WEEKS 1-2:
HISTORY OF DISCIPLINE, ETHICS, GOOD THEORY
The
history of the discipline of political science starts with the Classical
Era from 700 BC to 1850 AD, when we were a more philosophical field, and
asked such questions as: how the government and society “ought” to be organized;
and asked what “justice” truly is; and who should rule society, the “wise” or
the multitude; and what the obligations of citizens and of the government are.
Some class discussion may occur on these topics. The role of government as seen
by liberals and conservatives was discussed. These philosophical questions are
still important today, as you can see with the question of what is “social
justice,” and is the American criminal justice system fair to minorities, and
the debate over police misconduct. Who should rule- Democrats seek to make it
as easy as possible for as many people as possible to vote, while Republicans
are accused of “voter suppression”- trying to restrict the franchise to the
higher socioeconomic status. So while this course will not be focused on the
Classical Era, it is still an important part of our discipline; make sure that
you take some of Dr. Chamberlain’s classes to learn more.
Next is the Institutional Era, which was from 1850 to 1900 AD. That era of
political science focused on institutions and processes, such as the Congress
and the stages whereby a bill becomes a law. It focused heavily on the U.S. constitution,
and the structure and powers of the three branches of government. It took a
legalistic case study approach to the legal process, as constitutional and
civil liberty courses studied the historical evolution of how the bill of
rights has evolved in practice over the years. In the international relations
field, it viewed the nation as a “unitary actor,” acting as a united entity
with a rational goal, such as self-interest, maintaining spheres of influence,
and engaging in power politics.
Next is the Transitional Era from 1900 to 1945. Political scientists saw
some problems with the institutional approach of studying politics, such as
Irony of Form. An example is that non-democratic nations would have
constitutions that looked very democratic on paper (elected Parliaments,
Presidents, like Russia), but the governments acted in an authoritarian or
totalitarian manner, suppressing opposition parties and the freedom of the
press. Also, political scientists realized that one could not just study an
institution or a nation as a whole, since there was considerable Pluralism within
those institutions. For example, there is a big difference between how
Democrats and Republicans act and govern in Congress today. Pluralism exists
within a nation, such as the Blue (Democratic controlled) states and the Red
states (Republican) in the U.S. today. So there is no unitary state. As such,
political scientists turned to a new approach to studying politics.
The Behavioral era is from 1945 to present,
and it is extremely important to know, since it is the approach that this
course uses. Its characteristics are:
1) We use the scientific
method, which relies on theories that seek to explain and predict
political phenomenon. We look for patterns in the world.
An example is that we want to explain how and
why people vote the way they do for presidential candidates. A sociological
theory says that people are affected by their group membership, and that
African Americans and lower income people in general are more likely to vote
for the more liberal party, the Democrats, than are the majority white race and
people having a higher income. A social-psychological theory of voting says
that a person’s attitudes are more important in affecting their presidential
vote. This University of Michigan model published in The American Voter in the
1950s theorized that party identification, issue attitudes, and evaluations of
the candidates affected who people voted for. Party identification was a psychological
identification with a political party. Issue attitudes might be measured by
liberal-conservative self-identification. Therefore, Republican party identifiers, conservatives, and those viewing the Republican candidate as having more character and leadership were theorized to be more likely to vote for the Republican presidential candidate, compared to those who were Democrats, liberals, and those viewing the Democratic candidate as having more character and leadership. A third theory might seek to explain
how southern state legislators vote on roll call bills. One might argue that
race and party identification are the key factors, as African American
Democrats cast the most liberal votes, white Democrats are more
middle-of-the-road, and white Republicans are the most conservative in voting.
Each of these three models are theories that seek to explain and predict voting
patterns.
2) Your theory directs
your research and what data (information) you gather.
For example, in the University of Michigan
social psychological theory, you would gather data by conducting a national public
opinion poll of voters, and asking every person their party identification, their
attitudes on public issues, and their evaluations of the major party candidates
on important traits such as honesty and leadership. To test the sociological
model of voting, your national study would ask voters their race and income. In
Shaffer’s study of Mississippi state legislators, he collected data on each of
Mississippi’s 174 state legislators during the last 20 years of the 20th
century, specifically their: political party, from their state legislative website
or from the Secretary of State’s election results which listed the party they
ran under; their race, identified from their biographies that listed their
affiliations and membership in minority caucuses; their roll call votes on important
issues, published each year by the major state newspaper, the Clarion-Ledger.
