CRQSS‘oPRESSURE , COMW‘i-UNECATEGN. BERN/.308 AND THE HEREBITARY VOTE Thesis {or the Degree of M. A: MtCRJGRE‘é STARE. UNWERSITY {ECGLR E}. HANEY 1.967 L LIBRARY Michlgan rate Universxty IHESlS -IlEn ABSTRACT CROSS-PRESSURBS, COMMUNICATION BEHAVIOR AND THE HBREDITARY VOTE by Roger D. Haney This study developed from the voting works of Lazarsfeld and Campbell where it was found that 75 percent of the voters vote for the same party as their parents. One purpose of this study was to determine the effect of cross-pressures on the hereditary vote. A significant positive correlation was found between cross-pressures (the extent to which referent groups disagreed among themselves in party preference) and deviation from the hereditary vote. Earlier research also indicated that cross-pressures would lead to less political media use and less political conversations. It was further felt that those under cross-pressures who had high political media exposure and more political conversations would exhibit greater deviation from the hereditary vote than those who had low political media exposure and fewer political conversations. The final hypothesis predicted that non-voters would exhibit a . greater degree of cross-pressures than voters. However, none of the above hypotheses were confirmed. Cross-pressures were not significantly related to political media exposure, extent of political conversations nor incidence of voting. Nor did those under cross—pressures with a high degree of media exposure and conversations exhibit greater deviation from the hereditary vote. Roger D. Haney Possible explanations for the nonsignificant findings of the present study are discussed. One reason suggested is that amount of interest in the election should be taken into account when investigating the relationship between cross-pressures and communication behavior. CROSS-PRESSURES, COMMUNICATION BEHAVIOR AND THE HEREDITARY VOTE By Roger D . Haney A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Communication 1967 If ([6! ‘ é 45/99/97 Accepted by the faculty of the Department of Communication, College of Communication Arts, Michigan State University, in partial fulfillment of the re- quirements for the Easter of Arts degree. .5"! 4' , \/ (Hy/”ff". ,. - Dire ctor of Thesis ACKNOWLEDGMENTS The author claims full credit for the results of this study. However, he wishes to express his sincere appreciation to all those who contributed to the completion of the study. Special thanks go to Dr. Verling C. Troldahl, thesis advisor, and to Dr. Bradley S. Greenberg and Dr. Hideya Kumata for their assistance. Thanks go to Mr. Dan Costello fer his help in administering and coding the questionnaires and to Mr. George Robeck for his help in administering the questionnaire. Commendations go to Mrs. Shirley Sherman for not only typing the manuscript, but also ensuring that it was completed in time. Finally, the author wishes to thank his wife Joyce, without whom.... Ho Ho TABLE CHAPTER I THE RESEARCH PROBLEM . . _Introduction . . . . The Hereditary Vote . Cross-Pressures . . . OF CONTENTS Rationale For Hypotheses . . . . . II RESEARCH DESIGN . . . . The Sample . . . . . Operationalization of Variables . Deviation From Hereditary Vote . . Cross-Pressure . . . . Media . . . . . . . . Newspaper-Readership Magazine Readership Television . . . . . Radio . . . . . . . Data Analysis . . . . III FINDINGS O O O O O O 0 Description of the Sample . . . . Registration and Voting . . . . Description of the Vote . . . . Deviation From Hereditary Vote . . Test of the Hypotheses IV SUMMARY AND DISCUSSION Summary of Results . . Discussion . . . . . Sampling . . . . . Measurement . . . . Extraneous Variables Implications . . . . . BIBLIOGRAPHY . . . . . . . . . iii wMMH 10 10 11 12 1M 15 15 15 15 16 18 18 18 21 22 22 30 30 32 32 3M 37 38 H1 LIST OF TABLES Description of Sample by Registration Behavior . . . . Description of the Sample on Demographic Attributes by Voters and Registered, Non-Voters Description of Sample: Self Vote and Vote Estimation of Reference Groups . . . . . . . . . . . . . . Correlation Between Cross-Pressure and Political Content Exposure in the Mass Media . . . . . . . Correlations Between Cross-Pressure and Political Conversations . . . . . . . . . . . . Media Exposure, Cross-Pressure and Deviation From Hereditary Vote . . . . . . . . . . . . . Political Conversations, Cross-Pressure and Deviation From Hereditary Vote . . . . . . . . . Rage Of cmss-Pmssms e o e e o o e o e e o 0 iv 18 2O 21 23 2Q 26 27 28 LIST OF APPENDICES, Page APPENDIX A: QUESTIONNAIRE . . . . . . . . . . . . #2 APPENDIX B: EXTENT OF COMMUNICATION BEHAVIOR . . N7 CHAPTER I THE RESEARCH PROBLEM Introduction One of the more popular arguments used against those who use the methodology of science in the study of human behavior is that human behavior is fundamentally different from the subject matter of the natural sciences. Human beings do not follow some pre-determined "law" of action as do falling objects or the paths of the planets. In short, goes the argument, humans are "free" to do and choose what they wish. However, evidence in the social sciences is accumulating that tends to indicate otherwise. During recent television coverage of national elections, computers were utilized to make predictions of the final outcome much more quickly. In fact they were so fast that often the prediction would be made before more than one or two percent of the votes were tabulated. This led to a _ great many phone calls from the viewers, to whom the commentator explained the procedure as best he could. He explained that only "key" precincts were counted and from these political scientists could tell how the district would go, and in turn how the entire state would go. Of course, some errors were made, but the commentator aptly summarized, "people are more regular than we had imagined." The Hereditary Vote People do seem to be more regular than imagined. Voting studies have found that roughly 75% of first voters voted as their fathers did (Berelson, Lazarsfeld and McPhee, 195'4, p. 89; Campbell, 1951+, p. 201). This has come to be called the "hereditary vote" and by it is meant "the influence of the political tradition of one generation upon the political conviction of the next." (Berelson 3. 3:12., p. 88). The question then arises as to when this hereditary vote will break down, and, of especial interest to the communication researcher, what role the mass media play. Klapper (19614) in his extensive summary of mass media effects, concludes that conversions are relatively infrequent in comparison to reinforcement effects. However, he also notes that "persons under cross-pressure have been observed to be particularly susceptible to conversion by mass communication as well as by personal influence." (Klapper, p. 77). Cross-Pressures Cross-pressures have been defined as intra-personal conflicts which arise when the motives affecting a decision are incompatible (Pinner, p. 1). Thus if one's personal friends belong to one party and his business associates to another, his voting choice will be affected by cross- pressures. The cross-pressure hypothesis is a prediction concerning the outcome of such internal conflict and is usually given in forms of "withdrawal" (i.e., "loss of interest," "delay in time of final decision"). However, balance theory suggests that withdrawal from the field is only one of several ways to resolve the conflict. Another way would be to seek more information about each candidate in order to resolve the conflict. In the area of politics, as in other areas, one could do this principally in two ways: through the mass media and/or through personal