Sorry, but copying text is forbidden on this website!
The power of America lies within the heart of its people and the ability to have their voices heard. One of the best ways to accomplish this is through an electoral vote. By voting, the people of America or any Democratic country can control the route of the government and the decisions it makes. To decide if those decisions be new and current with society, or kept traditional is why such ability is granted under the 15th amendment. In addition, with voting behaviors determined by Democratic, Republican and Independent parties, it is important to follow the factors that affect one’s decision to identify themselves by certain political party. There are varieties of factors that form these distinct identities such as family, region, racial background, religiosity, and culture while these may be key determining causes, this paper will look at education level more precisely as it tends to be the most internally and externally expansive characteristic that applies to all or most individuals’ as well as their in voting behavior or political identification.
Thus, this paper is about the relationship between formal education and the affects it has on democratic citizenship, moreover it effect on political affiliation. It considers a question that has preoccupied political philosophers, theorists, and political scientists for centuries in understanding party formation and alignment among the masses, especially in Democratic society. To do this, one must consider the ways and the extent to which education influences how knowledgeable citizens are, how attentive they are, how regularly they vote, how active in politics they are beyond the vote, and finally how tolerant they are of the free expression of unpopular political views. Education is shown to have played a great deal in the voting behavior for this last presidential election voting. Statistics released from U.S. census Bureau’s American Community Survey and mentioned in foxbussiness.com (2012) showed how states, best and worst educated voted either republican or democratic, which residents over the age of 25 with a college degree or higher within the top ten most educated states all voted for president Obama in this last election (Democratic party) while inversely, 9 of the lowest 10 educated states in the U.S. voted for governor Romney (Republican).
While this information, looking solely on education by state alone, cannot give us a clear answer as to how significant education and voting behavior relate, it acknowledges that there is a correlative pattern between these phenomena. Another general social survey provided by secularright.org (2009) displays party affiliation as a function of education in whites. The statistics study breaks down Republicans, Democrats and Independent based on those with less than $150k net worth and more than $150k net worth (secularright.org 2009). With people separated based on level of education by less than college or college degree, the proportion of Democrats is highest for whites with the combination of college education or higher and net wealth of less than $150,000 (secularright.org 2009). Whereas, the proportion of Republican identification was highest among whites with the combination of college education or higher and with a net wealth of greater than $150,000 (secularright.org 2009).
Studying these two phenomena together is very beneficial in gaining a deeper understanding of how formal education influences what citizens know, how much they pay attention to politics, and how politically active and tolerant they are, with important implications for both democratic theory and long-term efficient educational policy. Additionally, for those who are working for a campaign in any level, determining whether education plays a role in political affiliation will help develop an idea in how citizens will vote serving as a campaigning prediction tool for political candidates to garner more votes for their political party. As society rapidly changes with an influx of new ideas and issues, studying the college educated and those who are not will help evaluate behaviors and attitudes towards the government, ultimately, clearing the way to adaption into a modern society that perhaps offer remedies of educational and voting discrepancies or even close the gaps between political ideology or identification. Hence, this paper proposes the research question: How does education level influence political party identification.
While there are many hypotheses and theories as to why education is important for democratic citizens, there is common and consistent agreement within the literature since the 1970s. There is consistency in the belief that education provides both the skills to become politically engaged and the knowledge to understand and accept democratic principles leading to correlative effects on party identification on both individual and aggregate levels (Golebiowska 1995; Galston 2004). Angus Campbell and Philip E. Converse (1972) describe education as the universal solvent, strongly and positively correlated with a host of valued civic attitudes and behaviors such as political party or ideology formation. One literature found that despite the steady increase in the average years of formal education attained by Americans, and the shrinking gap in education among citizens leading to a strong connection between individual measures of good citizenship and education, contradictory findings or aggregate indicators of civic involvement (electoral participation) show little evidence of increasing, and in many instances they have declined over the past several decades (Nie, Junn, and Stehlik-Barry 1996).
Nonetheless, despite this apparent paradox, Norman H. Nie, Jane Junn, and Kenneth Stehlik-Barry (1996) explored party alignment and found that it has been a consistent guide among people with differing educational backgrounds, even with voting apathy involved, when it is centrally affected by education’s ability to increase the cognitive capacity of citizens. This in turn, gives people the motivation and tools to appreciate both the logic of democratic norms and the potential legitimacy of views different from their own in that civic engagement. It also displays the ability to understand one’s own interests and effectively pursue them through the political system, is largely connected to education through one’s social position and the resulting access to centers of political power and decision-making dividing people across party lines (Nie et al. 1996).