3) Value free. Researchers
must be completely free of any personal, partisan, or ideological bias. They
must be completely neutral. Researchers simply try to explain how the world
works. They don’t try to change the world, an orientation that would hurt their
objectivity.
An example is that an
opinionated conservative
on Fox News may claim that Reagan won the 1980 presidential election because of
the popularity of his conservative philosophy, and an opinionated liberal may
claim that Obama won in 2008 because of his liberal philosophy, but our
research may indicate that each won merely because voters were dissatisfied
with the incumbent party’s performance in the Presidency. In 1980 voters
perceived Carter as a failed leader, due to the high unemployment (recession),
the 13% annual inflation, the Iranian hostage situation, and the Soviet
invasion of Afghanistan. When Reagan’s supporters tried to implement a very
conservative “mandate” and tried to significantly cut domestic spending, the
Democratic-controlled House of Representatives refused to do so, and
Republicans lost seats in the midterm election. Democrats won in 2008 because
of the financial crisis which occurred under a Republican President. When
President Obama implemented liberal programs like the Affordable Care Act and a
massive stimulus package, he lost control of the House in his first midterm
election and then lost the Senate in his second midterm year. If Democratic and Republican party activists had been more "scientific" and value free, they may have kept their party's Presidents from veering too far to the ideological extreme, and maybe they would have kept control of Congress.
4) Our research is Interdisciplinary,
as we draw on such fields as sociology, psychology, and economics.
An example from sociology is the
sociological model of voting, where mere group membership can influence how
someone votes. Note how about 90% of African Americans tend to vote Democratic,
and how a majority of whites vote Republican. Historically, sociologists also
examined voting differences in income groups, occupation, and religious groups.
During the New Deal Democratic majority era before 1968, blue collar workers
voted more Democratic, and white collar more Republican. The most Democratic
voting group were people of the Jewish faith, the most Republican were northern
Protestants, and Catholics leaned Democratic. Today, both blue collar workers
and Catholics are up for grabs. An example from psychology is Shaffer’s
study of balance theory, which tested the theory that voters seek to maintain
consistency in their beliefs and attitudes, as consistency reduces psychological
stress. As such, if you like a candidate, you will tend to believe that that
candidate agrees with you on the issues that you care about. That can actually result in your misperceiving
reality. If you really like Donald Trump, you maintain consistency by believing
that he was a “Great President,” and so he must have won the last election, so
the election was actually stolen! Sadly, this makes you psychologically
consistent, though you are divorced from reality. Indeed, former President Trump may be an example of this theory, as it can also apply to political leaders. Regarding economics, Shaffer
worked with economics professor (George Chressanthis) to determine that U.S.
Senate elections were somewhat accountable to the public. Senate election margins were affected by presidential
coattails, campaign spending, divisive primaries, and the preceding election
margin. Economic conditions in the state and federal pork barrel dollars did
not affect the election margins, however. Perhaps a better measure of Senators' job performance is the public's perception of their performance, which may be more affected by the publicity that they gain for federal money and projects that they bring into the state, so Senators' press secretaries may be instrumental in their electoral hopes.
5) We are methodologically
sophisticated. Our methods are not merely case studies of individuals, but we
study many people randomly selected and over many years.
