Univariate Statistics and Crosstabulation Analysis

   1   HW1 Poli 101

Terrorism, Education, and Ideology

1

Using the PPIC January 2016 data set contains Q30, an ordinal indicator for use as our

DV, measuring whether terrorism and security are perceived to be a problem in California. The

question wording is “On another topic, how much of a problem is terrorism and security in

California today? Is it a big problem, somewhat of a problem, or not much of a problem?” to

measure fear of terrorism. Missing Values are declared for respondents who say that they don’t

know (8) or refuse to answer (9), because these amount to less than 1% of the responses and

cannot be reasonably combined with another response.

2.

In the frequency distribution of Q30 it is clear that there is considerable variation across

respondents as to their perception of terrorism.

How much of a problem is Terrorism?

Valid Percent big problem 41.5

somewhat of a problem

35.9

not much of a problem

22.5

Total 1690

Source: PPIC Jan 2016

The most common (modal) response is that terrorism and security are a “big problem”

with somewhat fewer saying it is “somewhat of a problem”, and fewer still that it “not much of

a problem”. This slight positive skew (.34) is not a problem. And the negative values of the

   2   kurtosis score (-1.3) shows that responses are not concentrated in a particular response

category and the distribution is flatter than a bell curve. The median response is that there is

somewhat of a problem regarding to terrorism and security. Overall Q30 seems to be

appropriate for further analysis.

3.

Our first independent variable is the respondent’s highest grade of school completed, to

measure respondent’s level of education as measured by question D7. This is an ordinal

measure ranging from some high school or less to post-graduate training. Missing values are

declared for the 2% respondents who refused to answer, as unknown level of education does not

provide insight into education’s effect on the measured dependent variable Q30. It might be

considered to include those who refused to respond in the lowest level of education due to

respondent embarrassment over low levels of education, but was not done because respondents

may be uncomfortable answering at any level of education.

The second independent variable is political ideology, measured by item Q40 which asks

respondents about their political leanings “Next, would you consider yourself to be

politically…?” The response categories range from very liberal to very conservative, making it

an ordinal measure. Missing values include the less than 3% of respondents who answered that

they did not know or refused to answer, because such responses are unhelpful in measuring the

effect of political ideology on concern over terrorism.

   3   4

We expect that as level of education increases fear of terrorism will decrease because

terrorism does not rank highly in terms of frequency of killing in America, and we believe that

as education increases so does knowledge of the realities of a terrorist threat, hence perceived

threat will decrease with increased levels of education.

We hypothesize that as respondents move from liberal leaning towards conservative

leaning perceived threat of terrorism will increase, because conservatives tend to approve of the

War on Terror more than liberals.

5

Perceiving Terrorism & Security as a problem (Q30) by Education (D7) How much a problem is Terrorism & Security?

Less than HS

HS Grad

Some College

College Grad

Post Grad

big problem 68.2% 48.1% 39.6% 35.2% 26.8% somewhat a [problem

20.2% 36.0% 39.1% 36.3% 43.0%

not much a problem

11.6% 15.9% 21.2% 28.4% 30.3%

N= 198 308 396 443 314

Tauc =.211

Source: PPIC Jan 2016 Survey

Over two-thirds of the survey respondents who have some high school or less (68.2%)

see terrorism as a big problem. And reading across the top row of the table shows that this

percentage shrinks steadily as education increases to 48%, 40, 35 and finally 28% among those

with a post graduate education. A similar progression, only in the opposite direction is evident

in the bottom row of the table with just under 12% of in the lowest category of education seeing

   4   terrorism and security issues as not much of a problem while 30% of those most educated

grouo feel this way. Tau-c is used to measure strength of association, because both variables

are ordinal, and there are more categories on the independent variable than the dependent,

making a rectangular (5 column by 3 row) table. The tau-c measure of association is .211,

which shows that there is a weak to moderate positive association between the variables.

Perceiving Terrorism & Security as a problem (Q30) by Ideology (Q40) How much a problem is Terrorism & Security?

Very Liberal

Somewhat Liberal

Middle of the Road

Somewhat Conservative

Very Conservative

big problems 26.0% 26.8% 42.9% 52.2% 63.4% somewhat a [problem

30.7% 45.0% 36.8% 34.4% 26.7%

not much a problem

43.3% 28.2% 20.3% 13.4% 9.9%

N= 231 351 517 358 191

Tauc = -.262

Source: PPIC Jan 2016 Survey

As expected, it can be seen in the second cross tabulation between Q30 and Q40 that

more conservative resondents are more likely to believe that terrorism is a big problem, and that

liberal respondents are more likely see terrorism as not much of a problem. This can be seen in

the second cross tabulation’s column percentages. Among those who identify as very liberal

26% responded that terrorism was a big problem, while 43.3% believe it is not much of a

problem. In contrast, among respondents who identify as very conservative, 63.4% believe that

terrorism is a big problem, while 9.9% believe it is not much of a problem. We again used the

tau-c to measure strength of association, because both variables were ordinal, and the

   5   crosstabulation is rectangular. The tau-c measure shows that there is a moderately strong

negative association between variables, at -.262 meaning that if a respondent is liberal they are

less likely to see terror as a big problem.

6.

This first cross tabulation supports our view that respondent with lower education levels

are more likely to see terror as a problem, which may be caused by lack of knowledge

surrounding the frequency of terrorist attacks in America and California specifically.

This second rosstabulation allows us to claim that because conservatives are more

accepting of the war on terror, they are more likely to believe that terrorism in California is a

big problem.

7.

The independent variable political ideology or leaning is a better predictor of concern

over terrorism and security than is level of education. This can be seen in the stronger measure

of association for political leaning than for level of education. This stronger measure of

association may be caused by the underlying political ideologies of various levels of education,

for example a respondent who had some post graduate education may also identify as very

conservative.

   6   8.

Syntax

frequency variables=d7 q40 q30 /statistics mode median skew kurtosis. missing values q30 (8,9). missing values d7 (9). crosstabs /tables=q30 BY d7 /cells=column count /statistics ctau.

missing values q40 (8,9). missing values q30 (8,9). Crosstabs tables = q30 by q40 /cells = column count /statistics = ctau