3) downgrades tend to make more use of the media and interpersonal search factors and (4) disconnects make more use of the media. These results essentially confirm those results presented in tables 4-27 to 4-30. None of these correlations are very high, quite typical of consumer research of this type. This analysis tends to indicate that other variables are impact- ing on these decisions. The next chapter will present a summary of the key findings. Also included in the final chapter is a discussion of the major limitations of this study and some suggestions for future research in the area of household decision-making processes. CHAPTER V DISCUSSION AND CONCLUSIONS Introduction In chapter 1, four questions were presented which have been used to guide the direction of this study. The answers to and the results of the empirical evidence associated with these questions have been presented hisome detail in chapter IV. This chapter will summarize the main findings and discuss their implications to practitioners and researchers. A second purpose of this chapter is to note some of the limitations and contributions of this research. And finally, some suggestions for future research are presented. Summary of Key Findings In chapter I, it was stated that four questions will guide the present research. These four questions will be restated and briefly answered. 1. To what extent do relative spousal contributions influence the joint decision-making process in the cable purchase decision(s)? Relative spousal occupational status appears to be the most useful resource variable to analyze. When both spouses are approximately equal on this variable, the decision to subscribe or to retain cable tends to be joint. Equal educational attainment and income contribution are not conducive to joint decision-making behavior. Four possible explana- tions may account for the results. 85 86 First, the sample is not truly a representative one. It is a peculiar population that is being represented, those of short-term cable subscribers. It may be possible that the specific spousal characteristics hold for this specific sample and not for the larger population, and vice versa. If the sample were more representative on these characteristics, it is possible that more support for this set of hypotheses would be found. Second, the method utilized in operationalizing the equal and not equal categories could be problematic. If a range were employed for all the variables, as in hypothesis la, stronger support may have been found. It is felt, however, that this methodology for hypothesis la is necessary, owing to the nature of the variable being measured. Since the other two variables are continuous, the present methodtfiidelineat- ing the categories is most appropriate, albeit the most stringent. Third, it may be that the three spousal resource variables of occupational status, educational attainment and percentage contribution to total family income are measuring the same construct, often called socioeconomic status. The Pearson correlation among these variables for the females and males as shown in Table 5-1 and 5-2, respectively, confirms this. All of the variables are highly correlated for the females and the relationship between occupational status and educa- tional attainment for the males is significant. Occupational status may be "overshadowing" the effects of the other two variables. As a result, the findings on these two variables are contrary to those expected. Last, it is possible that a host of dynamic factors, both internal and external to the household, which directly affect joint decision 87 Table 5-1. Pearson Correlation Matrix Spousal Resource Variables Female Characteristics Income Occupational Educational Contribution Status Attainment Income Contribution (%) 1.00 r=.32 r=.l6 n=288 n=415 p=.001 p=.001 Occupational Status 1.00 r=.35 n=314 p=.001 Educational Attainment 1.00 Table 5-2. Pearson Correlation Matrix Spousal Resource Variables Male Characteristics Income Occupational Educational Contribution Status Attainment Income Contribution (%) 1.00 r=.02 r=.03 n=376 n=413 p=.37 p=.30 Occupational Status 1.00 r=.44 n=404 p=.001 Educational Attainment 1.00 88 making processes, has not been captured by the hypothesis. Some of these factors may be cultural influences, stage in the family life- cycle, personality of family members and spousal involvement in the product area of concern. These all have been found to be somewhat influential in previous research, but were not specifically addressed in the present research. 