contacts. It is the purpose of this thesis, then, to investigate the inter- relationship of four variables: 1. The Hereditary Vote 2. Cross—pressures 3. Political Media Exposure u. Frequency of Political Conversations. Rationale For Hypotheses The first question to be resolved is whether cross-pressures lead to withdrawal or information seeking. It was found in the Erie County study that the more interested voters are in the election, the more they expose themselves to campaign propaganda. Thus, those with high interest had a mean media exposure of 12.0, those with medium interest had a mean media exposure of 8.7 and those with no interest had a mean media exposure of 6.2 (Lazarsfeld, 23? gl,, 19u8, p. #2). The significance of this finding is realized (in terms of this study) when it is seen that it is those who are under cross-pressures who lose interest in the election (Lazarsfeld, 35, gl,, 19u8, p. 62). Thus it can be argued that cross- pressures lead to less political media exposure. A study by Greenberg (1965) also supports this latter notion. There the voter's expectation of who would win the election was determined. Greenberg found that it was those who thought their side would win who exposed themselves to more information. If intending to vote for a candidate whom one expects to lose is regarded as cross-pressures (See Chapter II), the notion that cross-pressures leads to less media exposure is again supported. Thus the following hypothesis is suggested. H1: The more a person experiences cross- pressures, the less his political media exposure. Evidence for the second hypothesis is conceptual rather than enpirical. Thus far it has been indicated that when a person is under cross-pressure he tends to avoid further information. Since information might come not only from the mass media, but also through personal in- fluence, a likely complement to the first hypothesis is the second: “2: The more a person experiences cross- pressures, the less frequently he has political conversatims. As referred to earlier, in the Elmira study it was found that four out of five _f_i_r_~_s_t_ voters who had Republican fathers voted Republican in the 19% presidential election, and that two out of three with Democratic fathers voted Democratic, "even in this Republican town" (p. 89). Thus about 75% of the first voters sided with their fathers in their political choice. Furthermore, a high proportion continued to do so regardless of age. The important thing to note here in terns of the present study is that (l) the town is Remtblican (61% of the voters voted Republican in a Democratic year, p. 15), and (2) the percentage of hereditary vote- for first voters varied from 80% for Republicans to 67% for Democrats. Even for the age, group "45 and over," the percentage was 86% Republican and only us% Democratic. In other words the hereditary» vote tended to break down when one is in a minority party. The Campbell study also found figures comparable to the 75% hereditary vote of the earlier studies but broke it down in a somewhat different manner. Thus while 72% of the sample with both parents Democratic tended to be Democratic, 36% were rated strongly Democratic and 36% weak Democratic. 0f the 63% Republicans who have both parents rated as Republican, 33% were rated as strong Republican, while 30% were rated as weak Republican. This comparatively high percentage of "weak" Republicans would lead one to suSpect that the hereditary vote isn't as impregnable to change as one might suppose. The question of interest now is identifying the factors that are coupled.with a less hereditary vote. As described in Voting, "it is people with 'cross-pressured' opinions on the issues, or candidates, or parties-~that is, gpinions or views simultaneously supporting different sidee--who are more likely to be unstable in their voting position during the campaign" (Berelson, 33 al;, 195“, p. 19). Thus on various issues the percentage of voters who remain constant in their voting intentions is 75-81% for those who have a "consistent" opinion, while it is only 66-68% for those who have an "inconsistent" opinion. In other words people with cross-pressures tend to be less constant in their voting intentions. This is, of course, a reasonable conclusion. One would expect a person who holds views in support of different sides of an issue to be indecisive in reaching a decision concerning them. Further evidence below indicates what factors have an effect in reaching a decision. In the "Erie County" study it was concluded that "of all the cross- pressures which we have identified, the single most effective one in delaying vote decision was the lack of complete agreement within the family (Lazarsfeld, gt_al;, 19u8, p. 60). They found that less than 3% of voters in homogeneous families change their mind between August and November, but in families where some relatives were undecided, 10% shifted between August and October. However, they tended to change toward the party favored by the rest of the family. Fully 81% of the members of Republican familieswho were originally undecided were pro— Republican in October; and 71% of those in Democratic families later came out for Roosevelt. A study by MaccOby (lQSH) also found that agreement was highest with parents. This is as expected; otherwise the hereditary vote would not be as high as it is. Maccoby also found that after parents, agreement is highest with friends and fellow workers. This evidence, and that which follows, suggests that when the hereditary vote does break down, personal influence is at least one factor to be considered. As noted earlier, Elmira, N.Y. at the time of the Voting Study was a "Republican" town. It was also noted that the Democrats exhibited a somewhat less hereditary vote. It was also found in that study that Republicans are more likely to have friends who are Republican and Democrats are more likely to have friends who are Democrats. "By and large," they concluded, "the voter is tied into a network of personal associations that is both homogeneous and congenial" (lQSu, p. 9“). However, while only one in five Republicans had a Democrat among his immediate associates, two in five Democrats had a Republican listed as a close associate. In other words, the "network of personal associations" was more homogeneous for the Republicans than the Democrats, and it was the Democrats who showed less hereditary vote in this study. Un- fortunately the data don't show whether or not the Democrats who had less hereditary vote are the same ones who had the higher percentage of Republican friends, and were thus possibly subject to cross-pressures. One might assume, however, that they are the same people. It was also found in that study that the hereditary vote becomes somewhat less with age (75% for the age group "21-25" compared to 66% for the age group "us and over"), and that the reason for this is due primarily to more voters of Democratic heritage to switch to Republican (33% of the "21-25" age group switched compared to 55% of the "HS and over" age group). This may imply a tendency for older people to vote Republican, but there is the added implication that it is the result of aSsociation with friends. In other words, there is the implication that as one grows older, the influence of one's parents on voting becomes less, while the influence of one's friends becomes greater. Finally the Elmira study found that, while 66% of the respondents tend to talk politics with people of the same party, those who discuss politics with the opposition are more likely than others to change parties. Thus of those who talked with someone of the same party, 1% changed to the other