Anda di halaman 1dari 17

1

Confidence in the Federal Government and its Contributing Factors


Jordan Kelch
Quantitative Methods 4308
Spring 2015

Abstract:
Responses in the 2008 GSS were used to compare church attendance, income, and
political party identification separately against the degree of confidence respondents had
in the executive branch of the government. Analysis concluded that there was a strong
relationship for religiosity and party identification, but not for income.

Introduction
The contemporary socio-political landscape can be perceived as one of the more
partisan in the history of the country. This study makes no claims as to the validity of,
the effects of mass media saturation on the propagation of, or the underlying political
commentary that defines this perception. Rather, it identifies that the perception exists
and that it indirectly affects public policy by virtue of existing. The linkage between the
two is fraught and difficult to define. In a democracy, the constellation of influencing
factors on individual representatives to the federal government alone represents a fairly
difficult methodological puzzle which must be solved, at least in part, prior to any
rigorous academic study. To attempt the same in the aggregate of representatives,
bureaucrats, agencies, employees, lobbies, and constituencies that form the social entity
heretofore termed government seems almost impossible.
The goal here, then, is to pick a discreet element of the puzzle and identify some
of the constructs that influence its construction. This piece of the puzzle is what might be
called the publics interpretation, in aggregate of the efficacy and competency of its
representative elements in the political sphere. Public confidence in government might
suffice as an alternative description. Even in that extremely reduced scope of
questioning, further clarification is necessary for the sake of pure practicality. This
confidence is no doubt a function of, at the very least and not limited to, regional trends

in political leaning, the voting trends of individuals, corporate influence in the form of
information dissemination in the media, environmental influences such as quality of life,
historical social inequalities directed at specific demographic elements, and so on and so
forth. In order to disentangle as many of these elements as possible, an effort must be
made to boil the variables down to a bare minimum that encompasses most or all of these
things in a few categories. The ones chosen for study, then, are religiosity, income, and
political affiliation as they relate to a proxy for overall confidence in our governing
institution, in this case confidence in the executive branch of the federal government. If
significant positive or negative relationships can be discerned from any or all of these
pairings, it might indicate that the confidence placed in each of the mammoth institutions
represented therein (religious organizations, class divisions, and political affiliations)
might be an important indicator of trust for modeling purposes.
Literature Review
Obviously, the field of political science is extremely interested in anything that
has to do with approval ratings and voting trends. Academics specializing in the
sociology of religion are similarly interested in the effect of levels of religiosity on
secular behavior. Conflict theorists studying norms, power or privilege, capitalist
influence in government, and any number of other relationships are likely to tie it all
together. This collective spotlight is represented by a surfeit of literature related to my
subject. In terms of wide ranging trends, it seems as though correlations are often
approached singly or in a highly generalized way.
In reviewing the literature which seems indicative of this trend, and which is
relevant to the study question, the first category that merits description is studies that

approaches the relationship between either my independent variable and a single


dependent variable, or two of my dependent variables. One set of scholars describes the
ebbs in overall religiosity in this country by way of replicating earlier tests relating social
class to the same. (Estus & Overington, 1970) An article found, not surprisingly, in the
journal Political Psychology studied ties between religiosity and political conservatism.
(Malka, Lelkes, Srivastava, Cohen, & Miller, 2012) This is superficially identical to my
own question. The difference lies primarily in the goal of the analysis within. The
largest scale comparative study found in the process of looking for relevant data
considered a relationship between the stringent application of party systems and
confidence in the institutions of government. (Miller & Listaug, 1990)
The second category of studies were those that studied a tangentially related
subject that bore upon my question. To wit, some of the material concerned the decline
in government confidence as it relates to public policy. (Brooks & Chang, 2001) This
primarily pertains to the question at hand in the sense that there is likely to be a
relationship between majority/minority held political affiliation and the amount of
satisfaction people have with public policy decisions. It might perhaps be called home
field advantage. More specifically, the study opts to use the issues of health care and
education. In the contemporary political climate, the more mobilized voting group on
these issues is likely to be political conservatives complaining of problems with new
health care legislation and educational standardization. The fact that the study did not
find a strong connection between government confidence and policy preference might be
due to this phenomena. That is to say, the connection could be ephemeral. If so, we