The theory that colleges and universities function as important socializing agents in the development in tolerance, and a belief in civil liberties and equality would not normally lead one to hypothesize a positive correlation between educational attainment and identification with the conservative republican party. However, in most national and regional studies on voting behavior, exempting the south, reveal such a relationship is typically observed (Knoke and Hout 1972). In attempts to uncover social and demographic factors in American political party affiliations, they delimited a set of causal variables. The first of which observed that the effects of socioeconomic origins on party preferences were completely mediated by respondents’ socioeconomic status achievements gained from education and occupational status (Knoke and Hout 1972).
Their analysis indicated that change in the variables might cause change in individual party affiliations over time. For example, the higher the educational or occupational status, the more republican respondents tend to be. David Knoke and Michael Hout (1972) explained if the functional form of these relationships remains stable, individuals who acquire more education or are upwardly mobile in the occupational hierarchy, would change their affiliations towards the Republican Party. Moreover, Terry S. Weiner and Bruce K. Eckland (1979) conducted similar research and express that the positive correlation between educational attainment and being a republican does not reflect the direct effects of higher education, but that it can be indirectly accounted for by two social processes. Both of which are unrelated to the direct influence that college has on political values, whose similar-minded political ideology plays a role leading to party identification.
The first of these being selective recruitment into higher education of persons from middle and upper-middle class republican backgrounds, paired with the tendency for party preferences to be acquired early in life from parents and carried over into adulthood by means of inertia (Weiner and Eckland 1979). Thus, if both educational attainment and adult partisan loyalties have common antecedents rooted in the family of origin, the association between going to college and voting could be entirely spurious, yet evidence still finds an indirect co-variation. Terry S. Weiner and Bruce K. Eckland (1979) acknowledges a second set of conditions that could very much produce a positive correlation between years of schooling and identification with the Republican Party.
Upon the completion of formal schooling people tend to adjust partisan loyalties to match those of the dominant status groups to which they belong and where at times membership depends on educational attainment. In turn, higher educational is a powerful determinant of occupational status and social mobility through marriage. If socialization were to continue after college and conservative political-economic values are associated with high status jobs and upper middle-class status, or conversely, if the Democratic Party is the worker’s party, educational attainment would be expected to correlate with the Republican Party (Weiner and Eckland 1979).
Another good determinant in understanding the relationship between education and party identification is to take into account of attitudes towards welfare reform such as governmental assistance or welfare often a contentious topic in modern day democracy in which the direction is often positive among less educated and lower income individuals with Democratic Party alignment (Abner 2011). Inversely, Kristin S. Abner (2011) determined negative attitudes towards welfare and big government are often found among people with higher education and ensuing republican notions even when controlling for income, age, employment status, gender, and race.
In other words, respondents residing in neighborhoods with a higher concentration of African-Americans and other non-whites (with little to none education) are more likely to endorse positive attitudes toward welfare policy aligning with the Democratic party that appeals to their interests, plight, or goals with income serving as a interaction effect (Abner 2011). Given the exploratory nature of this study, the author concludes with implications for such findings and proposes future research in the relationship between with education level and party identification in which the findings shows indirect if not direct linkages due to interactive effects from other controls (Abner 2011).
William A. Galston (2004) does not connect education to a specific political party, but rather explains it is vital in keeping political interest in America’s young adults, affecting the voting turnout for the future; influencing the vote of a democrat or republican. He explains why civic education is important by saying education by civic knowledge promotes support for democratic values. The more knowledge we have of the workings of government, the more likely we are to support the core values of democratic self-government, starting with tolerance. Civic knowledge promotes political participation. All other things being equal, the more knowledge people have, the more likely they are to participate in civic and political affairs. In addition, civic knowledge helps citizens to understand their interests as individuals and as members of groups.
There is a rational relationship between one’s interests and particular legislation: the more knowledge we have, the more readily and accurately we connect with and defend our interests in the political process (Galston 2004). Thus, education can support the core values of democratic self-government, promote political participation in our governmental affairs and provide an understanding of their own interests and political legislation, which are important as a result of their relation to political participation and affiliation.