party; whereas of those who had discussions with a person of the opposite party 13% changed. In a study by Deutschmann and Pinner (1960) it was feund that knowledge of two pro-Republican messages produced a net shift of 16% in voting intentions toward the Republicans. It was also found that conversation along with media exposure is important. Thus media exposure with no conversation accounted for only u% of the net shift, while media exposure with pro and con conversation accounted for 1n% and media ex- posure with only pro conversation accounted for 33% of the shift. But the study also reports that media exposure stimulated the conversation and that conversation played a supplementary role. The implication thus far then is as follows: One tends to vote as his parents did. When he is subject to differing views, whether from mass media or because of conversations with friends and fellow workers, he tends to withdraw from the situation (complete withdrawal, of course, would exhibit itself in non-voting). However, if he does vote, this con- tact with the other side of the issue does have an effect, viz., less hereditary vote. This argument leads one, then, to the final three hypotheses. “3‘ Those who experience cross-pressures and have high political media eXposure exhibit greater deviation from hereditary vote than 1 those who experience cross-pressures and have low political media exposure. Those who experience cross-pressures and have high frequency of political conversations exhibit greater deviation from hereditary vote than those who experience cross-pressures, but have low frequency of political conversations. E. The fifth and final hypothesis has been alluded to earlier. HS: Among registered voters, individuals who exhibit high cross-pressures will be less likely to vote than individuals who exhibit few cross-pressures. . The argument here is conceptual rather than empirical. It is simply that if cross-pressures in general lead to withdrawal, then those who withdraw completely (by not voting) must be under greater cross- pressures than those who actually vote. CHAPTER II RESEARCH DESIGN The Sample Respondents for the present study came from three \mdergraduate courses (Business Law, Journalism and Communication) taught at Michigan State University. All questionnaires were collected within three weeks after the November 8, 1966, state and national election. There were two groups used in the analysis: (1) those who voted in Michigan and (2) those who registered in Michigan, but did not vote.1 These groups were determined through a series of filter'questions. First the respondent was asked if he was a registered voter. All respondents answering Ng_to this question were not used in any of the analyses. Respondents answering Zg§_to this question were then asked in which state they were registered. Again, those respondents who were not registered in the state of Michigan were not used in any of the analyses. Those who answered that they had registered in the state of Michigan constituted the sample for this study. Respondents were then asked the question: In talking to peOple about the election, we find that a lot of people weren't able to vote because they weren't registered, or they were sick, or they just didn't have time. How about you, did you vote this time? l . . This latter group was needed to test Hypothesrs 5. 10 11 Those who had voted were placed in the "Michigan, voted"_ group, while those who had not voted were placed in the "Michigan registered, not voted"; group. Respondents who had voted were also asked how many times they had voted up to that time in the hopes that there would be enough first-time voters to use exclusively'in the analysis. This was not the case . @erationalization of Variables Deviation From Hereditary Vote To determine deviation from hereditary vote, twO things are necessary. First it must be determined how the reapondent voted. Thus respondents were asked: "In the election just held, which answer below best describes how you voted for state and national offices?1 1N Straight Democrat 12 Mostly Democrat 10 A few more Democrats than Republicans 8 About equally for both parties 6 A few more Republicans than Democrats It Mostly Republicans 2 Straight Republicans The higher scores indicated a Democratic inclination, lower scores indicated a Republican inclination, and middle scores indicated about equal or slight preference for one party or the other. The Elmira Study (Berelson, 93.- 31;, 195u, p. 363) only asked which presidential candidate one voted for; the Campbell study broke party affiliation down into The complete questionnaire (code sheet) used in this study can be found in APPENDIX A. l2 "strongly" or "weakly" Republican, etc. In the present study, it was felt that since deviation from hereditary vote was such an important variable, it was necessary to measure deviation as precisely as possible. To obtain the hereditary vote measure, re8pondents were asked to indicate how (1) their father and (2) their mother voted.in recent elections.1 The same categories were used as fer the self-vote, but they were coded from 7 to 1. This enabled the vote of each parent to be weighted equally. They were then summated to determine "total hereditary vote." The range of this index now correSponded to that of the resPondent's self-vote.‘ These two scores were then subtracted to determine "deviation from hereditary vote," the possible range of deviation being -12 to +12 (zero indicating no deviation and 12 indicating the greatest amount of deviation possible). Cross-Pressure There are several ways in which cross-pressures have been operationalized. Campbell, st, 21:, (1960, pp. 370-380) defined the individual as subject to cross-pressure if he identifies himself with a (higher) social stratum than that to which he actually belongs. In their Erie County Study, Lazarsfeld, Berelson and Gaudet (19u8) included.not only the economic social stratum (class), but also religion and area of residence (urban vs. rural). Thus an urban, upper-class Baptist was defined as being under cross-pressure. Pinner (1967) distinguishes between "attitudinal" and "affiliative" cross-pressures e It was also asked if their father and mother were still living to determine relevancy. 13 Attitudinal cross-pressures arise when decisions "between beliefs or courses of action bring into play attitudes which motivate different and Opposing selections of alternatives (Pinner, p. 1). Affiliative cross- pressures "result from a person's attachment to several groups which have preferences fer different alternatives" (Pinner, p. 1). Thus, in the area of voting, if a person agrees with the domestic policy of one party and the foreign policy of another, he will be subject to attitudinal cross-pressures. If an individual's personal friends belong to one party and his business associates to another, the individual's voting choice may be subject to affiliative cross-pressures. Pinner goes on to say that the two types of cross-pressures are not mutually exclusive and fre- quently coincide. (Pinner, p. 1). In the Elmira Study, Berelson, Lazarsfeld and McPhee (195M, pp. 118-127) operationalized cross-pressures in terms of disagreement among the voter's family, friends and cOdworkers. The authors felt that personal associates were more effective in influencing the voter's decision than his social aspirations (Campbell) or his class, religion and area of residence (Lazarsfeld, Berelson and Gaudet). In all three studies, however, it should be noted that cross-pressures are operationalized in terms of affiliations and not attitudes as such. The present study takes this same approach and operationalizes the