might expect to see the sought after connection when we consider it in the light of
polarized party affiliation.
Continuing along this vein, a group of political scientists out of Louisiana looked
at the overarching category of public confidence. It is significant that this study found
connections between short-term evaluations of political events and leaders. (Cook &
Gronke, 2005, p. 784) This can be interpreted as supporting evidence towards the
hypothetical relationship posited above. That is to say, it represents conflicting data
wherein the public was strongly influenced by the contemporary political environment.
That said, they still concluded that the polled public exhibited a trend towards distrust
rather than trust in large, operationally defined institutions.
Other scholars focused strictly on the relationship between voter engagement and
religious denomination. (Smith & Walker, 2013) It will be useful here in the sense that it
focuses inward at the partisanship central to my question without directly engaging with
it. Its results pertain to my variables in the sense that voter turnout might be seen as one
of several manifested attitudes that represent societal levels of government confidence. A
comparison between my results and the results here, which indicate that mainline
Protestants manifest voting patterns strongly indicated by their level of religious
observance, might indicate that the variable ATTEND is one of the stronger indicators
towards trust or distrust in government institutions.
The final relevant study of concern here primarily investigates what might be
defined as levels of confidence in micro level government, where the federal level would
be considered macro. It is relevant in the sense that one of the indexes used to contrast
polled levels of confidence is quality of life in communities being studied. While this is a

far broader category than income, it stands to reason that income bears a strong
relationship to quality of life on a strictly statistical basis. The study found that
confidence was, at least in part, a function of what they deem subnational politics, so if
nothing else this work is useful as a gentle warning against ecological fallacies in the
interpretation of my results and in any future modeling.
Gaps in the Literature
Having explored the available literature, at least one very evident gap is obvious.
It is difficult to find a holistic study which independently or concurrently interrogates the
relationship between all of the variables chosen. Also, the level of confidence in the
government is typically determined from a customized study on the topic or from another
form of specific polling. By using the confidence strictly in the executive, and by
looking at the opposition party so to speak, it might be possible to determine trends and
attitudes that get washed away when the data attempts to nullify political partisanship.
Alternatively, the extant body of literature tends to consider strongly issues of political
engagement along with issues of public trust. I am attempting to ascertain only what
variables might influence the latter. To what extent this might pertain to political
engagement and voting patterns my question has no bearing. It is conceivable that in the
event of strong statistical correlations the results might be useful in follow up work set
towards this determination.

The null hypotheses that will be used in my analysis are as follows.


1.) H0: Church attendance is not related to confidence in the executive
branch.
2.) H0: Income is not related to confidence in the executive branch.
3.) H0: Party identification is not related to confidence in the executive
branch.
Given strong correlations or multiple failed null hypothesis, a more multivariate format
might deliver more informative results, but these hypothesis should serve to cover the
identified gaps in the literature. Namely, that no study investigate this specific set of
relationships and that this specific and directed analysis investigates relationships
between confidence and lifestyle variables rather than confidence and voting patterns of
lifestyle variables and voting pattern.
Data and Methods
The raw data that will be used during the course of statistical testing here is
derived from the General Social Survey (GSS) taken in 2008. It is administered by the
National Opinion Research Center (NORC) on an annual basis to a randomly selected
sample of individuals from across the nation determined to be representative of the
greater population. It is of paramount use in investigations such as this one in that it does
not solely measure demographic information (race, age, gender, income, etc.), but also
prevalent attitudes towards important social issues (homosexuality, gun control,
religiosity, etc.).