The literature above displays empirical findings, providing many consistent evidence that tie education and party identification together whether directly or indirectly exists, yet there are conflicting evidence indicating that such co-variation is spurious or weak when controls are added into the equation. These literatures will be valuable to future research as well as providing cumulative and empirical evidence on education’s effect on party identification. However, outdated data and inconsistencies or contradictions exists within the findings between education and party identification on both individual and aggregate levels suggesting the need for further research and retesting to see the actual direct or indirect relationship between the variables if evident. Despite this, the findings still suggest an empirical relationship between the variables do exist. There is evidence that absolute levels of educational attainment is vital to shape levels of political knowledge, which in turn affect the acceptance of democratic principles, attitudes toward specific issues, and political participation. This paper’s theory will replicate those works that found a correlation between education and one’s political party affiliation.
It is theorized that the higher the level of education one receives, the more politically engaged they are with current events consistently on their minds that in turn align with a particular political party that will advance similar-minded goals and interests (such as the Republican party). Families with higher education levels at a college degree or higher make up a large percentage of the middle and upper class and desire less economic changes or interference to be brought upon their lives such as governmental assistance and higher taxing codes that will create greater economic negative costs or burdens for them. Republican views towards welfare are similar in the belief that because of religious organizations, charities, and fraternal benevolent societies in fostering benevolence and patriotism, they should not be subject to taxation, and donations to them should continue to be tax deductible. In this sense, higher educated people desire small government intervention and practice conservatism aligning with the Republican Party due to similar minded economic conservatism agendas as well as social morals. On the other hand, less educated people tend to align with the Democratic Party due to similar-minded attitudes towards government’s role being much highly involved in providing welfare assistance or benefits, increasing one’s reliance on the government as lower educated families often lack the income to provide for a better standard of living.
Since less educated persons need more assistance from the government through benefits like food stamps will in turn align with the Democratic party’s platform of social equality. These little-to-none educated people ask government to play a larger role in the citizen’s lives, ultimately for big government involvement and join the Democratic Party to advance and preserve a higher standard of living (including education, health welfare, and so forth). Such a relationship forms the hypothesis that, the higher the level of education one has, the more likely he or she is to identify him or herself as a Republican.
As previously stated the original hypothesis (H1) states, “The higher the level of education one has, the more likely he or she is to identify him or herself as a Republican.” A second, alternative hypothesis which explains the covariance between a control variable (X2-race) and the dependent variable (Y-political party identification), symbolized as H2, states, “Individuals that identify themselves as white Americans are more likely to affiliate themselves with the republican party, whereas those that identify themselves as black are more likely to be a democrat.” The null hypothesis (H0) states that there is no relationship or statistical difference between the variables. The primary independent variable is highest form of education degree, the control variable is race, and the dependent variable is political party affiliation. The unit of analysis used in this research is people.
The dataset in which the information is found is GSS2008.sav. In examining the respondent’s highest form of education degree (X1), there are five categories based on an ordinal level of measurement: (0) limited high school, (1) high school graduate, (2) junior college, and (3) bachelors degree, and (4) graduate degree. By looking under the frequency column, one can see that 293 and a valid percent of 14.5% of the people identified their level of education as 0 (some high school). 1025 and a valid percent of 50.7% of the people identified their level of education as 1 (high school degree). 180 and a valid percent of 8.9% of the people identified their level of education as 2 (junior college associate degree). 338 and a valid percent of 16.7% of the people identified their level of education as 3 (bachelors degree). 187 and a valid percent of 9.2% of the people identified their level of education as 5 (graduate degree). The modal category (most frequently chosen) category is 1 (high school degree), which was found by locating the largest number under the frequency column which it had 1025 respondents and a valid percent of 50.7%.