For example, we conduct national public opinion
polls that are representative of the nation's diversity. We do not conduct
shopping mall polls, or phone-in or internet polls that fail to reflect the
views of lower socioeconomic classes. So we can test whether the sociological
group, or the social-psychological model of presidential voting is upheld. We
have indeed found that party identification is the most important factor
affecting the vote, with about 95% of Democrats and Republicans voting for
their party’s presidential and congressional candidates. The issues of satisfaction and dissatisfaction
are also important, and when times are good such as in 1984 and 1996, the
incumbent President wins (Reagan, Clinton). When there is a recession, the
incumbent tends to lose (Carter in 1980, Bush 1 in 1992). Candidate traits are
also important, as Carter in 1980 was viewed as a failed leader. In another
published study, I used such national polls from 1960 thru 1976 to explain how
voter turnout declined during that period due to decreased political efficacy, decreased partisan
intensity, and decreased newspaper readership. In my southern state legislative
factions research, I started with one southern state, Mississippi, in only a
few years. We then became more methodologically sophisticated and expanded to a
twenty-year time frame in Mississippi. Then, we added other southern states,
like Georgia, Florida, Arkansas, and Texas. My book co-authored with Charles
Menifield, Politics in the New South: Representation of African Americans in
Southern State Legislatures, won the V.O. Key award for the best book in
southern politics. Shaffer's study of balance theory relied on the 1994-1996
American national panel study to examine cognition change over time; such panel
studies follow the same people over time. The Shaffer and Chressanthis study of
Senate accountability used a pooled time-series, cross-sectional approach. All
even-numbered years from 1976 thru 1986 were included, as were all 33 state
contests in each election year. Regression and probit were used, such
sophisticated methods that we will only briefly cover one of them late in the
course.
6) We rely on an Individual
and group level of analysis, rather than a whole institution or whole
nation focus.
For example, the presidential voting studies use
the individual voter as the unit of analysis. The southern state legislative
factions also looked at individuals (legislators in this case), and we combined
them into three groups based on their race and party (only 3 groups, since
there were no black Republicans). Balance theory and voter turnout studies also
looked at individuals. Even our modern legal studies focus more on the
individual justice rather than the evolution of case law over time. So, in my
Honors Government class notes (website: https://sds17.pspa.msstate.edu/classes/honor/supremecourt.html),
you can see how I have listed each of the 9 justices in each year since 1961,
and indicated how they voted. Names in italics indicate that a Republican
President appointed them. You can then see how they voted on each court case,
and read the case outcome at the end to determine whether that was a more liberal
or conservative vote. Our theory is that Republican appointed justices vote in
a more conservative direction than Democratic appointed justices, and Democratic
appointees vote in a more liberal direction. Hence, the controversy in 2016 of a
Republican-controlled Senate refusing to consider President Obama’s last Supreme
Court nominee, and Democratic threats in Biden's first years in office to expand the size of the court. Again,
we focus heavily on individuals, not institutions as a whole.
Criticisms of
Behavioralism- are people and events
predictable, can we be value free; class discussion may occur.
Can this scientific behavioral approach explain the 2022 midterm elections?
One theory of midterm congressional outcomes is that the President's job approval rating and the economy are related to how many seats the President's party loses in Congress, and that the President's party almost always loses seats in midterm elections. Because President Biden's approval rating was in the low 40% and the nation faced fairly high inflation and a relatively weak economy, most political analysts expected a big Republican gain of seats in Congress. However, I pointed out that there were two exceptions to this pattern of presidential party midterm seat losses- 1998 (when Clinton's impeachment was viewed as very partisan) and 2002 (after the 9-11 terrorist attack). We also see a President's popularity increase when he runs for re-election, as the public must dampen their unrealistic high expectations as they compare the incumbent with the other party's presidential nominee. Former President Trump was so active in the midterm campaigns that voters did not focus solely on President Biden, but were also reminded of an even more unpopular alternative. Plus, President Biden and the Democrats reminded voters of the January 6 Insurrection, an event that some viewed as comparable to the divisive or dangerous events of 1998 and 2002. Finally, political science studies have found that Candidate Quality is important, and it is measured by candidates having previously held elective office or at least had run for office. Trump was successful in getting non-politician Republicans nominated, who were weaker candidates (as noted by Republican Senate Leader McConnell). Consequently, Republicans made only minor gains in the House elections, and even lost one Senate seat and two Governorships.
Comment on the field of
Public Administration.
Methodological Issues in
PA
Your Research Paper
We
now turn to a discussion of your research paper. On January 31, you should turn
in a few pages of your proposed paper, which includes an introduction, a
drawing of your model, a list of your 5 hypotheses, and a few paragraphs on
your methodology. To get started, think of something you would like to explain
(your dependent variable), and then pick three variables that help you to
explain it (your independent variables). Pick these four variables from the
variables listed in the Mississippi Poll. They must all be asked in the same years,
and try to choose two or three adjacent years, in order to increase the sample
size and reduce the sample error. A nice summary of the questions asked in the
Mississippi Poll, and what years they are asked in is available on-line.