The statistical testing will be carries out using the Statistical Package for the
Social Sciences (SPSS) software made by IBM Inc. Each null hypothesis will be recoded
and run through the appropriate test. Only the responders who provided responses to the
individual questions on the survey will be counted in the analysis.
The first hypothesis attempts to discern the relationship between religiosity and
confidence. The variables used will be entitled ATTEND and CONFED in the GSS data.
The former, and independent, of the two is an ordinal variable and the latter, and
dependent, is nominal. The ATTEND variable will become NEWREL with three
categories. 1 will be never or annually, 2 will be monthly, 3 will be weekly or
daily. The old category values that will have been combined will have been 0-3 (never,
less than once a year, once a year, and several times a year) will become 1 (never or
annually), 4-6 (once a month, 2-3X a month, and nearly every week) combine into 2
(monthly), 7 and 8 (every week and more than once per week) combine into 3 (weekly).
The new variable will be ordinal. The recoded variables will be analyzed using a ChiSquare test.
The second hypothesis investigates the possibility of a relationship between
income and confidence. The independent variable will be the interval level income, called
RINCOM06, and the dependent will be CONFED again. ) The RINCOM06 variable will
be recoded into an ordinal form with 4 categories. 1 will be values 1-9, representing
incomes from 0-12499. This will likely be labeled At or Below Poverty Line. 2 will
be 10-18, representing incomes from 12500 to 49999. This will be labeled Lower
Middle Class, as they detail a portion of the population with incomes that lie between an
approximation of the poverty line and an approximation of the national average income.

3 will be 19-24, for incomes 50000-149999. The label will be Upper Middle Class.
Finally, 4 will be 25, for incomes 150000+. The label will be Wealthy. The recoded
variable will be NEWINC. I will also perform a Chi-Square analysis here.
The third hypothesis attempts to relate party identification to confidence in the
executive. The independent variable will be entitled PARTYID and the dependent will
remain CONFED. As is, PARTYID is a nominal variable, but it will be recoded to
include those respondents who reported leanings towards one party or another into that
partys category. ) PARTYID will be recoded to 1 Democrat, 2 Republican, and 3
other. The condensed values will be 0-2 for 1, 4-6 for 2, 3 and 7 for 3. The new
variable (NEWPID) will be nominal. As with the other hypothesis, a Chi-Square analysis
will be performed here.
Results
Table 1: Percentages
Percentages

Descriptive
Statistics

Only Some

INCOME (totals)
At or Below Poverty Line
Lower Middle Class
Upper Middle Class
Wealthy
RELIGIOSITY (totals)
Never or Annually Attend Church
Monthly Attend
Weekly or Daily Attend
POLITICAL AFFILIATION (totals)
Democrat
Republican
Independent

A Great Deal
10%
14.19%
8.45%
9.85%
11.11%
10.02%
9.60%
9.34%
11.44%
9.91%
7.12%
16.21%
5.80%

49.87%
47.10%
52.42%
46.80%
50%
48.68%
44.99%
58.37%
49.27%
48.64%
44.39%
55.02%
48.66%

Hardly Any
40.13%
38.71%
39.13%
43.35%
38.89%
41.30%
45.40%
32.30%
39.30%
41.45%
48.48%
28.77%
45.54%

Mode

10

The above table describes how respondents fell into each of the applicable
confidence categories after recoding and reorganization. It is noted where the majority of
the respondents fell after the recoding took place for the purposes of taking a reasoned
view of the results of the statistical analysis. That is to say, that if one were to imagine
this aggregate personified, it would be lower middle class economically, rarely if ever
attend church, and tend to vote either for democratic or independent candidates. It is also
notable that, economically, it would appear that over the entire income spectrum
respondents were more likely to opt for the lesser of the two confidence categories.
Religiously, the greatest portion of those who attend church the most also had the
highest degree of confidence in the fed, and the greatest portion of those who attend
church the least corresponded to the least confidence. Similarly, the most Republicans
reported the most confidence and the most Democrats reported the least. Distributions
are reported by the charts below.
Table 2: Income

Distribution of Trust in the Executive by Economic


Category
Wealthy
Upper Middle Class
Lower Middle Class
At or Below Poverty Line
0

20

A Great Deal

40
Only Some

60

80

Hardly Any

100

120

11

Table 3: Church Attendance

Distribution of Trust in the Executive


by Church Attendence
Weekly or Daily
Monthly
Never or Annually
0

20

40

A Great Deal

60

Only Some

80

100

120

Hardly Any

Table 4: Political Affiliation

Distribution of Trust in the Executive


by Political Affiliation
Other
Republican
Democrat
0

20

40

A Great Deal

60
Only Some

80
Hardly Any

100

120

12

Test #1 produced the following results.