The median is category 1 (high school) with a cumulative percentage of 50.7 that divides the distribution in half. The mean is 1.56 that falls between categories 1 (High School) and 2 (Junior College). This graph displays high external validity as only 1 respondent, representing 0.1% of the data are missing out of the total of 2023 making this sample size representative of other targeted populations observing the same phenomena. It also displays high internal validity because it only has a few categories that are mutually exclusive and exhaustive. The categories measure what they are supposed to measure, and the number of respondents for each category is roughly equal in their distribution of responses, so recoding is not necessary. Tables: I and 0 here
When looking at the respondent’s political party affiliation (Y) there are seven categories based on a nominal level of measurement: (0) Strong Democrat, (1) Weak Democrat, (2) Independent Democrat, (3) Independent, (4) Independent Republican, (5) Weak Republican, and (6) Strong Republican. By looking under the frequency column, one can see that 379 people identified their political party affiliation as 0 (Strong Democrat) with a valid percent of 19.2. 344 people identified their political party affiliation as 1 (weak democrat) with a valid percent of 17.4. 244 people identified their political party affiliation as 2 (independent democrat) with a valid percent of 12.4. 316 people identified themselves as 3 (independent) with a valid percent of 16.0. 163 people identified their political party as 4 (independent republican). 317 people identified themselves as 5 (weak republican). 211 people said the affiliated with 6 (strong republican). The modal category (most frequently chosen) is 0 (strong democrat), which is found by locating the largest number under the frequency column that is 379 with a valid percent of 19.2.
The median category is 3 (independent) with a cumulative percentage 65 that divides the percentage in half. The mean is 3 that roughly falls under category 3 (independent). This graph displays low external validity because 49 respondents representing 2.4% of the data are missing out of the 2023, making representativeness of the categories questionable. It displays high internal validity because the numbers of respondents’ responses are equally or evenly distributed between the categories since the categories are recoded into seven classifications that differentiate between strong and weak in both parties, sustaining mutual exclusiveness and exhaustiveness. Tables: III and 0 here
When analyzing the respondent’s race (X2) there are 2 dichotimal/nominal categories. These consist of (0) white and (1) black. A frequency of 1570 people identified their race as 0 (white) with a valid percent of 85.7. A frequency of 262 people identified their race as 1 (black) with a valid percent of 14.3. The modal category (most frequently chosen) is 0 (white), found by locating the largest number under the frequency column that is 1570 with a valid percent of 85.7. There is no median category with only two, white and black. The mean is .14, which falls between categories 0 (white) and 1 (black). This graph displays low external validity because a total of 191 respondents, representing 9.4% of the data are missing out of the total of 2023 that could make representativeness questionable. This graph displays high internal validity because it is evenly distributed and because the categories are mutually exclusive and exhaustive, measuring what they are supposed to measure, so no recoding is necessary. Tables: II and 0 here
The correlation between an individual’s level of education degree (X1) and their political party affiliation (Y) is a very weak to none, positive relationship of .025 and this relationship is not statistically significant based on 95% confidence because the observed sig value is .266, which is greater than the expected sig of .01. Therefore, the null hypothesis cannot be rejected. In regards to an individual’s race (X2) and their political party identification (Y), there is no positive relationship with .0 and this relationship is not statistically significant based on a 95% confidence because the observed sig value is .00, which is the same in relation to the expected sig of .00. Therefore, the null hypothesis can be rejected. Between an individual’s level of education degree (X1) and their race (X2), there is a weak to little and positive relationship of .405. This relationship is statistically significant based on 95% confidence because the observed sig is equal to .000, which is less than the expected sig of .05. Due to the fact that both are statistically significant together increases the chances of a spurious relationship. Table IV here
According to Table V, between an individual’s level of degree (X1) and their political party affiliation (Y), the adjusted R² is .001. This means that formal education explains 1% of the variance in strength of political party affiliation, leaving 99% of the variance unexplained. This is a terrible model fit summary because it leaves 99% of the variance unexplained, indicating that there are other factors or variables that could better explain affiliation to a political party. Table V here
