Upload
andrew-taylor
View
7
Download
0
Embed Size (px)
Citation preview
Research Methods Andrew Taylor
Final Research Paper
What is the effect of educational achievement on the level of political conviction in the American voting public?
Political conviction is an elusive topic of study the world over. What is it that compels one citizen
over another to engage in the politics of their people and effect as much change or establish as much
stability as is available to an individual? The connection between education and politics is one of the most
heavily researched relationships in the history of academic study. When you are endeavoring to ask the
fundamental question: “Why do people behave as they do?” you will invariably be led to the
corresponding inquiry: “What have they been taught to do?” Great minds throughout history have
considered the origins of civic virtue, social responsibility, political participation and the fundamental
rights of the individual within their culture. Many have identified factors that they believe to be intrinsic
of positional goods such as tolerance, justice, equality and participation. Others believe the subjective
nature of our existence precludes any such goods as situational in nature. However, it is commonly
believed as well as academically well established that the knowledge one generation passes on has a
direct effect on the future of the next.
My research question, “What are the effects of education on political conviction?” has been a
difficult ponderance. What is political conviction? How do you define it? How do you measure it? If you
ask one hundred people on the street if they have strength in their convictions you are likely to get ninety-
nine “of course I do” responses. Who wishes to consider the possibility that they are politically ineffectual
due to their own shortcomings? However, in a nation that boasts the freest citizens in the history of
mankind we can barely convince fifty percent of our people to show up once every two years and exercise
the most basic action of those convictions and vote.
After much grappling with the concept of conviction it has been defined for the purposes of this
study as the circumstances in which an individual will choose to act or refuse to act based on their
personal ideology. What causes someone to take time out of their day to march in a rally? What prompts
one to decrease their personal wealth in any capacity to aid in a common political cause? What compels
you to write a letter to the editor in strong support or dissent of an issue? Conviction. While this study
may not be able to answer the “What and the Why” people participate in the way that they do within the
political realm, I do believe it can shed some light on the “Who and the How” in regards to education.
Hypothesis –
In a study of individuals, those with a higher degree of educational achievement will have a
higher degree of political conviction which is expressed through various forms of political participation.
Literature review –
The review of previous works for this study begins with the foundational work of Converse
(1964) where the very concepts of cultural perceptions and the nature and formation of belief were
explored. Utilizing attitudinal survey techniques Converse and his team classified responses to civic and
issue based inquiries into categories ranging from the most informed to the least informed components of
the mass electorate. In his work, he concluded that as civic knowledge increases within an actor of the
body politic the individual develops what he termed to be cognitive “constraints.” These formalized
constraints allowed the actor to develop a more precise understanding of complex issues as well as the
ability to effectively link or “package” (as he terms it) inferential truths; basically this is the ability to
extrapolate complicated concepts from inferences into sophisticated and informed opinions regarding a
wide range of issues. He used the example of the word “conservative” which is meaningless without the
proper “package” of mental constraints to allow you to infer the likely parameters of an issue or concept
that has been labeled as conservative. Those that failed to acquire these cognitive constraints were left
without the proper contexts to adequately assess the nuances of political concepts and they consequently
fill in the gaps with simple conjecture. While impressed with his analysis I think that his conclusions
regarding the weak link between cues from elites to the mass public in civil discourse should be revisited
in the context of the information age.
A corroborating study, Bobo & Licari (1989) found a similar correlation in the effects of
education on political tolerance in the form of cognitive sophistication. They demonstrated that as verbal
aptitude increased, the subjects were more likely to comprehend the social stratification and universal
application of civil rights. In this study the subjects were ranked following a verbal acuity test and then
asked about extending basic civil rights to deviant or non-conformist groups. A strong correlation was
found between verbal cognition and tolerance in the face of unpopular views. It should be noted that this
did not extend past legal definitions of acceptable behavior. To the contrary, a sharp decline for support
for groups that advocated for violence was also demonstrated among the higher scoring respondents. The
major flaw in this study in my view is the limited parameters defining “educated.” A good vocabulary is a
strong indicator of education but it is far from conclusive.
This leads into the next piece for review, Emler & Frazer (1999), which delves into the
curriculum verses socialization debate. This study was an exhaustive review of previous works on the
subject in an attempt to ascertain precisely what the active ingredient in education is that compels its
participants to be more civilly engaged. Their findings were that curriculum is largely irrelevant. They
found that students of math or physical science were statistically just as likely to be engaged with social
action as students of social sciences. They hypothesized that due to the finite nature of opportunities for
political participation those with a higher aptitude for discourse are in a better position to exploit those
opportunities. They concluded that it is likely the exposure to formal institutions with political
substructures and the development of social networks that leads to future political participation. This
work failed to track how individuals were likely to express that political engagement.