NEWREL

Total

Never or
Annually
CONFID. IN EXEC BRANCH OF
FED GOVT

A GREAT
DEAL

Count
Expected Count

ONLY SOME

% within
NEWREL
Count
Expected Count

HARDLY ANY

% within
NEWREL
Count
Expected Count

Total

% within
NEWREL
Count
Expected Count
% within
NEWREL

Monthly

Weekly or
Daily

70

24

39

133

73.1

25.8

34.2

133.0

9.6%

9.3%

11.4%

10.0%

328

150

168

646

354.9

125.1

166.0

646.0

45.0%

58.4%

49.3%

48.7%

331

83

134

548

301.0

106.1

140.8

548.0

45.4%

32.3%

39.3%

41.3%

729

257

341

1327

729.0

257.0

341.0

1327.0

100.0%

100.0%

100.0%

100.0%

The Chi-Square value was 16.29, and the p value was .003, well below the 95%
confidence interval demanded by the test. There is a statistically significant relationship
between confidence in the executive branch and attendance at religious services. It can
be observed that the never or annually respondents exceeded their expected count only
in the hardly any category, the monthly only in the only some categories, and the
weekly or daily in both lower categories. In effect, this test indicates only that the
variance between the expected counts and the actual counts of responses is outside the
scope of plausible natural variability. It fails to indicate exactly which respondent
categories might point towards a trend, but succeeds at pointing towards a likely

13

relationship between the two greater categories.


Test #2 produced the following results.

The data did not optimally conform to this test, as the final count of the wealthy
respondents after recoding was very low. The recoding isnt at fault, as the same would
have been true for several rungs of the previous, much more precise organizational
structure of the variable. The ultimate pearson Chi-Square value was 5.62, and the p
NEWINC

Total

At or Below
Poverty
Line
CONFID. IN
EXEC BRANCH
OF FED GOVT

A GREAT
DEAL

ONLY
SOME

HARDLY
ANY

Total

Lower
Middle
Class

Upper
Middle
Class

Wealthy

Count

Expected
Count
% within
NEWINC
Count
Expected
Count
% within
NEWINC
Count
Expected
Count
% within
NEWINC
Count
Expected
Count
% within
NEWINC

22

35

20

79

15.5

41.4

20.3

1.8

79.0

14.2%

8.5%

9.9%

11.1%

10.0%

73

217

95

394

77.3

206.5

101.2

9.0

394.0

47.1%

52.4%

46.8%

50.0%

49.9%

60

162

88

317

62.2

166.1

81.5

7.2

317.0

38.7%

39.1%

43.3%

38.9%

40.1%

155

414

203

18

790

155.0

414.0

203.0

18.0

790.0

100.0%

100.0%

100.0%

100.0%

100.0%

value was .468. This considerably exceeds the .05 maximum allowed to prove a
relationship. Therefore, with a confidence interval of 95%, it is very unlikely that there is
a statistically significant relationship between respondent class status as measured by
reported income and confidence in the federal government. If any useful information is
hinted at by a comparison of expected and actual responses, it would seem to indicate that

14

there is a possible trend towards less trust in the government as income increases. The
chi-square analysis cannot make this claim, however.
The results to the third test are as follows.

NEWPID
CONFID. IN EXEC
BRANCH OF FED GOVT

A GREAT
DEAL

ONLY
SOME

HARDLY
ANY

Total

Count
Expected
Count
% within
NEWPID
Count
Expected
Count
% within
NEWPID
Count
Expected
Count
% within
NEWPID
Count
Expected
Count
% within
NEWPID

Total

Democrat

Republican

Other

47

71

13

131

65.4

43.4

22.2

131.0

7.1%

16.2%

5.8%

9.9%

293

241

109

643

321.0

213.0

109.0

643.0

44.4%

55.0%

48.7%

48.6%

320

126

102

548

273.6

181.6

92.9

548.0

48.5%

28.8%

45.5%

41.5%

660

438

224

1322

660.0

438.0

224.0

1322.0

100.0%

100.0%

100.0%

100.0%

The chi-square value here was 58.43, and the p value was .000. There is a very
strong statistical relationship between political party affiliation and confidence in the
federal government. The actual counts vary strongly from the expected counts in
opposing directions, with the democratic respondents being much more likely to respond
that they have very little trust and the republican responding that they have a great deal of
trust.