Most
papers will have one outside "early" variable, such as a demographic
characteristic, two intervening variables (in the middle), and one dependent
variable. Other papers have two outside variables, one intervening variable,
and one dependent variable. The most recent and helpful student research paper is
accessible here. Notice how our alumnus Ms. Moehrs (now Gardner) used sex,
race, and ideology to explain people’s attitudes towards the death penalty. You
can also see how her Introduction justified the importance of her topic, how she
drew up her model, how carefully written her 5 hypotheses were, and (skip the
lit review for now) how her methods section was written. More on all of that
later; at this point just come up with a model with 4 interrelated variables
that make sense in how they go together.
ETHICAL CONCERNS
The
Stanley Milgram study was a controversial study of obedience to authority. He
wanted to know whether what happened in Nazi Germany could ever happen in the
United States. Would Americans blindly follow orders and hurt their fellow
citizens? The researcher took “subjects” off of the street, and told them that
a person was hooked up to electrodes in the back room, and that the subject
would lead a “learning experiment.” The subject would read a list of objects
and names, and the person in the back room would try to list them back from
memory, and every time they made a mistake the subject would increase the dial
and give the person an electric shock. When subjects would resist hurting them,
the researcher would remind them that it was a very important experiment, so
please increase the dial and pain. Most people followed orders, even when the
person in the back room was moaning and yelling in pain. At the conclusion of
the experiment, the researcher showed the subject that the person in the back
room had never been physically harmed, as they weren’t really hooked up to any
electricity. The researcher just lied to the subject. Is this an ethical
experiment? No, because the subject of the experiment was lied to, and they
could have been psychologically hurt (or even had a heart attack!). As such,
today we have federal regulations over how we do our research. All universities
must follow the protocol of Informed Consent, a very important topic for you to
know. We must follow this protocol whenever we study human beings, including
just doing a public opinion poll.
The four components of Informed Consent are:
Another important topic is the difference
between Anonymity and Confidentiality.
In Anonymity, no one can identify a
person with their responses. So, for example, if you were in a room with the
subjects, and you asked them to vote for a candidate, you might give each of
them a Red (Republican) and Blue (Democratic) marble, and they would drop the
appropriate marble in a covered collection box and discard the other marble in
a covered waste basket. Nobody would know how they voted, so it would be
completely anonymous.
In Confidentiality, the researcher knows how
the respondent answered, but promises not to tell anyone. That is like when we
did a mail survey of county party executive committee members across
Mississippi, and included a number on their questionnaires so that we could
determine that they had returned their forms. As I got the forms back, I could
determine what each individual had said on each of the sensitive questions that
we asked them, but I didn’t look at their responses. So I held their responses
in Confidentiality. Most of our studies, including the Mississippi Poll, are
merely confidential, and not anonymous. As such, it is important to delink your
poll results from the identity of the person who gave those responses. For
example, in the early years of the Mississippi Poll, we had printed
questionnaires and a first page with the phone numbers on them. Once pollsters
completed the surveys, I immediate tore off the first page with the phone
number on it, and threw it away. Therefore, I had turned Confidentiality into
Anonymity, since I now just had 600 completed surveys but didn’t know who had
answered what. You should also then destroy that list of 600 phone numbers. Yet
another problem is if you are doing a survey of a small group of people, and
you have enough specific questions that their responses could reveal who they
are. In our NSF study of Mississippi grassroots party activists, we asked
respondents their county, their party, and whether they were the county chair
or just a member of the county committee. Therefore, the county chairs could be
linked up with their responses. Therefore, the University of Michigan in
archiving the dataset deleted not only the county identifiers but even the
state identifiers. They ensured anonymity.