degree of cross-pressures as the amount of disagreement in voting among three primary groups: parents, neighbors, and those majoring in the same area as the resPondent (a student's "codworkers"). Thus the respondent 11+ is asked to estimate how he thought (1) his father, (2) his mother, (3) people who _l_ile; near him ("people in your neighborhood, in your apartment area, or in your dam") and (1}) students majoring in the same subject voted in the last election. Responses were again coded along a continuum from Straight Democrat (coded as a 7) to Straight Republican (coded as a 1). Two measures were then used as an index of cross-pressures. First the 529.89. of these four questions was determined. By this is meant the _ greatest single difference among the four items. The possible range is from 0 to 6, a 0 signifying that the respondent estimates the vote of all four of these persons exactly the same. A 6 would mean that at least one of the estimations was "Straight Democrat" while at least one of the other estimations was "Straight Republican." Since the above measure only considers the greatest single difference among the four items, the standard deviation of the four responses was also used as an index of cross-pressures. Because this index took into account all four items at once, it was felt that this might be a more precise index. In fact, the two measures of cross-pressures were correlated .98, so it doesn't matter which measure is used. The standard deviation measure was used in this study. Media The media questions pertained only to media content relevant to politics. Two questions involve print media, and twO questions involve electronic media. By breaking mass media usage into these four separate 15 media, each medium could be analyzed separately to determine its particular relation to cross-pressures and the hereditary vote. The correlation between usage in each medium and every other medium was also determined. Newspgper Readership. -- To obtain an index of how much political content the resPondent read in newspapers the question was asked, "During the last week of the election campaign, about how many news stories dealing with political candidates did you read in newspapers?" Magazine Readership. -- To gain an index of how much political content the respondent read in magazines, the question was asked: "During the last month of the campaign, about how many magazine articles on political candidates did you read?" Television. -- For television the question was asked, "During the last $9939. of the campaign, about how many political talks or shows on T.V. (of 15 minutes or more) concerning political candidates did you listen to?" m. -- To determine political content exposure in the medium of radio, the reSpondent was asked, "During the last month of the campaign, about how many political talks or shows on radio (of 15 minutes or more) concerning political candidates did you listen to?" Political Conversations Two questions were asked to determine how often the respondent engaged in conversations of a political nature. He was first asked, "During the last month of the campaign, about how often did you have 16 conversations with people about political candidates?" Then he was asked, "And finally, during the last Eggth_of the campaign, about how often would you say people asked you for your opinion on politics?" The responses were coded as follows for each question: u several times a week once or twice a week 3 2 once or twice a month less often Data Analysis Four of the five hypotheses utilized a correlational approach. The fifth was tested by means of a t-test. Only the respondents who had zgtgg_in Michigan were used to test the first fbur hypotheses.. To determine if voters experiencing cross-pressures had less use of the mass media zero-order correlations were obtained between the standard deviation of the cross-pressure scores and scores of political eXposure to each of the following four media: (1) Newspapers, (2) Magazines, (3) Radio, and (u) T.V. A significant negative correlation was predicted. The same procedure was used to test the second hypothesis, except that scores for political discussion and opinion leadership were substituted for the media scores. Again a significant negative correlation was pre- dicted. ‘. To test Ha, the sample was first Split as close to the median as possible into a high—low dichotomy of political media exposure on each of the four media measures. Then a zero-order correlation between cross- pressures and the deviation from hereditary vote was determined for each 17 of these subgroups. It was predicted that the correlation in the "high exposure" subgroup would be significantly higher than the correlation in the "low exposure" subgroup. Thus, as an example, the sample of 103 voters was first Split along the median on the basis of the reSpondent's newspaper exposure. For each of these two sub-groups, the correlation between cross-pressures and deviation from hereditary vote was computed. This same procedure was then followed for magazines, then for radio and then for T.V. A similar procedure was followed to test H“, In this case the sample was split into a high-low dichotomy on the basis of (1) political conversations, then (2) political opinion leadership. Again zero- order correlations between cross-pressures and deviation from hereditary vote were determined. T0 test H5, the sample of Michigan voters and the sample of Michigan registered, but non-voters, were used. The mean of the cross- pressume scores was determined for each group. The significance of the difference between these tw0 means was then tested by use of the t-test. CHAPTER III FINDINGS Description of the Sample Registration and Voting As stated earlier, respondents came from three higher-level undergraduate classes taught at Michigan State University. The main group (hereafter referred to as Group I) was to consist only of those individuals who had registered and voted in Michigan. The second group (used only in the analysis for H5) was to consist of those who registered in Michigan, but did not vote. The description of the sample in terms of registration and voting appears in Table 1. Table 1. Description of Sample by Registration Behavior Percentage of Total Not Registered: Under 21 21% Over 21 30 Registered, Out-of-State Non-Voted 10 Voted on Registered, Michigan" Non-Voted 09 Voted ' 25 Incompleted Questionnaires 01 100% *These were the two groups used in the study N = #17 18 19 In total, #17 questionnaires were administered. Of these, 25 percent voted in Michigan and nine percent registered in Michigan, but did not vote. Another one percent of the total could not be used due to incomplete questionnaires. In all six cases this was due to incomplete information on the media items. This left a total of 270 questionnaires, or 65%, which were not used in the analysis either due to non-regiStration or registration in states other than Michigan. Table 2 represents a description of the sample on several demographic.attributes, by Veters and Registered, Non-Voters in Michigan. Pearson's chi—square was used to test the similarity of both groups on these attributes. They differ significantly on all traits except "class" and animation of parents. Since the group of primary importance in terms of the hypotheses of this study is the group of voters, the verbal description of the sample is in terms of that group. Approximately three-fourths of the sample was male. Three-fifths of the sample were of a minimum voting age; however, 13 percent of the voters were 31 or over. Eight of every 10 were undergraduates, and six of every 10 were still single. Three-fourths of the sample lived off-campus. An important question in terms of the hereditary vote was whether or not the reSpondent's parents were still living. The two groups had almost equal percentages here, with the father of 8 of every 10 still living and the mother of 9 of every 10 still living. Thus the sample may be described.as predominantly male, 21, undergraduates living off-campus. 