Does education influence voter turnout?
Elections are usually regarded as the core of a democracy. It is the duty and privilege of the citizens in a democratic country to elect their leader.
One of the most striking things about US elections though is the low voter turnout. While other democratic countries have turnout percentages of 80 percent or higher - Germany for example had a voter turnout of 82.2 percent in its last general election in 19981- voter turnout in the US was at low of 49.8 percent in the 1996 presidential elections2.
So why does one half of Americans stay at home in November? What are the factors behind voting and non-voting?
There are different explanations for the phenomenon of low voter turnout. Miller and Shanks (1996) claim that the main reason is that "the average American is uninterested, uninformed and uninvolved".3They acknowledge that turnout in American elections used to be in a higher level a century and even 30 years ago. A generational exchange in the composition of the electorate has taken place, the politically more interested pre-New Deal generation has been replaced by the post-New Deal generation which is less engaged with politics. Even though, this analysis accounts for the steady decline in voter turnout over the last 30 years, it does not explain why the turnout level was always below the figures of other democracies.
Teixeira (1992) sees the reason for low voter turnout as a consequence of the high costs and comparably low benefits of voting.4In his answer to the question why people vote he mentions the symbolic and instrumental benefits of voting. However benefits of voting are actually low due to the American political system and since voting is not a zero-cost activity - US citizens have to register to vote - most people tend not to vote. So, voting seems to be an affair where those who benefit the most from doing it take part. Who are those who benefit from voting? How can they be classified demographically?
Wolfinger and Rosenstone (1980) address this question by looking at turnout rates dividing people by components of socioeconomic status (SES)5such as education, occupation or income. According to them, those SES variables differ in their impact on turnout.
Starting from here, this study will examine the particular effect education has on turnout.
The notion that education is strongly connected to turnout is usually acknowledged. Even Miller and Shanks accept that "it is hard to imagine questioning the effect of education - be it direct or indirect - on turnout"6.
In what ways are the phenomena education and voter turnout connected?
Generally, it is assumed that more formal education results in a higher probability of voting. People who have been to college are regarded as having received more information about politics than those, who did not attend college, and are therefore more likely to turnout. They have fewer difficulties mastering the bureaucratic registration procedures and most likely have been taught some kind of feeling of voting as a civic duty.
On the other hand, people who have not received education beyond high school are more likely to be estranged from the political life in general, registering and voting in particular, due to a lack of understanding.
Sometimes it is supposed that people who are in higher occupations are more likely to vote for they are often in direct contact with politicians. Even though education might be connected to occupation and income, it is not necessary the case7and therefore I will concentrate solely on education, which I reckon to be the strongest SES-variable, disregarding the others. Since the relationship of those variables is one- directional - education might influence occupation or income, while it does not work the other way around
- it is fair to put the emphasis on education.
After considering all those explanations, a hypothesis can be deduced.
The hypothesis is a statement that can be tested and will be found to be true if the theory was correct. Furthermore, a good hypothesis is empirical and not normative, plausible, general and specific8. A potential hypothesis would be: The number of years of formal education affects one's chances of turning out to vote9. While this kind of hypothesis, which is called a correlative hypothesis, assumes that there exists a relationship between the two phenomenon or variables (a term which will be explained more clearly later), it does not specify the nature this relationship10.
For I expect the nature of the relationship to be of a certain kind, I am going to propose a directional hypothesis. In this case, this would either Less years of formal education decrease the chance of turning out to vote or The more formal education a person has received, the more likely that person is to vote. In both these hypotheses there is a direct relationship between education and turnout, meaning an increase of one variable leads to an increase in the other. Each of the hypotheses is empirical (e.g. they do not imply that people with more education should vote), general (not only concentrating on one special election or one group of people) and specific (increase in one leads to an increase in the other).
Since both hypotheses seem to fit the purpose of the study and one has to be chosen, I will go on to deal with the one relating more education to higher turnout.
Before testing this hypothesis though, the phenomena, whose specific relationship will be examined, have to be accurately defined. In political science, the words that are used to describe these attributes are called concepts11and the act of defining them is called operationalisation.
As mentioned above, the two phenomena between which there is an assumed relationship are called variables. The phenomenon that is supposed to influence other phenomena is the independent variable.
Dependent variables on the other hand are those which are influenced by or dependent on the independent variables. In this study, education is the independent variable, while voter turnout is the dependent one. Often there are intervening variables in between the independent and dependent variable that help to explain the process by which one influences the other12. Those might include the ability of mastering the bureaucratic registration procedures or a higher interest in politics. These intervening variables are identical with the explanations offered in the first place why there seems to be a connection between education and voter turnout. If the theory and hypothesis in the end turn out to be wrong, those explanations will be proven wrong too. Occupation or income would also be possible intervening variables in between education and voter turnout.