ASPA Code of Ethics: 5 sources of ethics for public administrators
1) Serve Public Interest: oppose discrimination and harassment, promote
affirmative action; public right to know; involve citizens in decision-making
2) Respect Law and Constitution: change obsolete, counterproductive laws;
prevent mismanagement of public funds, need audits; protect privileged
information; whistleblower protect
3) Personal Integrity: give others credit for their work-avoid plagiarism; avoid
appearance of conflict-of-interest, such as nepotism, gift acceptance, misusing
public resources, improper outside employment; act nonpartisan in actions;
admit own errors
4) Ethical Organizations: promote creativity, open communication among workers;
permit dissent, no reprisal, due process used; merit use
5) Professional Excellence: keep current on new issues, problems, upgrade
professional competence; be active in professional associations; help public service
students, provide them internships
THEORY BUILDING
Four characteristics of
a good theory (this is an important topic):
1. Explanation- why does something happen.
An example is from the Presidential voting
models. People vote more Democratic because they psychologically identify with
the Democratic party, because they are liberal, and because they prefer the
Democratic presidential candidate's characteristics. A different presidential
vote model is based on satisfaction or dissatisfaction. People hold the
President's party responsible for economic conditions in the country, so they
tend to vote for the President or his party's successor when things are going
well, and they tend to vote against him or his party's successor when things
are going badly. An example from the Southern state legislative factions theory is
that white conservatives are gravitating toward the more conservative party
nationally, the Republicans, therefore white Republican legislators tend to
vote conservatively. Liberal African-Americans tend to join the more liberal
party nationally, the Democrats, so African-American Democratic legislators
tend to vote liberally. Moderate whites tended to join the more ideologically
inclusive party in the 1990s South, so they tended to be Democrats; hence,
white Democratic legislators tended to vote moderately.
2) Prediction- if we know people's positions on the
independent variables, we can predict their positions on the dependent variable.
An example from the University of Michigan presidential vote model is
that if a voter is a Democrat, a liberal, and prefers the Democratic
candidate's attributes, we predict that they would vote for the Democratic
presidential candidate. If a voter is a Republican, a conservative, and prefers
the Republican candidate's attributes, we predict they would vote for the
Republican presidential candidate. An example from the southern state
legislative project is that we predict that African-American Democratic
legislators will tend to vote more liberally, against anti-crime measures, for
public education projects, and for affirmative action programs. We predict that
white Republican legislators will tend to vote in the opposite manner, in a
conservative direction. We also predict that white Democrats will tend to vote
somewhere in between these two groups, being supportive of pro-education and
anti-crime measures.
3) Generalizability- the theory should apply
to different situations and circumstances and different times and geographic
areas.
An example from the University of Michigan Presidential vote model is
whether this theory can apply to other offices, such as U.S. Congress, governor,
state legislature. It can. Can it apply to any time span (yes, but the 19th century
would have different parties, such as the Whigs and Democrats, Federalists and
Democratic-Republicans)? Can it apply to different geographic areas, such as
other nations (yes, because Ohio State professor Bradley
Richardson successfully tested the party identification model
in Japan, Netherlands, Germany, France, Britain, and Italy)? Shaffer extended
the University of Michigan party-issues-candidate model to the U.S. House, Senate,
governors, and even to state legislatures. Scholars have extended the party
identification model across decades. Richardson extended the model to other
countries. So this is a very generalizable and powerful theory. Another example
is the Southern state legislative factions project of Shaffer and Menifield. It
was generalized to other southern states, even to northern states and to the
Congress. It was generalized over time, such as 1980 to the present. Can it be
generalized to other nations having a newly empowered group, such as South
Africa is the final question?
4) Parsimony is simplicity with few independent variables.
The simplest theory is best, if everything else is equal.
An example for the Presidential Vote model is
that the Social-psychological model is parsimonious, as it has only three
predictors--party identification, issues, candidates. The economic
dissatisfaction presidential vote model has even fewer predictors—one (with two
categories of the variable- satisfied or dissatisfied). So both of these
presidential vote models are parsimonious theories. The Southern state
legislative factions project has only two predictors--party and race of
legislator. The dependent variable is less parsimonious, as it is not merely
ideology, but different types of issues such as education, crime, race issues.
Example of Predictive
Ability of a Theory.