The unstandardized b coefficient, or slope, in the first bivariate regression for X1(level of degree education) is .043. This means that for every one unit change or increase in education, there is a .043 increase in political party affiliation (from strong democrat to strong republican). This indicates the higher the education degree a person has, the more likely he or she will be a democrat. According to the t-statistic rule, the relationship between X1and Y is not statistically significant at 95% confidence because the observed t-score is 1.113, which falls within or below the critical value of ±2.461. Thus, we cannot reject the null hypothesis. The relationship between X1and Y is also shown to be statistically insignificant based on the sig-rule, which requires the observed sig to be below the expected value of .05, based on the 95% confidence interval. The observed sig value between X1 and Y is equal to .266, which is greater than the expected value of .05. Therefore, we cannot reject the null hypothesis. The relationship between X1 and Y is again shown to be statistically insignificant by observing the confidence interval values of -.033 and .118 including 0 within this range, so the relationship is not statistically significant at 95% confidence and the null hypothesis cannot be rejected while rejecting hypothesis one instead. Table VI here
For the second bivariate regression, consisting of X2 (race) and Y (political party affiliation), there is an adjusted R² of is .111. This means that race explains 11% of the variance in political party identification, leaving 89% of the variance unexplained. This is a better model fit summary because it leaves 89% of the variance unexplained, but indicating that there are other factors or variables that could better explain strength of religious affiliation. Table VII here
The unstandardized b coefficient, or slope, in the second bivariate regression for X2 (race) is -1.958. This means that for every one-unit change in race, there is a -1.958 decrease in political party affiliation (from strong democrat to strong republican). This indicates the more likely an individual is black, the more likely he or she will affiliate themselves as an independent to strong democrat. According to the t-statistic rule, the relationship between X2 and Y is somewhat statistically significant at 95% confidence because the observed t-score is -14.961, which falls higher and above of the critical value of ±2.906. Thus, we can reject the null hypothesis. The relationship between X2 and Y is also shown to be statistically more significant based on the sig-rule, which requires the observed sig to be below the expected value of .05, based on 95% confidence and the observed sig value is .00, which is less than the expected value of .05. Therefore, we can reject the null hypothesis. The relationship between X2 and Y is again shown to be statistically more significant by observing the confidence interval range values of -.2.214and -1.701 that indicate 0 does not fall within this range, the relationship is statistically significant at 95% level confidence and the null hypothesis can be rejected while hypothesis two is rejected instead.
Table VIII here
When looking at the multivariate regression table, one can observe an adjusted R2 of .111 when examining how both X1 (level of education degree) and X2 (race) explain the variance in Y (political party affiliation). Combined together, X1 and X2 accounts for .111% of the variance in political party affiliation, leaving 89% of the variance unexplained. Therefore, this is not a great model fit because it leaves 89% of the variance unexplained, meaning there are many other variables that can better explain the variance in Y. Table IX here
The multivariate regression also reveals an unstandardized b coefficient of -.033 for X1, meaning that, for every one unit increase of formal education, there is a .033 decrease in political party affiliation (from strong democrat to strong republican) when controlling for race. When controlling for race, the higher one’s education degree, the less likely he or she will be strong republican (or greater education equals stronger democrat). The unstandardized b coefficient in regards to the control variable (X2) is -1.968 and this indicates that for every one unit increase of total family income, there is a -1.968 increase in democrat when controlling for level of education. This means that even when accounting for education, the more likely oneself is black, the more likely he or she will be a strong democrat.
In determining the statistical significance of X1 on Y and X2 on Y with respect to the controls, one must again reexamine the t-statistic rule, the sig rule, and the confidence interval ranges. First, looking at X1 (level of education degree), it has an observed t-score value of -.833 when controlling for X2 (race) that falls below the critical value of ±1.98, therefore it is not statistically significant on a 95% confidence. The sig value of .405 is well above the expected sig value of .05, meaning again that the relationship between education and political identity while controlling for race is not significant at a 95% confidence. Also, 0 does fall within the confidence interval ranges of -.110 and .044 that makes the relationship between X1 and Y insignificant when controlling for X2 based on a 95% confidence. This means that the null hypothesis cannot be rejected based on 95% confidence and rejects hypothesis one.
Now looking at X2 (race) and its relationship with Y (political party affiliation) when controlling for X1 (level of education degree), it has an observed t-score value of -14.947 that falls above the critical value of ±1.96, and therefore it is more statistically significant at a 95% confidence. The sig value of .000 is not above the expected sig value of .05, meaning again that the relationship between race and party identification when controlling for race is more significant at a 95% confidence. Also, 0 does not fall within the confidence interval range of -2.226 and -1.710 which makes the relationship between X1 and Y insignificant when controlling for X2 based on a 95% confidence. This means that the null hypothesis can be rejected based on 95% confidence and instead accepts hypothesis two. Thus, X1 (level of education degree) has no statistical effect on Y (political party affiliation) whereas X2 (race) does.