The operationalisation of these concepts does not only include defining the independent and the dependent variables but also the specification of other concepts such as person and voting/turnout. First of all, what is meant by education? Is it the education a person receives from his/her parents or external factors such as television, books or people outside the family when growing up? Or is it the kind of formal education a person receives in educational institutions such as schools or universities? Since the hypothesis in question relates the years of formal education to voter turnout, it is evident that the study will deal with the length a person has been educated in schools or universities and not domestic education.
The usual way to look at this is to categorise people by years of school completed. These categories might be: less than 5th grade, 5th to 8th grade, 9th to 12th grade/no diploma, high school graduate, some college or associate degree, bachelor's degree and advanced degree13.
Voter turnout, the dependent variable, is not as easily definable. It has to be clarified what voting means and what kind of elections will be dealt with.
Will only on-year (presidential) elections be considered or also elections in off-years, elections at state or local level? What is meant by voting? Has a person to vote for all the offices and propositions on the ballot or is it enough, for example, to only have voted for Senate but not for the President? Most scholars in the past have concentrated on presidential elections and a vote for the president. Presidential elections have higher turnouts than off-year elections and in contrast to local elections they are demographically more representative.
More over, the presidential is the most interesting of all elections and it is probably the election where there is the most at stake. Congress elections take place every two years but the President stays at least for four years.
Finally, after clarifying how the analysis of education and turnout is being operationalised, there is a need to specify which people the study is going to deal with. The first question is whether to deal with the entire Voting Age Population (VAP) or just with the eligible electorate.
The term Voting Age Population refers to the total number of persons in the United States who are 18 years of age or older regardless of citizenship, military status, felony conviction, or mental state. In comparison here to, the number of eligible voters excludes all people who are not allowed to vote:
resident aliens, convicted felons and patients of mental institutions. Since neither the Bureau of Census nor any other organisation can define with complete accuracy exactly how many eligible voters there are in the United States14, this study will concern itself with data referring to the VAP. The second question in reference to the people the study is interested in, is that of registration. There is a significant difference if a person did not register and consequently did not vote or if a person actually registered but then did not vote. There might be a difference in registration turnout between the different education categories, but not a difference in turnout among those registered. In the end it might turn out that education does not affect voting turnout but actually registration turnout.
Therefore this study will be concerned as well with overall voter turnout of the educational categories as with their registration rates as with the turnout among those who have registered. Considering this new intervening variable, the hypothesis will have to be revised to The more formal education a person has received, the more likely that person is to register and even more likely vote. The next step in proving this hypothesis true is to come up with a research design that shows how the goals of the study are intended to be fulfilled15. This includes the collection and interpretation of data. There are two ways of data collection that would suit this sort of study. On the one hand, it might be possible to obtain interview data for primary analysis by face-to-face questioning. On the other hand though, one could also conduct analysis of data, which is based on surveys and interviews by other scholars or organisations. The latter method might save time and money but it is possible that the existing data does not fit the needs of the survey.
If one were to conduct a survey on the connection between education and turnout, the first step would be drafting a questionnaire which elicits quickly and constantly the desired information. A good questionnaire has well-worded questions and appropriate question types and question order16. The questions should not be ambiguous, double-barreled, vague, leading or inappropriate.
The question type most appropriate for this survey is the closed-ended one. Close-ended questions provide the respondent with a set of answer to choose from. Successful completion of an interview or questionnaire is often directly connected to the order of questions. If the interviewee gets the feeling that the questionnaire is about questioning him/her rather than to attain some general knowledge, he/she is likely to terminate the interview. Therefore the first few questions are so-called ice-breakers, general questions that can be easily answered such as demographic questions, while specific and personal question are to be put to the end.
Another factor for the success of the survey lies in the questionnaire design and the way the survey is conducted. A questionnaire has not only to be self-explanative, but also easy and attractive to complete properly.17
While all of the three basic ways of conducting an interview - mailed questionnaire, telephone interview and personal interview - have their pros and cons, they all have one thing in common: they must produce a high response rate in order for the survey to be representative.
The questionnaire for this survey should start of with some general, demographic questions concerning age, gender and family with the following question about educational attainment bridging the way to the more specific and sensitive question:
"Which of the following statements concerning education is true for you: less than 5th grade, 5th to 8th grade, 9th to 12th grade/no diploma, high school graduate, some college or associate degree, bachelor's degree and advanced degree?"