A conflicting review of previous works, Galston (2001) finds that curriculum is far from
irrelevant in the formation of effective civic actors. In this piece the author considers the fact that
aggregate levels of education in the United States have substantially increased over the past fifty years yet
political knowledge and participation rates remain unchanged. Galston postulated that it is not simply the
knowledge of these institutions that compels participation but the way that we learn about them. His
conclusion based on other statistical analysis is that basic classroom instruction on civil institution is
insufficient to compel the individual to participate. Open debate, discussion and even civil conflict
coupled with service learning initiatives produce a greater degree of participation. While he makes a good
argument, it is a bit vague on specifics and relies heavily on the works of others. Largely it is a call to
action on greater study of the link between education, participation and service learning.
Finally, Berinsky, Lenz (2010) was particularly interesting considering my personal background.
They hypothesized that there is no true specific correlation to education and political participation and
that the correlation previously observed by other researchers is simply one example of a larger
phenomenon of socialization. Utilizing CPS data from the 1960’s and 1970’s they were able to provide a
cross-sectional analysis indicating that there is no statistical difference in the levels of political
participation between college graduates and Vietnam War veterans. They concluded that education can
likely lead to greater political participation but it is the broader experience that motivates and that
experience can be achieved through many forms of personal growth and self-discovery. The weakness of
this study is that no attitudinal analysis was possible. The specific motivation behind veteran participation
was not explored.
So, while the direct mechanism of effect regarding a causal relationship between education and
political participation is still far from an academic absolute, the academic community is comfortable in
utilizing it as an independent variable in exploring modes of political expression. The continuing central
debate concerning the effects of education and political participation revolves around the concept of
curriculum verses institution. Is it the level of enlightenment that one receives in continued education that
affects one’s political attitudes or is it the broader socialization and familiarization with culture that leads
individuals to engage, or is the phenomenon being analyzed from the finish-line? Perhaps it is individuals
that are predisposed from their formative years and earlier socializations towards civic participation that
seek out advanced education as a means to satisfy those ends.
Data and Methods –
For the purposes of this study we will utilize the 2012 ANES Times survey data set. The history
of the ANES election survey is a fairly interesting product of academic knowledge. According to Prof.
Nancy Burns of Michigan State the survey was the intellectual response to the results of an inquiry into
Ohio public perceptions of foreign policy in 1940. From there, under the direction of Angus Campbell
and Robert Kahn in 1948 and later Warren Miller in 1952 the ANES survey grew in scope and complicity
to its modern incarnation that models political perceptions and attitudes on a national scale.
The 2012 ANES Time Series Study is a cross sectional survey of all U.S. eligible voters utilizing
face-to-face cluster interviews as well as a web-based format producing a total sample size of 5,914
participants (Interview: n = 2,054, Web: n = 3,860). The study was conducted in six waves: two in the
face-to-face format (Pre and Post-election), four in the web based format (2 Pre and 2 Post). The data set
has specific weights concerning the face-to-face format as well as the web format.
Moving forward with the ANES Time 2012 data set, three major forms of advanced political
participation beyond voting were identified in which the average citizen expresses his or her political
conviction. These were expressed in the form of financial support, vocal or written support and physical
support in the form of time and effort. For the purposes of this study they have been coded as: Money,
Voice and Time.
Utilizing these parameters, three sets of four variables within the Post-election survey that
directly measure the respondent’s participation in each of the three forms of political action were
established. They are as follows:
“Money” Variables –
1. Did R contribute money to specific candidate campaign? (MOBILPO_CTBCAND)
2. Did R contribute money to political party? (MOBILPO_CTBPTY)
3. Did R contribute to any other group for/against a candidate? (MOBILPO_CTBOTH)
4. Has R in past 4 years: given money to a soc/pol organization? (DHSINVOLV_EVERGIVSOC)
“Voice” Variables –
1. Has R in past 4 yrs: called radio/TV about a political issue? (DHSINVOLV_EVERTVCALL)
2. Has R in past 4 yrs : written a letter to newspaper/mag about political issue?
(DHSINVOLV_EVERLETT)