In analyzing the Beta (standardized coefficient) levels for X1 (level of education degree), it can be determined that the Beta level for X1 is -.019 which indicates the relationship is negative and very weak to no association when controlling for X2 (race). The Beta level for X2 is -.336 that indicates the relationship is negative and has better association when controlling for X1 (level of education degree). Race seems to have a stronger affect on political party identification than education does when assessing the Beta level. This is relevant because one can conclude that there is statistically significant relationship between the variables since both X2 is statistically significant with the dependent in all bivariate and multivariate regressions, not supporting the null hypothesis based on 95% confidence. Table X here
Based on the above findings, my proposed hypothesis that the greater an individual’s level of education, the more likely he or she is a republican was not confirmed as true based on 95% confidence despite the inconsistencies in the above empirical literature. Therefore, I cannot reject the null hypothesis, so there is no statistically significant relationship between education and political party affiliation. In return, the data included in this research did show that race was an important determining factor, much greater than education, in addressing one’s political identity. Altogether, the second alternative hypothesis which explains the covariance between a control variable race and the dependent variable political party identification, which stated Individuals that identify themselves as white Americans are more likely to affiliate themselves with the republican party, whereas those that identify themselves as black are more likely to be a democrat, did prove to be true.
Therefore, I can reject the null hypothesis, as there is a statistically significant relationship between race and political party affiliation. However, it does show that in general there is no significant relationship between education and political party affiliation. This paper will contribute to empirical literature because it is broad in its scope and observes the relationship between education and party affiliation in general as opposed to most of the previous literature. Hopefully, this research will lead to more empirical trends in this direction to allow for a better understanding of the relationship between the two variables in general. I found the conclusion of this research fairly surprising. I really expected to find a causal relationship between education and one’s political party identification.
Abner, Kristin S. 2011. Determinants Of Welfare Policy Attitudes: A Contextual Level Analysis. Sociological Spectrum, Volume 31, Issue 4, pp. 466-497. UTSA Databases- m.search.serials solutions.com/ (accessed February 16, 2013).
Campbell, Angus, and Phillip E. Converse. 1972. The Human Meaning of Social Change. New York: Russell Sage Foundation. UTSA Databaseshttp://ucat.lib.utsa.edu /vwebv/holdingsInfo?bibId= 181794/ (accessed February 20, 2013).
Foxbusiness.com. 2012. “America’s Best and Worst Educated States.” October 15. http://www.foxbusiness.com/personal-finance/2012/10/15/americas-best-and-worst-educated-states (accessed February 14, 2013).
Galtson, William A. 2004. Civic Education and Participation. Commentary Scholarly Journals PS: Political Science and Politics , Vol. 37, No. 2 (Apr., 2004), pp. 263-266. Published by: American Political Science Association UTSA Databases http://search.proquest.com.libweb.lib.utsa.edu/docview/235182949/abstract?accountid=7122/ (accessed February 19, 2013).
Golebiowska, Ewa A. 1995. Individual Value Priorities, Education and Political Tolerance. Political Behavior, Volume 17, Issue 1, pp. 23 – 48. UTSA Databases- http://www.jstor.org.libweb.lib.utsa.edu/ stable/586556?seq=3/ (accessed February 21, 2013).
Knoke, David, and Michael Hout. 1974. Social and Demographic Factors in American Political Party Affiliations. American Sociological Review, Volume 39, Issue 5, pp. 700 – 713. UTSA Databases http://www.jstor.org.libweb.lib.utsa. edu/stable/2094315?seq=2/ (accessed February 20, 2013).
Nie, H. Norman, Jane Junn, and Kenneth Stehlick-Barry. 1996. Education and Democratic Citizenship in America. Chicago: University of Chicago Press. UTSA Databases- http://ucat.lib.utsa.edu/vwebv/holdingsInfo?bibId=59087/ (accessed February 19, 2013).
Secularright.org. 2009. “ Reality and Reason. Party Affiliation by Wealth and
Education.” July 8. http://secularright.org/SR/wordpress/party-affiliation-by-wealth-education-2004-2008/ (accessed February 16, 2013).
Weiner. Terry S. and Bruce Eckland. 1979. Education and Political Party: The Effects of College or Social Class?. American Journal of Sociology, Volume 84, Issue 4, pp. 911 – 928. UTSA Databases http://www.jstor.org.libweb.lib.utsa.edu/stable /2778029/ (accessed: February 13, 2013). Frequencies