The other main questions of this survey would then be "Did you register to vote in the last federal election?" and "If you did register, did you actually vote for president?" Further possible questions, concerning the reasons for non-registration or non-voting, are "Which of the following reasons for not registering to vote is true for you: not a citizen; failing to meet residence requirements; missed closing date; did not know where to register; was purged from registration rolls; not interested; other reasons" and "Which of the following reasons for not voting despite of registering is true for you: had no way to get to polls; could not take time off from work/school/too busy; out of town or away from home; sick, disabled, or family emergency; did not prefer any of the candidates; not interested; forgot to vote; lines too long at polls; other reasons."18
One problem about this questionnaire though is its fixation on only two topics, education and voter turnout. Despite of some ice-breaking questions in the beginning, the respondent could easily get the impression that the survey is only on his/her political behaviour and might therefore either misreport or in the worst case, terminate the interview.
Restricted representativeness and costs are further weaknesses of this potential survey. If the survey is to be representative it will not be enough just to interview the people in one neighbourhood. People from all different demographic groups have to be interviewed. In order to do this, one would either have to spend a lot of money on telephone interviews and mailed questionnaires or hire trained and experienced interviewers to conduct personal interviews.
Weighing the advantages of a primary analysis - formulation of questions best suited for the survey - against its disadvantages - high cost, high risk of terminating the interview - it would be more efficient, less expensive and less time consuming to turn to secondary analysis of existing data, collected by record keeping agencies such as US Census Bureau or the American National Election Studies (NES). Particularly the November Voting and Registration Supplement File from the Bureau of the Census Current Population Survey (CPS) produces data which in size and reliability is even superior to surveys such as the NES.
The CPS is a monthly labor force survey in which interviews are conducted in approximately 48,000 households across the US.19Starting in 1964, the November survey of even-numbered years features a supplement asking question about registration and voting20. The large sample size of data of every demographic group, combined with the data concerning educational attainment, which is obtained
through the regular survey, makes this set of data the best suited for this study.
One of the major advantages of the CPS is the fact that it is not a survey just about political involvement and voting behaviour. Since the questions about education, voting and registration make up just a small part of the survey, people are more likely to answer the question honestly, rather than overreporting on their turnout. A feeling of embarrassment for not voting is probably less likely to be arisen here than in a survey dealing just with voting behaviour.
The use of this running record which covers a long time-span allows a longitudinal analysis21of the hypothesis while a single survey might produce a result which is just caused by short term forces at the time the survey is carried out.
Still, the CPS has its shortcomings too. One of it is the regular over-estimation of voter turnout. While the official turnout is based on the actual votes cast for the President, the CPS percentage records voting for any office or public issue.22According to Teixeira, this overestimation is fairly small compared to other data sets23and therefore the CPS data is superior to any other data.
Sometimes, the data from the CPS or other surveys is not organised and weighed against each other in the way a study needs it to be. In this case, data has to be combined and tabulated in the way suited best for the study.
This is partly the case with the data in table 1-1, which was tabulated by the US Census Bureau from data obtained through the CPS over the years.
Table 1-1. Reported Voted and Registered by Educational Attainment: November 1964 to 1996
Abbildung in dieser Leseprobe nicht enthalten
Note: Prior to 1972, data are for people 21 to 24 years of age with the exception of those aged 18 to 24 in Georgia and Kentucky, 19 to 24 in Alaska, and 20 to 24 in Hawaii.
Registration data were not collected in the 1964 Current Population Survey.
NA Not available.
Source: Current Population Reports, Series P20-504, "Voting and Registration in the Election of November 1996," and earlier reports.
The data in table 1-1 allows us to answer the question if the first hypothesis - The more formal education a person has received, the more likely that person is to vote - is true or not (tentatively speaking, it is true) and it shows that the same is true for registration rates. However, it does not directly prove the revised hypothesis (The more formal education a person has received, the more likely that person is to register and even more likely vote) to be true or false, for it there is no direct data of turnout for those who registered.
It is no problem though to calculate the desired percentages from the given data. The result would be result in a table like table 1-2.
Table 1-2. Reported Registered Who Voted by Educational Attainment: November 1964 to 1996
Abbildung in dieser Leseprobe nicht enthalten
Source: Author's tabulation of table 1-1 data
A tentative conclusion drawn from this aggregated data is that the revised hypothesis is true: people with more years of formal education are not only more likely to register, but they are more likely to actually vote than registered people who received less education.
This conclusion might be explained by the above mentioned proposals such as a feeling of civic duty or a higher interest in politics.