15

Conclusions
The most important result from this study must be that given three variables to
compare to a political attitude, the only one that produced an extremely strong result was
the one that was also a political attitude. Furthermore, the other relationship proven was
between two large institutions with known to homogenize social ideas and behaviors in
their membership (Malka et al, 2012). Often, other research in the field relies upon
secularization theory to explain diverging political attitudes, the amplification of what
might be termed the religious right, and the overall trend towards a decrease in
American (specifically) religiosity (Klieman, 1996). This study tested two variables
separately against a third instead of against each other, as is the practice in much of the
relevant research to date.
The result of this juxtaposition is the conclusion that there is definitely a
relationship between confidence and religiosity, and confidence and political affiliation.
With further study, this could provide evidence to show that the link between the two
independent variables that is often sought is at least a little bit specious. There may be
some greater organizational structure at work that causes individuals to formulate and
express their political ideas.
There were several weaknesses in this approach as well. First, there must be a
better way to measure income in order for a good result. The GSS data is insufficient to
the task because it is self-reported and then delivered in an ordinal format. The selfreporting is probably an unavoidable issue, and fortunately one that will wash itself out
given a large enough sample. The other is not, if there were a similarly representative

16

poll that asked for an approximate numerical value so that a ratio level of measurement
could be used, it might produce more reliable results in statistical analysis.
Second, it would be preferable to perform secondary and tertiary stages of
analysis on my chosen variables. These tests do not in any way rule out speciousness,
and in one case may reveal only a tautology. That is in the political affiliation versus
public confidence test. It seems entirely feasible that the amount of confidence in the
government one has contributes to the party identification one chooses. If this is the case,
then the relationship is dependent upon itself. Given an expanded schedule of testing, it
might be possible not only to rule out some of these vagaries, but to search for an actual
trend or correlation between data points instead of just a relationship. All of these steps
must be left to further research.

References
Brooks, C., & Cheng, S. (2001). Declining Government Confidence and Policy
Preferences in the U.S.: Devolution, Regime Effects, or Symbolic Change? Social
Forces, 79(4), 1343-1375. doi:10.1353/sof.2001.0032
Gallup.Com - Daily News, Polls, Public Opinion on Politics, Economy, Wellbeing, and
World. (n.d.). Retrieved February 11, 2015, from http://www.gallup.com/home.aspx

17

Hout, M., & Fischer, C. S. (2002). Why More Americans Have No Religious Preference:
Politics and Generations. American Sociological Review, 67(2), 165.
Kelleher, C. A., & Wolak, J. (2007). Explaining Public Confidence in the Branches of
State Government. Political Research Quarterly, 60(4), 707-721.
doi:10.1177/1065912907304496
Kleiman, M. B. (1996). Public Confidence in Religious Leaders: A Perspective from
Secularization Theory. Review of Religious Research, 38(1), 79-87.
Malka, A., Lelkes, Y., Srivastava, S., Cohen, A. B., & Miller, D. T. (2012). The
Association of Religiosity and Political Conservatism: The Role of Political
Engagement. Political Psychology, 33(2), 275-299.
Miller, A. H., & Listhaug, O. (1990). Political Parties and Confidence in Government: A
Comparison of Norway, Sweden and the United States. British Journal of Political
Science, 20(03), 357. doi:10.1017/S0007123400005883
Wolak, J., & Palus, C. K. (2010). The Dynamics of Public Confidence in U.S. State and
Local Government. State Politics & Policy Quarterly, 10(4), 421-445.
doi:10.1177/153244001001000407

Anda mungkin juga menyukai