Public opinion has played an important role in the efficiency of the United States government ever since its inception; as a country founded on the principles of liberty and democracy, the attitudes of citizens on political issues, leaders, and institutions have a lasting impact on the nation’s political processes. Therefore, due to the increasing importance of public opinion in American politics and the approaching presidential election, it is of the utmost importance to measure public opinion through the use of surveys and polls.
In order to achieve a greater understanding of the key components of public opinion, one must understand America’s dominant political ideologies and the socialization factors that cause one to form their political opinions and values. The term “political ideology” refers to the ideas and beliefs through which individuals interpret politics. In the United States, two of the most common ideologies are that of liberalism and conservatism; liberals often advocate for the civil rights of minorities, political reform, and express deep concern for climate change and ecosystem endangerment, while conservatives are known to support the “status quo” when it comes to the nation’s current policies, often opposing efforts for government expansion. These ideologies are often formed through the political socialization of the individual. Individuals are initially socialized by their families, often identifying with similar political ideologies to their caregivers. They are also influenced through their social groups, race, party affiliation (Democrat or Republican, liberal or conservative, etc.), highest level of education achieved, and their political environment. With a great variety of political ideologies in the United States, it is likely unsurprising that it is difficult to ensure that most citizens feel pleased with the outcome of elections and policy changes. Therefore, in order to gain a better understanding of the concerns and needs of the public, public-opinion polls have become exceedingly common.
In order to discover the true opinions of the citizens of Cuyahoga County, the Baldwin Wallace political science department launched several identical surveys among the convenience samples of each student in the American Politics course. This paper compares the results from nationally representative public opinion polls to the results from a public opinion poll I administered to family and friends from October 13, 2016 to October 27. Although the polls used identical question wording, several differences emerged. In what follows, I describe three main differences between the two sets of data. I argue we see these differences because of the size of the population, differences in time period in which the surveys were distributed, and differences in age of the general population surveyed. I conclude with a discussion about the credibility of polls that draw information from convenience samples, as opposed to information drawn from simple random samples.
The first comparison is the number of individuals reported to be voting for Hillary Clinton in the Bloomberg Politics poll, versus the amount who reported supporting Clinton in the poll I distributed among my peers via the social media platforms Facebook, Tumblr, and Twitter. According to the results received, 43% of Bloomberg survey recipients expressed support for a Hillary Clinton presidency, while over 56% of the individuals who responded to the survey I distributed expressed support of the candidate.
While this does not seem to be a large difference, there is a majority support for Clinton in the social media poll (noted in Figure 1 as “Qualtrics Survey”), whereas Clinton trails Trump in the Bloomberg poll results. Those responding to the Bloomberg poll were mostly part of an older generation; most of Trump’s supporters are older and more conservative, which may explain why only 43% of the Bloomberg respondents expressed support for Clinton. However, the Qualtrics survey expresses a much greater support for Clinton over her opponent, showing that 56% of respondents would favor a Clinton presidency. This is likely due to most of the respondents being a part of younger generations (most who participated were part of the “Milennial” generation, being under the age of 25), and therefore supporting reform over conservative values. If I had distributed the Qualtrics survey to a larger audience, perhaps the results would have been closer to the Bloomberg results. Only 36 individuals responded to my request to complete the poll, as opposed to the more than 1,000 who participated in Bloomberg’s survey.
This data is further supported when respondents are questioned about their opinion on businessman Donald Trump.
According to the results in Figure 2, 48% of those surveyed in the Bloomberg poll expressed support for Trump, while in the social media poll, only 22% of the respondents expressed support for the businessman (many only supporting him to prevent a Clinton presidency), which I found quite surprising at first. One possible factor is likely the time period in which the Bloomberg poll was distributed versus when the Qualtrics survey was released on the Internet. Bloomberg’s poll was released between the dates of September 9 through to September 12, as opposed to the Qualtrics survey, which was released less than a week after the release of a video in which Trump is heard describing various sexual privileges he is entitled to due to being in a position of power. Many argued that his statements described sexual assault, and thus the candidate lost several of his supporters to his opponent and Libertarian Gary Johnson. Therefore, if the individuals had taken the Qualtrics survey prior to the release of the video, there may have been a closer similarity in results.