20 Table 2. Description of the Sample on Demographic Attributes by Voters and Registered, Non-Voters Personal Characteristics Voters Registered, Non-Voters Sex: Male 76% 87% Female 2a 13 W96 mm N=103 N=38 Age: 21-22 57% 50% 23-2“ 10 29 25-26 7 10 27-28 6 3 29-30 3 5 31 and over 13 3 Too—96 160% Class: Graduate 23% 21% Undergraduate 77 79 E696 1639s Marital Status: Married “2% 32% Single _§§_ 68 100% 166% Place of Residence: Married Housing 15% 5% Dormitory 10 13 Off-Campus Housing 75 82 m “136% Parents Still Living: Father: ' Yes 82% 3u% No _13 16 100% 163% Mother: Yes 91% 92% No 9 8 163% 100% 21 Description of the Vote Table 3 presents a description of the way resPondents voted and the way they felt certain referents (parents, neighbors and those in the same major'as the reSpondent) voted. As can be seen by inspection of the table, the sample not only tended to vote Republican, but also Table 3. Description of Sample: Reference Groups Self Vote and Vote Estimation of Vote Self Father Mother Neighbors Same Major Republican: Straight 25% 18% 17% 9% u% Mostly 38 37 38 38 35 A Few More 9 8 8 27 28 About the Same 9 5 9 1a 18 For Both Parties Democrat: A Few More 6 3 3 5 u Mostly 10 18 15 5 11 Straight 3 11 10 2 0 Total 100% 100% 100% 100% 100% N=103 22 tended to estimate the votes of their referents as Republican. Thus approximately 3 of every '4 respondents voted Republican, while only 2 of every 10 voted Democratic. Six of every 10 respondents felt that their parents had voted Republican and 7 of every 10 felt that their neighbors and those in the same major voted Republican. Deviation Prom Hereditary Vote As a comparison measure, it was also determined how many respondents voted for the same p§££z_as their parents. Only 26 of the 103 respondents voted for a different party than their father did, and only 21 of the 103 reSpondents voted for a different party than their mother. Thus approximately 77% of the sample voted the same way as their parents, a figure quite comparable to previous research. In terms of the hypotheses for the present study, however, any deviation, even within the same party is considered a deviation from hereditary vote. Test of the Hypotheses H1: The more a person experiences cross-pressure, the less his political media exposure. The first hypothesis predicts that those who have a high degree of cross-pressure will have less exposure to political content in the mass media than those who have a low degree of cross—pressure. In other words cross-pressure and media exposure should.be negatively correlated. To test this hypotheses zero—order Pearsonian correlation coefficients were oomuted between cross-pressures and exposure to political content in 23 the following four media: (1) newspapers, (2) magazines, (3) radio, and (M) television. Thus for the single theoretic hypothesis, four empirical hypotheses were tested. For these tests, only the voter group was wed in the analyses. The results of the computatims relevant to “1 are summarized in Table u. The results do not support the hypothesis. Only'the correlation between cross-pressure and newsPaper readership and cross-pressure and T.V. viewing approach significance, but in a direction opposite of that predicted. Table u. Correlations Between Cross-Pressure and Political Content Exposure in the Mass Media Type of Media Zero-order r (N=103) Political Newspaper ;* Readership .13 Political Magazine * Readership .02 Political Radio * Listening -o0“ Political T.V. * Viewing .15 *A correlation of .16 was needed to be statistically significant at the .05 level, 2 tailed test. 21! H2: The more a person experiences cross- pressures the less frequently he has political conversations. This hypothesis predicts that those who have a high degree of cross-pressures will have less political conversations than those who experience a low degree of cross-pressures. Thus cross—pressures and political conversations are predicted to be negatively correlated. Questions were asked to determine amount of political discussion and also extent of political opinion leadership. Zero-order correlations were computed bemeen each of these measures and cross-pressures. The results are sumnarized in Table 5. The hypothesis is not supported by the data. Only the relationship between opinion leadership and cross- pressures approaches significance. Table 5. Correlatims Betwaen Cross-Pressure and Political Conversations Political Conversations Zero-order r (N=103) Amount of Political * Conversations -.09 Opinion Leadership - . * -015 *A correlation of .16 was needed to be statistically significant at the .05 level, 2 tailed test. 25 H3: Those who experience cross-pressure and have high political media eXposure exhibit greater deviation from hereditary vote than those who experience cross-pressure and have low political media exposure. This hypothesis predicts that the correlation between cross- pressures and deviation from the hereditary vote will be greater for those who have a high degree of political media exposure than it would be for those who have a low degree of political media exposure. Since a re- lationship between cross-pressures and a deviation from the hereditary vote was a crucial starting point for this hypothesis and hypothesis u, the product moment correlation between these two variables was determined. It was found to be significant1 (.50). Media exposure was operationalized as political exposure to (1) Newspapers, (2) Magazines, (3) Radio, and (u) T.V. Each medium was analyzed separately and the results are reported in Table 6. The results do not support the hypothesis. A negative difference between the correlations indicates a direction Opposite than that predicted. Thus the correlation between cross-pressures and deviation from hereditary vote is somewhat greater for those with low exposure in the case of magazine readership and radio listening, but not significantly. J'A product moment correlation of .23 was needed to be significant at the .001 level, 2 tailed test. 26 Table 6. Media Exposure, Cross-Pressure and Deviation From Hereditary Vote Type of Media r z n Political NewSpaper Readership: High .su .60a 53 * Low .ng .536 50 .32u Political Magazine Readership: High .38 .uoo us * Political Radio Listening: High .34 .35u us * Political T.V. Viewing . ' High .50 .5u9 us * Low .51 .563 57 -.07 *p (z = 1.95): .05 27 H“: Those who experience cross-pressures and have high frequency of political con- versations exhibit greater deviation from hereditary vote than those who experience cross-pressures, but have low frequency of political conversations. This hypothesis predicts that the correlation between cross- pressures and deviation from the hereditary vote will be greater for those who have a large number of political conversations than it will be for those who have a small number of political conversations. Table 7 summarizes the results fir this hypothesis. The hypothesis is not supported by the data. Table 7. Political Conversations, Cross-Pressure and Deviation From Hereditary Vote r z n Political Conversations: High .u2 .uua 5n * Low .57 .648 49 -l.00 Opinion leadership: High .ue .523 65 * Low .5” .60”, 38 .39 *p (z=l.96)= .05 28 “5‘ Among registered voters, individuals who exhibit high cross-pressures will be less likely to vote than individuals who exhibit few cross-pressures. The fifth and last hypothesis predicts that in a group of registered respondents, those who vote will have a lesser degree of cross-pressures than those who do not vote. A summary of the results appears in Table 8, along with a description of the sample on the cross-pressures measure. The results approach, but do not reach significance. Table 8. Range of Cross-Pressures Range Voters Non-Voters 0 12% 3% l 16 18 2 23 19 3 19 18 u 16 29 5 13 13 6 1 0 100% 100% N=103 N=38 2': 2.55 E”: 2.92 t = 1.27* *p (t = 1.95) N O O 01 29 To summarize then, the five hypotheses for this study were not supported. The implications of these results for future research, as well as some possible explanations for the failure of the predictions are outlined in Chapter IV. CHAPTER IV SUMMARY AND DISCUSSION Summary of Results The motivations fer this study came from the voting studies of Lazarsfeld gt glyfllQuu, 1950) and Campbell 33: al.