The party identification
model. The five
presidential elections from 1992 thru 2008 were very competitive with Democrats
winning three and Republicans winning two. So if we had no other information
about a state like Mississippi, we would predict that a Mississippi survey
respondent would have a 50-50 chance of voting Democratic or Republican. Our
predictive success improves once we ask a respondent what their party
identification (a 7-point scale, but we cite only 5 categories by omitting independents leaning to a party) is. How
they vote follows (using the Mississippi Poll data):
We can then apply this
theory to the last presidential election of 2012 that was asked in the
Mississippi Poll. The results were very similar to previous years:
Hypothesis Testing
Independent variable is
the predictor; it comes first temporally and causally, it causes the dependent
variable.
Dependent variable is
the effect; it is being caused by the independent variable.
Ideology
--------------------------> Presidential Vote
(Independent var.).......................(Dependent Variable)
Hypothesis is a
statement of a relationship between concepts.
Example: self-identified
conservatives are more likely to vote Republican, compared to self-identified
liberals.
Hypothesis test- example
with crosstabulations, put independent variable at top, dependent variable at
the side. Calculate column percents.
VOTE FOR: |
LIBERAL |
MODERATE |
CONSERVATIVE |
BARACK OBAMA |
65% |
54% |
32% |
JOHN MCCAIN |
35% |
46% |
68% |
100% |
100% |
100% |
...................................Theory
.....................................|
....................................\|/
................................-Hypothesis-
..............Concept
<------------------------> Concept
.......................(Relationship between concepts)
The hypothesis above is
at the theoretical level- general, abstract.
............Indicator
<------------------------> Indicator
..............(Relationship between indicators; hypothesis testing)
Operationalizing your
concept is to select specific indicators of your abstract concepts. Hypothesis
testing occurs at the indicator level, and it measures the relationship between
the indicators. The indicator in this example may be ideological self-identification, just asking respondents: "In politics today, do you think of yourself as very liberal, somewhat liberal, moderate or middle of the road, somewhat conservative, or very conservative (we combined the "very" and "somewhat" categories).
If the hypothesis is
rejected, maybe the indicator is not valid.
Religiosity example of a
theory.
At the theoretical level,
the two principal concepts are Social Deprivation and Religiosity. The
principal hypothesis at the theoretical level is that people who are socially
deprived are more likely to be intensely religious than are people who are not
socially deprived.
Operationalizing the concepts is to choose valid, specific
indicators of those concepts. One indicator of religiosity might be frequency
of church attendance. An indicator of social deprivation might be annual family
income before taxes. The major problem with operationalizing one's concepts is
whether the indicators are valid measures of those theoretical concepts. Is a
person who attends church twice a week necessarily more religious than someone
who never attends church, but who reads the Bible and prays daily? Is a person
with a large family income, but who also has a large family size, necessarily
well-off financially? Can you think of more valid indicators of these concepts
of social deprivation and religiosity?
Hypothesis Testing measures the relationship between the
indicators. Are people with low family incomes more likely to attend church
weekly, compared to people with high family incomes? Are people with lower net
financial worth more likely to pray daily, compared to people with high net
financial worth? If your hypothesis is rejected, there may be two reasons.
Perhaps your theory is rejected, or perhaps your indicators are not valid
measures of your concepts.
Actual Results of This
Hypothesis Test:
Using the 2004-2010
Mississippi Poll, no significant relationship was found between reported family
income and reported frequency of church attendance.
YOUR RESEARCH PAPER (the
first part of your paper which is due January 31)
1) Introduction- discuss
the importance of your subject. Discuss your initial expectations. Example of
gender gap in party identification--why are women slightly more Democratic than
are men? Why is this subject important? Why do you think this female Democratic
bias is occurring?
2) Your model and
hypotheses. List all five of your hypotheses, and draw your model.
Example of a model and
its hypotheses:
Assume that sex is the earliest, independent variable; party identification is
the latest, dependent variable; ideological self-identification and income are
the two intervening variables located between sex and party identification.
SEX........(H1).......>
Ideology .....(H2).....> PARTY
Male or...................(H3)..............................> IDENTIFICATION
Female.....(H4)........> Income ......(H5)........> (D or R)
The hypotheses are:
H1: Women are more likely to be liberal, compared to men.