2. To what extent do children alter the influence structure of both spouses in the cable purchase decision(s)? It appears that children do not significantly alter spousal joint, husband-dominant and wife-dominant decisions. One finding may be of some importance, however. Tables 4-19 and 4-20, when analyzed together, indicate that approximately four percent fewer decisions are jointly made when children are present. Moreover, this four percent increase seems to be derived at the husband's expense. That is, when children are present in the household, the wife tends to become more dominant in the purchase/retention decision. Although this finding is not statistically significant, it does suggest that the wife is the "gatekeeper" in cable purchase decisions, when children are present in the household. This may have promotional targeting implications in maintenance marketing and remarketing campaigns. Also, there is a positive association between age of the children and overall influence of the children, when present in the household. The average child influence is 18 (on a scale ranging from 0 to 100), with the two highest influence levels being associated with the Disney and H80 services. 89 3. How does "traditional" versus "contemporary" family ideology affect the distribution of family decision influence among all family members in the cable purchase/retention decision(s)? A larger proportion of family members tends to be involved in the household decision-making when the respondent is more inclined toward a "contemporary" family ideology, with respect to a spousal relation- shipsfactor. Because of the small sample size, however, the statisti— cal significance of this finding is not established. Also, the scale used to determine "traditional" and "contemporary" family ideology may be somewhat dated for today's more sophisticated respondent. It may be possible that the "traditional" market has disappeared, or become very small, at least by the present definition used. Moreover, "social desirability" factors may be altering respon- dents' answers to the questions asked. In other words, the respondent may either know how they should respond to these types of questions or they are indeed truly ambivalent about their attitudes. Whichever may be true, it is possible that if a different scale were utilized to measure this construct, the results may be different. 4. To what extent do prepurchase planning and search activities impact on the family's decision to subscribe or not subscribe to cable television? Overall, it appears that cable subscribers engage in more retailer search and interpersonal search activities than non-subscribers. It appears that both subscribers and non-subscribers are exposed to media campaigns, but it is not clear as to how active the consumers are in seeking the various media for information about cable television. 90 This is a weakness in the measuring device and will be discussed more fully later. Proprietary research by the author has shown that cable is a high interest product among consumers,but along with that interest goes a great deal of complication--misinformation and apprehension. Given the present research, it appears that non-subscribers are exposed to the commercial messages touting cable television, but they are rela- tively ineffective. The advertising campaigns positioning cable tele- vision as a family institution that appeals to discriminating people or portraying domestic warmth by suggesting that TV viewing can be a shared family experience, seem to be more suited for maintenance and retention marketing strategies. The strategies that may be most effec- tive at engendering favorable attitudes toward cable may be free—trial offers. By allowing viewers to sample cable's benefits for a specified time interval, perceptions of the service may become more accurate and churn would reduce. Concomitant with this tactic may be an increase in subscriber sales, an increase in pay unit sales and an overall in- crease in cable penetration. It was found that downgrades tend to engage in more search and deliberation activities, contrary to expectations. One possible explanation for this finding may be that downgrades discussed the basic service more than the premium services. If this is true, this would actually lend some support to the hypothesis. In addition, it may be that the premium service(s) were purchased on an impulse, that is, little or no discussion took place in the household, thus further supporting the hypothesis. Further, the downgrade decision may be regarded as an economic decision; the cable channel is a luxury item 91 and can be discontinued without any harm, thus there is little or no discussion taking place. Whichever, it must be noted that the method used in the present research assumes that the same types of search and deliberation activities occur for the basic service and pay ser- vices alike, when added or discontinued at a later time. Although this may be a reasonable assumption, it is a limitation of the methodology. Finally, there is a conceptual weakness in assigning a value of one to those prepurchase activities engaged in and a value of zero for those prepurchase activities not engaged in. This method assumes that all prepurchase activities are equivalent in importance when in actuality this may not be true. It would be more precise methodologi- cally to weight the individual components of search factors which would reflect