(1954). There it was found that approximately 75% of the voters voted the same way their fathers did. It was felt that cross-pressures and exposure to political content in the media and in discussions would have a direct and inter- related.bearing on the hereditary vote. As discussed in Chapter II, cross-pressures have been operationalized in many ways. Pinner (1967) distinguishes between "attitudinal" and "affiliative" cross-pressures. ‘The present study operationalized cross- pressures in terms of disagreement _a_m_on_g_ primary affiliations. Since a relationship between cross-pressures and a deviation from the hereditary vote was a crucial starting point for hypotheses 3, u and 5 of this study, the product-moment correlation between these two variables was deter— mined. It was found to be highly significant (.50). It was felt one could react to cross-pressures in one of two ways: he could either withdraw from the situation or he could seek more in- formation. From previous research, it was implied that one would tend to withdraw. This 16d 1:0 H1! H2 and H5. If cross-pressures are resolved by withdrawing from the field, this should exhibit itself in at least 30 31 three ways. The respondent should have less political media exposure (H1), fewer political conversations (H2) and a lower incidence of voting as). None of these hypotheses were confirmed. People under cross- pressures tended to engage in political conversations about as often as those who experienced little cross-pressure. Nor did they tend to per- ceive themselves as being asked less often for their political opinions. Third, non-voters did not tend to be under a greater degree of cross- pressures. Respondents under cross-pressures did not tend to have less Ipolitical media exposure. If anything, for newspapers and television they may be higher in media exposure. Earlier research had found that when a person changes his mind on voting intentions, he cites media exposure and/or conversations as the stimulus for the change. This led to the hypothesis that people under cross-pressures who had a greater degree of political media exposure (H3) and a greater degree of political conversations (Ha) would exhibit less hereditary vote than those with few conversations and low media exposure. In other words it was hypothesized that the correlation be- tween cross-pressures and deviation from hereditary vote would be 'greatest for those with high degrees of conversation and media exposure. Again, the hypotheses were not supported. The correlation between cross-pressures and deviation from hereditary vote tended to be about the same regardless of nxmmber of conversations and amount of media exposure. 32 Discussion The reason a hypothesis is not confirmed is due either to the lack of a real relationship in the population or to weak methodology. Methodological prOblems may be further subdivided into those of (l) sampling, (2) measurement, and (3) extraneous variables. Each of these areas will now be discussed in relation to the five hypotheses of this study. Then certain implications for further research will be pre- sented. Sampling The resPondents in the sample for the present study came from upper-level undergraduate classes taught at Michigan State University. Only those respondents who registered in the state of Michigan were used. Thus the sample was not random. This does not preclude, however, the possibility that the sample was representative of the college population (in relation to the variables under study. RandomiZation ensures that the sample is representative of the population from which it is drawn. Unfortunately, however, it is not always possible to employ randomization in social science research. Few researchers would disclaim significant findings based on non-random samples, unless generalized far beyond reason. It is equally unreasonable to claim that insignificant findings are due to lack of randomization, if one can give evidence to show that the sample is representative of its population. This was not possible. There is no reason to suspect that the resPondents in the present sample 33 differ radically from other college students in relation to the variables under consideration. The fact that the sample was composed of college students may have a more important bearing on the nonsignificant findings of the present study.) Hypotheses were derived from research on the general population. It may be that college students differ from the general population with respect to the variables under consideration. First, college students may react differently to cross-pressures than the rest of the population. It may be that because of their education they are more open to, and therefore less affected by, disagreements be- tween their primary groups. Thus cross-pressures may not affect the media behavior, nor the amount of political conversations, nor the incidence of voting of college students. The present study indicates that that is the case . Second, the sample was young. Eighty percent of the sample of non-voters were between 21 and 2a years of age. Lazarsfeld, g£_al;_(19u8, p. u5) and Campbell and Kahn (1952, p. 37) found that the older age . groups tend to have a greater percentage of voters than younger age groups. Lazarsfeld, 23, El, (1908, 195u) found interest to be an important variable in relation to voting. Younger age groups (21-25) tend to have less interest in the election (Lazarsfeld 33, al,, 19u3, p. #6; 195n, p. 25). Thus, in relation t0 HS’ the non-voters may not have voted because they were not interested in the election. The relationship between cross- 3‘4 pressures and non—voting may be greater for older age groups, however. The effect of interest in relation to the hypotheses of this study will be discussed in greater detail below. Measurement There are two problems in the area of measurement that may lead to insignificant findings. The first is whether or not the question asked is precise enough for the reSpondent to answer it meaningfully. The second is whether or not the question actually measures what the researcher wants it to measure. Both would seem to have a bearing on the present study. Since there were only single items used for each measure, and the items were administered only once to the sample, it is not possible to test the reliability of the questions. However, certain inferences might be made based on the relative ambiguities of the questions. The media exposure and political conversation questions asked the resPondent to estimate exposure during the last Egnth_of'the election (except in the case of newspapers which asked for an estimate on the basis of a week). This may have been difficult for the respondent to do. It also may be that the respondent