Of course, after collecting and evaluating all the data, it might have turned out that the hypothesis is proven wrong. In this case, the failure of the hypothesis has to be explained. One explanation that comes to mind is that of educated people realising that their preferred candidate has no chance of winning and therefore they do not vote or even register. Recognition of the corruption of politics and a consequential turning away from it could be mentioned as another explanation.
If a hypothesis is proven wrong, it either has to be reformulated or totally abandoned and a new hypothesis has to be deduced.
In this study, perhaps the opposite of the hypothesis is true and the less educated people are the more likely they are to vote because they are not as disillusioned about politics.
Another possible result is that there is no connection between voter turnout and education at all. Registration and voting percentages might be higher for a college graduate than for someone with a Bachelor's degree but their registration turnout could be below that of a high school graduate. If this is the case, a new hypothesis concerning the relationship of two other variables (e.g. region and turnout) has to be formulated.
Does education influence voter turnout? 9
Federal Election Commission, "A Few Words About Voting Age Population (VAP)," <http://www.fec.gov/pages/vapwords.htm>, 2000.
Federal Election Commission, "National Voter Turnout in Federal Elections: 1960-1996," <http://www.fec.gov/pages/htmlto5.htm>, 2000.
Friedhelm, Jerry Warden. Where Are The Voters?. Washington D.C.: The National Press Inc., 1968.
Johnson, Janet Buttolph and Richard A. Joslyn. 1995. Political Science Research Methods 3rd ed. Washington D.C.: Congressional Quarterly Inc., 1995.
Miller, Warren E. and J. Merrill Shanks. The New American Voter. Cambridge: Harvard University Press, 1996.
Statistisches Bundesamt, "Endgültiges Ergebnis Bundestagswahl 1998 - Bundesgebiet," <http://184.108.40.206/wahlen/ergeb98/d/t/bun999_02.htm>, 2000.
Teixiera, Ruy A. The Disappearing American Voter. Washington, D.C.: The Brooking Institution, 1992.
US Bureau of the Census, "Voting and Registration in the Election of November 1996," <http://www.census.gov/prod/3/98pubs/p20-504.pdf>, 1998.
US Bureau of the Census, "CPS Voting and Registration Supp - 1996 Public Use File Overview," <http://www.bls.census.gov/cps/vote/1996/suppovrw.htm>, 1999.
Wolfinger, Raymond E. and Steven J. Rosenstone. Who votes?. New Haven: Yale University Press, 1980.
1Statistisches Bundesamt, "Endgültiges Ergebnis Bundestagswahl 1998 - Bundesgebiet," <http://220.127.116.11/wahlen/ergeb98/d/t/bun999_02.htm>, 2000.
2US Bureau of the Census, "Voting and Registration in the Election of November 1996," <http://www.census.gov/prod/3/98pubs/p20-504.pdf>, July 1998, p. 2.
3Warren E. Miller and J. Merrill Shanks, The New American Voter (Cambridge,1996), p. 36.
4Ruy A. Teixeira, The Disappearing American Voter (Washington D.C., 1992), p. 23.
5 Raymond E. Wolfinger and Steven J. Rosenstone, Who votes? (New Haven, 1980), p. 14.
6Miller and Shanks 1996, p. 56.
7Wolfinger and Rosenstone 1980, pp. 23-30.
8Janet Buttolph Johnson and Richard A. Joslyn, Political Science Research Methods 3rd ed., (Washington D.C., 1995), pp. 57-63.
9ibid., p. 46.
10ibid., p. 54-55.
11 ibid., p. 49.
12Johnson and Joslyn 1995, p. 46.
13 US Bureau of the Census, "Voting and Registration in the Election of November 1996," <http://www.census.gov/prod/3/98pubs/p20-504u.pdf>, July 1998, p. 34, table 7.
14Federal Election Commission, "A Few Words About Voting Age Population (VAP)," <http://www.fec.gov/pages/vapwords.htm>, 2000.
15Johnson and Joslyn 1995, p. 111.
16 Johnson and Joslyn 1995, p. 266.
17Johnson and Joslyn 1995, p. 276.
18US Bureau of the Census, "Voting and Registration in the Election of November 1996," <http://www.census.gov/prod/3/98pubs/p20-504.pdf>, July 1998, table 3.
19US Bureau of the Census, "CPS Voting and Registration Supp - 1996 Public Use File Overview," <http://www.bls.census.gov/cps/vote/1996/suppovrw.htm>, January 28, 1999.
20 Teixeira 1992, p. 186.
21Johnson and Josyln 1995, p. 240.
22US Bureau of the Census, "Voting and Registration in the Election of November 1996," <http://www.census.gov/prod/3/98pubs/p20-504.pdf>, July 1998, p. 2.
23 Teixeira 1992, p. 187.