H2: Liberals are more likely to be Democratic in party identification, compared
to conservatives.
H3: Women are more likely to be Democratic in party identification, compared to
men.
H4: Women are more likely to have lower incomes, compared to men.
H5: Lower income people are more likely to be Democratic in party identification,
compared to higher income people.
3) Literature review. Skip
this for this first part of your paper. It will be due at a later time. In your final, rewritten research paper, you should put the Literature Review after the Model and Hypotheses section and before the Methodology section.
4) Methodology section. Check
the sample
student paper, which you can cut and paste from for a basic outline (it
follows):
METHODS
To test my model, I use data taken from The Mississippi Poll project, which has been carried out over the years through a series of statewide public opinion polls conducted by the Survey Research Unit of the Social Science Research Center (SSRC) at Mississippi State University and led by political science professor Stephen D. Shaffer. I combined or pooled telephone surveys conducted in three years-2010, 2012, and 2014 to maximize my sample size and minimize my sample error. The 2010 Mississippi Poll surveyed 604 adult Mississippi residents from April 5 to April 25, 2010, and had a response rate of 42%. The 2012 Mississippi Poll surveyed 439 adult Mississippi residents from April 2 to April 25, 2012, for a response rate of 26%. The 2014 Mississippi Poll surveyed 350 residents from April 7 to April 30, 2014, with a response rate of 31%. The three years combined contained 1,393 respondents. With 1,393 respondents interviewed, the sample error is 2.7%, which means that if every adult Mississippian had been interviewed, the results could differ from those reported here by as much as 2.7%. The pooled sample was adjusted or weighted by demographic characteristics to ensure that social groups less likely to answer the surveys or to own telephones were also represented in the sample in rough proportion to their presence in the state population. In all three years, a random sampling technique was used to select the households and each individual within the household to be interviewed, and no substitutions were permitted. The SSRC's Computer Assisted Telephone Interviewing System (CATI) was used to collect the data.
For my analysis I chose four variables that were included in three years of the Mississippi Poll. The first variable, sex was a simple question as respondents were asked to choose between male or female. The second variable, race, was also straightforward as respondents chose between white, black, or other. For Ideology respondents were asked to self-identify based on the following question: “What are your political beliefs? Do you consider yourself very liberal, somewhat liberal, moderate or middle of the road, somewhat conservative, or very conservative?” Finally, attitudes on the death penalty were measured by asking the following question: “For someone who is convicted of murder, do you generally favor the death penalty, life in prison without parole, or a jail term that is shorter than for the rest of someone’s life?”
For two of my variables I recoded or combined categories in order to make sure I had enough people to analyze using multivariate tables. The five ideological self-identification categories became three groups: liberals now included anyone who self-identified as “very liberal” or “somewhat liberal”, conservatives now included anyone who self-identified as “very conservative” or “somewhat conservative”, and the category in the middle, “moderate or middle of the road” became an intermediate “moderate” group. The three answers to the death penalty question became two dichotomous groups: those who answered in favor of using the death penalty for someone convicted of murder were now the “favor” group, and those who answered in favor of life without parole or a jail term shorter than the rest of someone’s life were now the “oppose” group. Sex did not have to be recorded because it only had two options. Race also only had two options so did not have to be recorded.
Notice
that she has three short paragraphs: 1) Information about the years studied; 2)
The exact wording of the questions examined; 3) How she plans to combine and
recode categories of each variable so that she has enough people to analyze in
the multivariate analyses. Where do you get all of this information?
The first
paragraph which has the number of people survey each year and when they were
interviewed is found here.
The
sample error of your combined polls can be calculated from this table for calculating sample error for your pooled dataset. Just
use the first column percentages, which assumes a closely divided population with
a 50-50 split on your political subject.
The
second paragraph that has the exact question wording of each of your variables
is found in the cumulative Mississippi Poll codebook.
The
third paragraph can rely on the codebook above and your own thoughts about how
you would like to recode your variables. I will provide you feedback on your
own ideas.
Remember, a quick
and easy to read summary of the Mississippi Poll codebook to
choose your four variables is available here.