based his answer on exposure at the time of the study (2-3 weeks after the election). This would decrease validity. The fact that measures of media exposure and extent of political conversations were taken at only one time may explain why H1 and H2 were not supported. Only one measure can not indicate how a person's media habits or personal interaction habits change over time. Thus, 35 in the present study, the individual's political media exposure and his extent of political conversations was in relation to the rest of the s anple rather than himself. It may be that an individual under cross- pressures has less media exposure and less conversations than when not under cross-pressures, yet still have relatively high media exposure and more conversations in relation to the rest of the sample. That possibility was not investigated in the present study. However, the results of the present study do indicate that if people are affected by cross-pressures in relation to their media exposure and number of conversations, other variables have to be taken into account and more precise measurements made. Another possibility that would explain the non-significant product- moment correlations of the present study is whether or not the measures satisfy the requirement of linearity. As McNemar (p. 202) puts it "the degree of correlation is a function of the error of estimate variance relative to the total variance of the variable being predicted by a linear regression line." If the array means fail to fall on a straight line, it can be argued that better prediction can be made by using a curve which really "fits" the means. Such a measure is the correlation ratio or Eta. To test the possibility that the correlation was a curvilinear relationship. Etas were computed between the range of cross-pressures and, in turn, political exposure in newspapers, magazines, television, radio, extent of political conversations and the degree to which the respondent was asked for his Opinion in the area of politics. None of the 36 Etas were significant. The Etas ranged from .01 for range of cross- pressures and magazines (F = .11, approximate significance of F .99) to an Eta of .06 fer range of cross-pressures and television (F 1.06, approximate significance of F = .39). Thus the zero-order product- moment correlations were not significant and allowing for curvilinearity does not have a significant effect on the cross-pressures and communication behavior relationship. The measures of cross-pressure and the hereditary vote were based on the respondent's estimation of how certain referent groups had voted. It may be that the reSpondent based his estimations to a certain degree on his own vote. The correlations between self-vote and the vote estimations for father, mother, neighbors and people in the same major ranged from .21 to .ua, all of which were significant.1 Since the individual must be aware of how his referent groups voted to be under cross-pressures (as operationalized in this study), asking fer an estimation is not in itself, an invalid procedure. However, the fact that the estimations were made after the election allows the possibility that they were restricted in a direction toward the individual's own vote. In other words, the fact that a person does vote indicates that he has resolved any internal conflict within himself concerning a vote decision. To reduce cross-pressures, the individual might then bias his estimation of how his referent groups voted in a direction toward his own vote. This would restrict the range of the cross-pressures measure and thus result in a lower correlation. 1 P (r = .16) = .05 37 An argument against the above having taken place, however, is the fact that the cross-pressures index does correlate significantly with deviation from hereditary vote. This indicates that the cross-pressures measure used in this study is a valid one .. Extraneous Variables As mentioned earlier, a random sample was not used in the present study. Thus extraneous variables were not controlled in this study. Previous research indicates that "interest" is an important variable related to reaching a voting decision (Lazarsfeld, 5511:, 1948, pp. 52-65). They found that as cross-pressures increase, interest in the election tends to decrease (p. 62). This was a basic notion for the hypotheses of the present study. It was felt that indicants of less interest would be less political media use (H1), less political con- versations (H2) and less voting. As indicated earlier, the non- significant findings of the present study may be due in part to the fact that college students made up the sample. The communication behavior of college students in an election may not be affected by cross-pressures due to their training in investigating issues. However, that possibility was not investigated. Also a relationship between cross-pressures, interest and communi- cation behavior may not be a direct one. The possibility remains that cross-pressures have no influence on communication behavior unless interest is taken into account. Given a great deal of interest in the outcome of an election it seems more likely that one's communication behavior will be affected by cross-pressures than given little interest in the election. 38 Again this possibility was not investigated. Another relevant variable may be the individual's relative ability to tolerate cross-pressure. Conceivably peOple with low toleration might tend to withdraw from the situation, whereas those with a high degree of toleration would be able to withstand the cross-pressure and actively seek information to resolve it. This would account for the insignificant findings of the present study. Implications The present study feund that cross-pressures, operationalized as disagreement in voting choice among one's reference groups, is related to deviation from hereditary vote. The greater the disagreement among one's parents and one's neighbors and fellow students, the less he tends to vote the way his parents did. However, cross-pressures were 225 found to be related to amount of communication exposure or incidence of voting. Nor*was amount of communication exposure found to intervene between the cross-pressures and deviation from hereditary vote relationship. The ccurelation was not stronger for either high or low political media users or political conversationalists. In other words those who have referent groups who differ among themselves tend to deviate from the hereditary vote. Correlational analysis can not determine causality, but the implication is there that if people talk to others who have a voting choice different from their parents, they will be affected by it. However, they do not tend to engage in fewer or greater conversations. Nor do they tend to have more or less political media exposure. 