Politics In America

Politics In America: A Data-Driven Analysis

Section One: Introduction

Politics are incredibly important to any functioning society. Essentially, politics is the practice of governance over a group of people. Governance maintains societal order and the function of a state. Politics affects everyone, from billionaires to the common man. The impact of the decisions of our federal, state, and local governments is felt on a daily basis. As people hold different priorities and values, they have different views on how government should be run. Our American system is dominated by two parties, the liberal-leaning Democrats, and the conservative-leaning Republicans.

The reason I did my research paper on political views in America is because of how intertwined American society and American politics are. From social media to education, to television, to buying property, the American political system has immense influence. I have seen this firsthand. Growing up as a member of Gen Z in post-2016 election America, I have seen the significance politics holds in everyday life. It feels that as I have gotten older, political polarization has increased. From the news and media I consume to everyday conversations, political views play a factor. It was my goal with this paper to investigate the causes of variance in political views in the United States and the truth of common beliefs and stereotypes in our political system. My main research questions include the following: How does one's age affect their political views? How does one's sex affect their political views? How does one's wealth affect their political views? Do college-educated Americans lean in one direction politically? Do religious groups tend to have political leanings? Does sexual orientation play a role in political views?

Section Two: Previously Published Sources

As politics and values are such important facets of American society, there has been extensive research done before on the causes of the variances of political views. Johnathan C. Peterson, Kevin B. Smith, and John R. Hibbing from the University of Chicago investigated the aphorism "if you are not a liberal at 20 you have no heart and if you are not a conservative at 30 you have no brain." The academic article utilizes data from the Michigan Youth-Parent Socialization Panel Study, which documented the behaviors and attitudes of a random sample of high-school seniors and their parents every decade in the mid-to-late 20th century. Their research found that 34% of liberals at age 26 were conservatives by 50, and only 19% of conservatives at 26 were liberals by 50 (J. Peterson, K. Smith, & J. Hibbing, 2020). This panel study allows an interesting insight into the development of political views by age. Nearly double the number of liberals at age 26 switched political views by the time they became middle-aged adults than conservatives at age 26. This suggests a strong correlation between older age and a conservative political view.

Ruth Igielnik wrote an academic article titled, "Men and women in the U.S. continue to differ in voter turnout rate, party identification" in 2020, using Pew Research Center data. The data shows a gap in both liberal and conservative identification among the two sexes. Pew Research Center data found that only 38% of registered female voters had Republican views, compared to 56% of the same group identifying as liberal. The contrast is flipped when it comes to registered male voters, where 50% identify as conservative and 42% identify as liberal (Igielnik, 2020). Previous academic articles suggest strong correlations between sex and political leanings.

A 2014 Pew Research Center study asked for the political views of over 25,000 Americans. The data found significant trends between family income and political identification. First, there was a negative correlation between liberal identification and income. 35% of respondents in households that had an income of less than $30,000 identify as liberal. This figure drops by 6% to only 29% when compared to liberal identification in respondents from households earning over $150,000. The most significant relationship in the Pew Research Center's survey was the positive correlation between conservative identification and increased family income. The under $30,000 group had a conservative identification of only 17%, a 12% difference compared to the 29% seen in the $150,000+ group. Interestingly, moderate identification is the most common among the family income groups. The percentage of moderate identifications averages 39.4% among the seven groups (Pew Research Center, 2014).

The Pew Research Center released a cross-sectional survey, in which political beliefs were asked every decade since 1994. Strong relationships can be seen between liberalism, conservatism, moderatism, and education. The higher level of education a respondent has, the more likely they are to identify as liberal. In 2015, 54% of respondents with postgraduate experience identify as liberal. This percentage drops to 44% for respondents with graduate degrees, 36% for respondents with some college, and 26% for respondents with high school or less. A less drastic, yet opposite, trend can be seen for conservative identification;, in 2015, 24% of respondents with postgraduate experience identify as conservative. This rate climbs to its peak, which is college graduates, with 29% identifying as conservative. The high school or less and some college figures are within 2% of the 29% figure for college graduates (Pew Research Center, 2015).

The Pew Research Center also conducted the Religious Landscape Study, where it studies religious affiliations, beliefs, practices, and social and political views. They estimate that 83% of conservatives identify as Christian, while both Jews and Muslims only make up 1% of conservatives. For moderates, they estimate that 69% identify as Christian, 2% identify as Jewish, and 1% identify as Muslim. In comparison, Christians make up only 52% of liberals, while Jews make up 3%, and Muslims make up 1% (Pew Research Center, 2022).