According to Figure 3, only 9% of the Bloomberg respondents were under the age of 25 (those over the age of 65 were the largest age group, counting for 24% of the total number); however, 84% of those responding to the Qualtrics survey were younger than 25. The overwhelming support for a Clinton presidency expressed in the Qualtrics poll is likely due to the fact that most of the respondents were young, liberal students who disagree with most of Trump’s policies and stances on immigration reform and reproductive health. Generally, younger individuals tend to favor more liberal policies than older, more conservative generations. Because the Qualtrics poll was distributed to individuals who were “friends” with me on Facebook, Tumblr, and Twitter, many of the individuals had similar opinions. Most of those who participated in the experiment were from Dayton, Ohio or Colorado Springs, Colorado. Most individuals who I am acquainted with are also in a similar age group to me (myself being 17 years old); identify as Christian or Agnostic; are white, African- or Asian-American; and are part of the middle class. Most supporters of more liberal candidates are younger than 30, are a part of a minority, and are in the middle or working class. Therefore, it is likely that due to several of the individuals being under the age of 25 (17 individuals advocating for universal health care and improved unemployment care), their support for Hillary Rodham Clinton over Donald Trump is likely due to the changes they have experienced in their lifetime, leading them to be more accepting of reform.
To further explain the difference of results between the two polls, it is important to note that public-opinion polls are crucial in educating officials on the concerns of their constituents; all recent presidents and politicians have utilized public opinion polls and worked closely with media outlets. Because it is impossible to survey each and every individual in the population, a political survey must provide an accurate representation for the general public through avoiding selection bias and carefully selecting a sample method and size.
First, one creating a public opinion poll must choose between two types of representative samples: a simple random sample or convenience sample. A common example of a simple random sample is the random digit dialing of citizens nationwide to respond to political surveys. This is considered a simple random sample because every person in the population has a probability of being selected. A convenience sample, however, is a type on non-probability sampling, meaning that not everyone in the population has a possibility of being selected. Instead, individuals are chosen from places where they can be conveniently found, such as a public library or shopping center. When sampling, it is important to avoid selection bias, which occurs when the sample used does not represent the entire population accurately, or if too few individuals of certain demographics are surveyed.
Once the poll creator has found a group to collect data from, it is important to ensure that the sample size is large enough to provide results that accurately represent the population. Finally, one must be cautious about a survey’s question wording and design, in order to prevent errors in data. An example of such is the social desirability effect, which occurs when respondents are asked a direct question about a sensitive topic in which certain answers may not be deemed socially appropriate. Therefore, in an attempt to conform to the response that they believe is the most accepted, they respond with an answer that does not accurately represent their true opinion. This can also happen with push polling, in which respondents are asked questions that are meant to influence their opinion. While I believe that the survey I distributed ultimately avoided these fatal mistakes, it is still not considered to be a scientific source, due to the lack of respondents. Because my poll used a convenience sample (individuals chosen from the amount of “friends” on my social media accounts) rather than a simple random sample in which every individual in Cuyahoga County (or the entire United States) would have a chance of being selected, I only received the opinions of those who are similar to each other through geographic location, ideology, and age.
Therefore, due to the extreme similarity in answers in age, location, race, and other demographics, it is likely that the results from the Qualtrics poll are not as credible as the Bloomberg poll due to its use of convenience sampling with a small population size. In the future, if I were to distribute another public opinion poll, I would distribute it in several public locations, such as public libraries, parks, college campuses, and some social media sites as well. Because of all of these factors in ensuring that a public-opinion poll is accurate, it is important to understand that not every poll will be completely accurate. There will always be improvements to be made, and the world of social science will continue to improve.
Bloomberg Politics. (2016). Bloomberg Politics Ohio Poll, September 2016 [United States]. Retrieved from: https://assets.bwbx.io/documents/users/iqjWHBFdfxIU/r2.771xfmKOI/v0
Ezell, Hope. (2016). Qualtrics Survey, October 2016 [United States].