3. Has R in past 4 yrs : sent a message on Facebook/Twitter about political issue?
(DHSINVOLV_EVERTWIT)
4. Has R worn a campaign button or post sign or bumper sticker? (MOBILPO_RBUTTN )
“Time” Variables –
1. Has R in past 4 years: attended city/school board? (DHSINVOLV_EVERMTG)
2. Has R in past 4 years: joined a protest march? (DHSINVOLV_RALLY)
3. Does R go to any political meetings, rallies, speeches? (MOBILPO_MARCH)
4. Does R do any (other) work for party or candidate? (MOBILPO_OTHERWORK)
Each dependent variable was recoded utilizing SPSS to remove non-respondents and reduce the
variable to nominal responses of Yes or No (Yes = 1 / No = 0). From there the twelve variables were
combined in to their respective groups producing three individual, interval level variables named
“TotalMoney, TotalVoice and TotalTime” and coded as 0 – 4 (0 = No participation and 4 = Participation
in all). The independent variable: What is the highest level of school you have completed or the highest
degree you have received? (DEM_EDUCR) was recoded to exclude all non-relevant codes outside of
responses 1 through 16. These are interval level responses covering the spectrum of American educational
achievement. (1 = Less than first grade and 16 = Doctorate degree.)
Variable baselines and adjusted sample size are as follows:
Statistics
Education Level Money Time Voice
NValid 5846 5495 5503 5498
Missing 68 419 411 416
Mean 10.6259 .5119 .3653 .4456
Mode 9.00 .00 .00 .00
The primary challenge faced was a large degree of negative skew in each of the dependent
variables due to a high rate of non-participation in any form of increased political activity as shown in the
following histograms:
The secondary challenge to the data was the bi-modal response to educational achievement shown below:
In fact, 45.9% of respondents fell within the 9 and 10 intervals of High school Graduate and some
college. Another 27.9% fall within the 13 and 14 intervals of Bachelor’s or Master’s degree. The primary
concern is that the concentration of respondents within these intervals increases the size of the 95%
confidence interval within each of the other independent variable’s intervals. In some cases to nearly 50%
of the data range. The final difficulty with this data set is that it fails to provide for frequency of
participation. There is no way to know how often or to what extent those that choose to act, do so.
Results –
Considering that both the dependent variables as well as the independent variable in this study are
interval level data, a means comparison was run to determine the average participation within each
educational interval with the following results:
As you can see following a means test for the average level of participation in multiple forms of
political expression, there is a statistical upward trend in all three composite dependent variables as the
educational achievement increases. However, given the fact that so few individuals within all educational
intervals choose to participate at all I was compelled to calculate the correlation coefficient of each
variable with the following results:
Education Level Money
Education Level
Pearson Correlation 1 .229 **
Sig. (2-tailed) .000
N 5846 5436
Money
Pearson Correlation .229** 1
Sig. (2-tailed) .000
N 5436 5495
Education Level Time
Education Level
Pearson Correlation 1 .172 **
Sig. (2-tailed) .000
N 5846 5444
Time
Pearson Correlation .172** 1
Sig. (2-tailed) .000
N 5444 5503
Education Level Voice
Education Level
Pearson Correlation 1 .123 **
Sig. (2-tailed) .000
N 5846 5440
Voice
Pearson Correlation .123** 1
Sig. (2-tailed) .000
N 5440 5498
**. Correlation is significant at the 0.01 level (2-tailed).
Clearly, preliminary results do show a positive trend for each composite variable, however, they are of
limited significance with the donation of money having the highest degree of correlation at r = .229
Discussion –
Following the statistical analysis of this study one may say that the original hypothesis of “In a
study of individuals, those with a higher degree of educational achievement will have a higher degree of
political conviction which is expressed through various forms of political participation” is in fact true.
Those with a higher degree of educational achievement do participate more than those with a lower
degree of achievement. However, one must also conclude that education is not predictive of a higher
degree of political participation or conviction.
If you refer back to the means comparison graph, one particularly interesting generalization in the
participation data pattern was the fact that all three dependent variables maintained a fairly tight and
consistent trend of uniform increase, all roughly within a .2 range of one another. This trend continues
right up until respondents reach the Bachelor’s level education interval where the “TotalTime and
TotalVoice” variables level out at around .6 and the “TotalMoney” variable more than doubles to roughly
1.2 in the last four educational intervals. This exception is likely due to an increase in wealth and
disposable income in relation to higher education but more study would be needed on the phenomenon.
Conclusion –
While this study does not establish a strong correlation between education and political
conviction expressed through participation, it is clear that a relationship does exist. It is my belief that the
composite variables of: TotalTime, TotalMoney TotalVoice, are valid measures of political conviction
and that education is one thread out of many that ties them together in a larger predictive parameter.
Future studies should be focused on exploring those other threads which may include: Gender, Age, Class
or Partisanship as the following preliminary means comparisons suggest, particularly in the expression of
conviction through financial support as it is uniformly the most utilized form of political participation.
Finally, as we all know, much of message politics revolves around the phrase “Get involved!” Future
research could look into the percentage of participants who did so simply because they were asked to.
This could be very useful in measuring the effectiveness of future support drives and campaigns.