A Gallup poll released in 2012 discovered a strong association between being LGBTQ+ and identifying as independent, as well as being LGBTQ+ and identifying as liberal. Only 13% of LGBTQ+ respondents identify as conservative, in comparison to 30% of heterosexual respondents identifying as such. Liberal identification jumps 12% to 44% on the LGBTQ+ respondents. Moderate identification only jumps 4% for LGBTQ+ respondents from the 39% of heterosexual respondents identifying as moderate (Gallup, 2012).

Section Three: Methods

The goal of my final project is to measure the variances in political views and examine the possible causes and factors. My primary research question is what causes variances in political views? I will be utilizing the General Social Survey from the University of Chicago to analyze the American public. To study this, I am using six independent variables, which are the following: AGE (respondent's age), SEX (respondent's sex), INCOME16 (total household income), EDUC (respondent's years of education), RELIG (respondent's religious identification), and SEXORNT (respondent's sexual orientation). My sole independent variable is polviews (political views of respondents), which I will be using to measure political leanings and trends.

My seven null hypotheses from my bivariate and multivariate analyses are the following: There is no relationship between age and political views, there is no relationship between sex and political views, there is no relationship between income and political views, there is no relationship between years of education and political views, there is no relationship between religious identity and political views, among LGBTQ+ respondents, there is no relationship between sex and political views, and among heterosexual respondents, there is no relationship between sex and political views.

My seven alternative hypotheses are the following: There is a relationship between age and political views, there is a relationship between sex and political views, there is a relationship between income and political views, and there is no relationship between years of education and political views, there is a relationship between religious identity and political views, among LGBT+ respondents, there is a relationship between sex and political views, and among heterosexual respondents, there is a relationship between sex and political views.

To sufficiently conduct my data analysis, much recoding of my variables had to be done. First, to simplify polviews into the three primary groups of liberal, moderate, and conservative, I recoded 1 through 3 into 1, 4 into 2, and 5 through 7 into 3. I then labeled 1 as "liberal", 2 as "moderate", and 3 as "conservative".

For the independent variable AGE, I planned to group respondents into 4 separate groups of similar response numbers. The recoding I did for AGE was 18 through 29 into 1, 30 through 49 into 2, 50 through 69 into 3, and 70 through 120 into 4. I then labeled each accordingly to their age bracket.

SEX did not require any recoding, as the variable only has males and females for primary response options. INCOME16 required me to divide the responses into 5 groups of wealth classes. I recoded 1 through 11 into 1. This was then labeled lower class, as it contained the responses of households of less than $20k annual income. I recoded 12 through 16 into 2 and labeled this group between $20k and $35k lower middle class. 17 through 20 was recoded into three and held the responses of respondents whose households brought in between $35k and $74K a year. This section was labeled middle class. 21 through 23 correspond to $75K to $129K, which is the upper-middle class. Finally, 24 through 26 were recoded to 5 and labeled upper class. EDUC was recoded into two groups, 0 through 12 into 1 and 13 through 20 into 2, dividing respondents with no college education and college education. RELIG required a lot of recoding. 1, 2, 10, and 11 were recoded into 1 and labeled "christian." 9 was recoded to 2 and was labeled "muslim.” 3 was labeled "jewish," and 4 through 8 and 12 and 13 were recoded to 4 and labeled "other". SEXORNT only had one recode, which was 1 and 2 into 1. 1 was labeled "lgbtq+" and 3 was labeled "heterosexual."

Section Four: Findings

To study the effect of age on political views, I used AGE as my independent variable and polviews as my dependent variable. My null hypothesis states there is no relationship between age and political views, while my research hypothesis is there is a relationship between age and political views.

The strongest relationship seen in Graph 1 (polviews * AGE crosstab) is the positive correlation between conservative identification and older age. In respondents 18-29 years old, only 20.6% of respondents identify as conservative. This figure jumps from 4.9% to 25.5% among respondents 30-49 years old. Among 50–69-year-olds, 36.5% identify as conservatives, and nearly 40% (39.5%) of respondents older than 70 identify as conservative. Respondents over 50 years old are around 10% more likely to identify as conservative than respondents under 50, revealing a strong relationship. A strong negative relationship can be seen between increased age and liberal identification. The largest group by percentage on Graph 1 is liberals identifiers 18-29 years old of about 44%. As age goes up, the percentage identifying as liberal decreases to just 29.5% for respondents over 70. The weakest correlation can be seen in moderate identification, where there is also a negative relationship between increased age and moderate identification. 30–49-year-olds have the highest moderate identification (36.8%). Among the next oldest group, 50–69-year-olds, this number decreases to 33.3%, and 31.0% for respondents over 70. Using Graph 2, the chi-square test for the polviews * AGE crosstab, there is an asymptotic significance, or p-value, of .000. Since this p-value is under .05, we can recognize the strong statistical significance between age and political views and reject the null hypothesis stating there is no relationship between age and political views.