the degree of active versus passive discussions and/or delibera- tions on the part of the consumer. These individual factors could then be added together in some fashion to create an aggregate index of search. This was done in Bennett and Mandel's (1969) study of auto- mobile purchasing behavior and in Duncan and Olshavsky's (1982) study of purchasers of new color TV sets. The weights will vary somewhat, dependent upon the nature of the product and the salience of the product to the consumer. Other factors to consider would include the expected gains from search activities, the opportunity costs of time as perceived by the consumer and search efficiency. This is an impor- tant topic for research and much work has been done in this area. One of the major theories dealing with the determinants of consumer search is the "economics of information" model extensively used by economists (Stigler, 1961). Since this was not central to the present 92 research, the simplified methodology was employed. As such, it is recognized as a limitation and may be one of the possible explanations for the results obtained. Limitations of the Research The primary limitation of this research study is the small, re- strictive sample used. Only couples and families subscribing to cable for six months or less in one cable franchise were included in the subscriber group. This was necessary, however, in an attempt to better ensure accuracy of responses related to prepurchase search and delib- eration activities and to determine the extent of joint spousal decision-making. The confirmed non-subscriber group was randomly selected, but those households without telephones were excluded from the sampling frame. Because the sample is not reflective of the general national cable population, it is unclear to what extent the results would be applicable in a different geographic setting. Also, analyses of several of the proposed relationships were limited by the sample size. Methodologically, the study has been limited in several ways. First, only one purchase decision was explored. A fuller under- standing of household decision-making would require the examination of different types of decisions under different environmental and household situations. Second, the discussion and deliberation prepurchase activities were analyzed on a post hoc basis. That is, the data were collected after the cable disposition decision was made. Although attempts were made to alleviate this weakness, further researchers should gather these~types of data in a new-build franchise. If this were done, 93 the stages of the buying process could be more clearly delineated and the data would probably be more valid and reliable. Third, only one respondent in each household was interviewed in determining the influence levels of all household members. Although some researchers (Davis, 1971; Engel, Kollat and Blackwell, 1973) have argued that family-role structure responses from either spouse are adequate if one's purpose is to describe behavior on an aggregated basis, all household members should be interviewed to add more validity to the findings. However, in an attempt to validate the influence measures gathered in this research effort, t-tests comparing the male and female perceptions of their respective influence patterns were conducted. Tables 5-3 through 5-5 present these data for the influence structure of the husband, wife and children respectively, for the basic and HBO services. These data indicate that there is no difference in the perception of ascribed familial influence for all the household members. Thus, the assertions of Davis and Engel, Kollat and Blackwell are corroborated and the validity of the influence structures collected for this research is strengthened. Contributions of this Research ' The essence of marketers' studies of household decision-making has been an effort to identify relationships between specific products or product attributes and the exercise of influence on purchase choices by the husband, wife and children (if present), either individually or jointly. This product-centered scheme is intuitively appealing, and it has suggested viable marketing strategies. However, household consumer choices involve interpersonal relations and information searches as well as product-centered responses, and a fuller 94 mucmuwmwcm_m we pump umpvmu-~m mum. mp.o- Fm.o~ pm.o¢ mm mp.o~ m~.oe Pm om: mmm. mo.o- op.~N mm.p¢ em en.e~ m~.~¢ mu owmwm mmuchPwpcmwm p cowpmw>mo Axv z copump>mo qu z muw>gmm uemucmym cowuamoemm m.ucmnmaz cemocmum cowuamugma m.m$w3 ”mucmzpwcfi m.mmw3 "mucmszcH m_mwrz Ammov>smm mpnmu 03» com mucm=~wcH um>mmucma mmmsm>mo Amy 2 cowumw>wo Axv z mcw>ewm uemccmam :owpamoema m.ucmam=: cemucmym cowpamuema m.oew3 "mocmzpwcH m.ucmnm:: ”mucmzpmcH m.u:mam=: Ammup>gmm mpnmu axe Low mocmzpwcH uw>wmugma mmmem>mo Axv z cowwmw>ma qul 2 muw>emm nemucmpm cowuamoema m.ucmnm:: uemucmum compamuema m.mww2 "mucmzpmcH m.va;o "mucmzpmcH m.u_w;u Ammuw>gmm mfinmu o3» toe mucmapmcH um>mmucma mumsm>