39 The reason for this may be that cross-pressures in terms of referent disagreements, which a£g_related to deviation from the hereditary vote, may ngt_be related to internal cross-pressures. In other words, the distinction Pinner (1967) makes between affiliative and attitudinal cross-pressures may be an important one in relation to communication behavior. Affiliative cross-pressures are related to deviation from hereditary vote, but do not seem to be related to communi- cation behavior. However, attitudinal cross-pressures, or vote indecision, might very well be. The possibility exists that a person may be under affiliative cross-pressures, yet not be under attitudinal cross-pressures. The individual may be aware that there are disagreements concerning a voting decision gmgng_his referent groups, yet have no disagreements concerning a voting decision within himself. The individual who is subject to attitudinal cross-pressures might then resolve them in ways related to their communication behavior. Such a possibility is worthy of further research. It would also be worthwhile to determine the relation- ship between affiliative and attitudinal cross-pressures more exactly. It was pointed out earlier that interest in the election is an important variable in relation to cross-pressures, communication behavior and the hereditary vote. Future research in the area should take interest into account. Those under cross-pressures may have differing amounts of media exposure and conversations, depending on the amount of'interest they have in the election. Those who have high interest may tend to resolve attitudinal cross-pressures - vote indecision - by seeking more information. Those with low interest may resolve the indecision by withdrawing from no the situation. Thus, if interest is not controlled for, the relation- ship between cross-pressures and communication behavior would be obscured. In conclusion it can be said that non-significant findings are always anathema to the researcher. However, even non-significant findings can be useful if they lead to further, significant, research. Hopefully this thesis will be a case in point. Ml BIBLIOGRAPHY Berelson, Bernard, Lazarsfeld, Paul F. and McPhee, W. Votipg: A Study of Opinion Formation in a Presidential Campaign, Chicago: University of Chicago Press, 1954. Campbell, Angus, Garin, Gerald and Miller, Warren E. The Voter Decides, Evanston, 111.: Row, Peterson and Co., 195a. Campbell, Angus and Kahn, Robert L. The People Elect a President, Ann Arbor: The University of Michigan Press, 1952. Greenberg, Bradley S. "Voting Intentions, Election Expectations and Exposure to Campaign Information," Journal of Communication, September, 1965, 15, lug-160. Lazarsfeld, Paul F., Berelson, Bernard and Gaudet, Hazel. The People's Choice, New York: Columbia University Press, l9u8. Maccoby, Eleanor E., Matthew, Richard E. and Morton, Anton 8. "Youth and Political Change," Public Opinion Quarterly, Spring, 19514, 18, 23-390 McNemar, Quinn. Psychological Statistics, New York: John Wiley and Sons, Inc., 1962. Pinner, Frank. Cross-Pressure, Written for the International Encyclopedia of the Social Sciences, in progress. APPENDIX A QUESTIONNAIRE POLITICAL INTEREST STUDY Roger D. Haney Dept. of Communication Michigan State University 1. On an average day, how much time do you spend reading your daily newspapers? . minutes 2. About how frequently do you listen to the Egdig? Several times a day Once or twice a day ____ Every other day About once a week Less often 3. How about television...in an average 3255, about how many 3225! do you spend viewing television? hours 4. How many magazines do you read regularly, that is, at least three out of every four issues? Four or more ,____ Three ____.Two One None 5. What was your age on election day, November 8th? Years of Age o-a-.. 10. Are you a registered voter? Yes No If No, skip to question ll. In which state are you a registered voter? In talking to pe0p1e about the election, we find that a lot of pe0p1e weren't able to vote because they weren't registered, or they were sick, or they just didn't have time. How about you, did you vote this time? Yes No If No, skip to question 1;. How many times have you voted up to this time? This election was the first time Two times, including this election Three times, including this election Four or more times, including this election In the election just held, which answer below best describes how you voted for gpggg and national offices? ____ Straight Democrat _____Most1y Democrat A few more Democrats than Republicans _____About equally for both parties A few more Republicans than Democrats ____ Mostly Republicans Straight Republicans 11. 12. 13. 14. 15. 16. During the last week of the election campaign, about how many news stories dealing with political candidates did you read in newspapers? During the last month of the campaign, about how many magazine articles on political candidates did you read? During the last month of the campaign, about how many political talks or shows on T.V. (of 15 minutes or more) for political can- didates did you see? During the last month of the campaign, about how many political talks or shows on radio (of 15 minutes or more) concerning political candidates did you listen to? During the last month of the campaign, about how often did you have conversations with people about political candidates? Several times a week Once or twice a week Once or twice a month Less often And finally, during the last Epppp Of the campaign, about how often would you say pe0p1e asked you for your opinion on politics? Several times a week ____ Once or twice a week Once or twice a month Less often 17. 18. 19. 4 Please estimate how you think your father has voted in recent elections for gggpg and national political candidates? (If deceased, please estimate his most recent voting) Straight Democrat ____ Mostly Democrat A few more Democrats than Republicans _____ About equally for both parties A few more Republicans than Democrats ____ Mostly Republicans ____.Straight Republicans Please give an estimate of how you think your mother has voted in recent elections for EEEEE and national candidates? (If deceased, please estimate her most recent voting) ____ Straight Democrat Mostly Democrat A few more Democrats than Republicans ____ About equally for both parties A few more Republicans than Democrats _____ Mostly Republican .____ Straight Republican Now think of the people you 532! who liyg near you...people in your neighborhood, in your apartment area, or in your own dorm. Which phrase below best describes how you think, in general, they voted in the last election? ____ Straight Democrat Mostly Democrat A few more Democrats than Republicans About equally for both parties A few more Republicans than Democrats Mostly Republicans Straight Republican 5 20. What about the students at Michigan State who are majoring in the same subject as you? Which phrase below best describes how you think, in general, they voted in the last election? Straight Democrat ____ Mostly Democrat A few more Democrats than Republicans .____ About equally for both parties ____ A few more Republicans than Democrats Mostly Republicans Straight Republican 21. Are you: Male Female 22. Are you: Graduate Undergraduate 23. Are you: Married Single 24. Where do you live: Married Housing Dormitory Off-Campus Housing 25. Are your parents still living? Father Yes No Mother Yes No APPENDIX B EXTENT OF COMMUNICATION BEHAVIOR 47 Voters Registered, Non-Voters Medium Frequency Standard Deviation Frequency Standard Deviation NeWSpapers 2.27 2.27 None 8% 2696 1-3 17 13 ”-6 23 13 7-9 5 0 10-12 23 29 13-15 S 8 16-18 3 3 19-21 12 8 22 and over u 0 T5696 T6696 N=1-3 N=38 Magazines 2.67 2.614 None 1% 1396 1-2 13 21 3-H 10 13 5-6 21 19 7-8 8 3 9-10 7 13 11-12 5 0 13-1“ 10 5 15 and over _l_2_ _13 100% 100% T.V. 1.83 2.09 None 3096 #795 1-2 25 16 3-4 21 8 5-6 11 11 7-8 2 5 9-10 6 8 11-12 2 0 13-1” 1 5 15 and over 2 O rm 155% M8 Voters Registered, Non-Voters Medium Frequency Standard Deviation Frequency Standard Deviation Radio 2.22 2.u6 None 5 896 14296 1 9 16 2 9 13 3 6 10 l} 1+ 5 5 7 3 6 2 0 7 2 3 8 and over 3 8 155% 155% Political Conversation .71 .91 Several times a week 52% 3% Once or twice a week #1 8 Once or twice a month u 39 Less often _3 us 100% 366% Extent Asked For Political Opinion .95 .98 Several times a week 24% 16% Once or twice a week 39 29 Once or twice a month 25 37 Less Often 12 18 100% 100% M'Tlllfilmfllllfl 3 12 ATE U 0 93 Ni 3 VER S ITY L 014 8422 ifilfi ARIES