To analyze the differences between political beliefs between men and women, I made Graph 3, a crosstab where polviews is the dependent variable and SEX is the independent variable. Graph 3 paints an interesting story of the relationships between gender and political identification. The greatest variance between the two sexes is in conservative identification. 36.4% of men identify as conservative, in comparison to only 28.4% of women, meaning men are 8% more likely to identify as conservative than women. A less significant, yet opposite variation is seen in moderate identification, where 36.4% of women identify as moderate, relative to 32.3% of men. Women and men are the most similar in their liberal identification, in which women are only less than 4% more likely to identify as liberal than men. Using Graph 4, the chi-square test for the polviews * SEX crosstab, there is a p-value of .000. Such a low asymptotic significance suggests a statistically significant correlation between the two variables and enables us to reject the null hypothesis suggesting there is no relationship between sex and political views.

Graph 5 in the appendix examines the relationship between wealth and political views. The General Social Survey variable I used as my independent variable to measure wealth was INCOME16. INCOME16 measures a respondent’s total household income, giving better insight into what socioeconomic class respondents fit into. My dependent variable remained polviews. My null hypothesis stated there is no relationship between income and political views, while my research hypothesis stated there is a relationship between income and political views. When looking into the data on the polviews * INCOME16 crosstab (Graph 5), conservative identification remains within 5% of the 30% identification mark throughout the five socioeconomic groups. The middle class has the highest conservative identification (33.9%), and the lower middle class has the lowest conservative identification (28.1%). The strongest association can be seen in the negative correlation between moderate identification and increased wealth. Both the lower class and lower middle class have remarkably high moderate identification, with 40% and 40.4% each. Middle class moderate identification falls to 33.6%, followed by 31.6% for the upper middle class and 25.3% for the upper class. A reserve association is seen for liberal identification, where 41.8% of the upper class identify as liberal, while this figure drops between each socioeconomic group, to where only 30.8% of the lower class identify as liberal. With the assistance of Graph 6, the chi-square test for the analysis, we are given an asymptotic significance of .000. A p-value under .05 suggests a statistically significant relationship between the two variables. In addition, we can reject the null hypothesis claiming there is no relationship between income and political views.

Graph 7 is a crosstab, in which the independent variable is EDUC and the dependent variable remains polviews. The data shows two negative correlations and one positive relationship between increased education and political identification. The two negative correlations are with moderate and conservative identification. The stronger relationship is with moderate identification, where 42.6% of non-college educated Americans identify as moderate, yet only 31.9% of college educated Americans identify as moderate. In comparison, 35.5% of non-college educated Americans identify as conservative, while 30.6% of college educated Americans are conservative. The strongest relationship is between education and liberal identification, where nearly 40% of college educated Americans are liberal. Utilizing Graph 8, the chi-square test for the polviews * EDUC crosstab, we see an asymptotic significance of .000. This figure allows us to reject the null hypothesis stating there is no relationship between education and political views, due to the strong statistical significance.

The polviews * RELIG crosstab (Graph 9) has some of the most significant relationships in the entire paper, as seen in the above chart. Beginning with Jewish respondents, they are more likely to be liberal than moderate or conservative combined, with a liberal identification of 56%. The next largest group conservative Jews, with about 22.7%. This mark is only 1.4% greater than the percent of moderate identifying Jews. Muslims have the greatest figure in Graph 9, in which 69.6% of Muslim respondents identify as moderate. Moderate beliefs dominate the Muslim-American community, as only 8.7% of Muslims identify as conservative. Christian responses were much more evenly divided. The largest group was conservative Christians, which was 40.6% of respondents. This was closely followed by moderate identification, which was 36.3% for Christians. Then came liberal identification, with only 23.1%, or less than ¼ of Christians identifying as such. With the assistance of the asymptotic significance from Graph 10 in the appendix, the p-value for the data is .000, revealing a strong statistical significance. We can reject the null hypothesis which claims there is no relationship between religious identity and political views.

My multivariate analysis can be seen in Graph 11 and is a crosstab with polviews as the dependent variable, SEX as the independent variable, and SEXORNT as the control variable. I chose SEXORNT, as I thought it would be interesting to see the differences in political identification between the general population and the LGBTQ+ community. It turns out, the data is quite interesting. It is important to note as a baseline that overall, men lean conservative, while women tend to be more liberal or moderate. This trend stands for heterosexual respondents, where heterosexual men are 37.7% conservative, 32.7% moderate, and 29.6% liberal. This strong conservative correlation is in contrast with heterosexual women, where 29.4% are liberal, 37.3% are conservative, and 33.3% are liberal. Liberal identification dominates both LGBTQ+ men and women. 69.9% of LGBTQ+ men identify as liberal, as well as 58.6% of LGBTQ+ women. Using Graph 12, the chi-square test for the multivariate analysis, we can find the asymptotic significance of .271. This p-value means there is no statistical significance, and we cannot reject the null hypotheses stating among LGBTQ+ respondents, there is no relationship between sex and political views, and among heterosexual respondents, there is no relationship between sex and political views.

Section Five: Conclusions

Using data from the General Social Survey, it is evident that there are significant associations between political views and age. The strongest relationship is between conservative identification and increased age, where Americans over 70 are nearly 20% more likely to identify as conservative compared to Americans 18-29 years old. The opposite is seen for liberal identification, where Americans 18-29 years old are nearly 15% more likely to identify as liberal than Americans over 70. The most interesting correlation is the negative relationship between increased age and political moderation. Though not as drastic as the changes in conservative and liberal identification, it interests me that as the older Americans get, the less likely they are to have picked a side politically.

The relationships between sex and political views are not necessarily strong yet are statistically significant. Men are 8% more likely to identify as conservative than women, while women are 3.7% more likely to lean liberal than men. Interestingly, women are most likely to be moderate, while men are most likely to be conservative.

There are strong trends among both liberal identification and income, as well as moderate identification and income. Moderate identification drops nearly 15% between the lower class and the upper class, while liberal identification jumps 11% when climbing up the socioeconomic ladder. As stated before, conservative identification remains around the same throughout the socioeconomic groups.

Education undoubtedly plays an impact on political identification. Non-college educated Americans are much less politically polarized than college educated Americans. Interestingly, college educated Americans are around 16% more likely to identify as liberal than non-college educated Americans. Throughout the two groups, conservatism remains the least popular ideology, while moderate views are the most popular.

It is evident from the data that religious identification plays a significant role in political identification. The most telling facts from the data show the significance of these relationships: Christians are 17.5% more likely to be conservative than liberal. Jews are 6% more likely to be liberal, than identifying as moderate or conservative combined. Muslims are more than 60% more likely to identify as liberal than conservative. Finally, Jews are the most politically polarized group.

In the multivariate analysis of polviews * SEX * SEXORNT, we see our first data set without statistical significance. The meaning of the p-value of .271 for LGBTQ+ men and women means that regardless of whether the respondent is male or female, an LGBTQ+ respondent will likely lean liberal. This is evident, as between both LGBTQ+ men and LGBTQ+ women, respondents are more likely to be liberal than moderate or conservative combined.

There are a number of arguments that could be made about the causes of these relationships, yet it is evident that there are many factors that affect Americans’ political views. Arguments have be made that psychological factors or societal influences cause relationships between the independent and political views, yet the true answers to these correlations are widely unknown.

References:

Igielnik, R. (2020, September 2). Men and women in the U.S. continue to differ in voter turnout rate, party identification. Pew Research Center. Retrieved May 7, 2022, from https://www.pewresearch.org/fact-tank/2020/08/18/men-and-women-in-the-u-s-continue-to-differ-in-voter-turnout-rate-party-identification/

Peterson, J. C., Palo Alto College Search for more articles by this author, Smith, K. B., University of Nebraska–Lincoln Search for more articles by this author, & Hibbing, J. R. (2020, April 1). Do people really become more conservative as they age?: The Journal of Politics: Vol 82, no 2. The Journal of Politics. Retrieved May 2, 2022, from https://www.journals.uchicago.edu/doi/10.1086/706889#_i7

Pew Research Center. (2020, August 28). A deep dive into party affiliation. Pew Research Center - U.S. Politics & Policy. Retrieved May 7, 2022, from https://www.pewresearch.org/politics/2015/04/07/a-deep-dive-into-party-affiliation/#party-id-by-race-education

Pew Research Center. (2020, May 30). A wider ideological gap between more and less educated adults. Pew Research Center - U.S. Politics & Policy. Retrieved May 7, 2022, from https://www.pewresearch.org/politics/2016/04/26/a-wider-ideological-gap-between-more-and-less-educated-adults

Pew Research Center. (2022, March 31). Religious landscape study. Pew Research Center's Religion & Public Life Project. Retrieved May 7, 2022, from https://www.pewresearch.org/religion/religious-landscape-study/

Appendix:

Graph 1: polviews * AGE crosstab