Tuitorial By: Prof

4. Assessment Tasks/Activities

(Indicative of likely activities and tasks designed to assess how well the students achieve the CILOs. Final details will be provided to students in their first week of attendance in this course)

CILO No

Type of assessment tasks/activities

Weighting

(if applicable)

Remarks

CILO 1-3

AT1: Data Analysis Report Writing and Critique

60%

CILO 1-3

AT2: Exercise

30%

CILO 1-3

AT3: Class Participation

10%

Further description of ATs:

AT1: Research report writing

Each student is to write a data analysis report (2,000 words) of a given dataset demonstrating mastery of advanced knowledge and skills in applying practical statistical methods to social research. Each student needs to develop and work on his/her report continuously throughout the semester. The student needs to consult the instructor at workshops about the writing.

Research report writing and critique need to be a coherent task to report the student’s own statistical analysis. The whole task is to proceed in the following step:

Identify a series of research questions from an existing dataset

Review and therefore critique existing statistical analyses related to the research questions or the like

Conducting statistical analyses to answer the research questions or hypotheses, using at least 5 different analytic methods introduced in the course

Writing results from the analyses

Concluding by showing the lesson acquired in the task—discussion of findings being paramount

Overall, the writing needs to be tidy, read in a good format, and avoid any problem of “copy and paste.” It is necessary to consult the literature for reference about the presentation format. References are necessary.

AT2: Exercise

Each student is to complete exercises to solve problems of the application of practical statistical methods to social research. The student needs to work independently and save work done in each workshop in a flash drive for later compilation. Eventually, each student is to write up key findings in about 1,000 words, to discuss results in five statistical tables refined from the statistical output saved in the classes. Essentially, the report needs to have a brief introduction about research questions and hypotheses and a substantial discussion summarizing the results and insights. References are desirable.

AT3: Class participation

Each student needs to participate actively in classes, including workshops, presentations, discussions, and sharing sessions, to demonstrate thorough and intensive knowledge about applying practical statistical methods to social research. At the end, each student needs to write a one-page performance report (about 500 words) about the class participation and learning in concrete ways.

Assessment Task

Criterion

1. Data Analysis Report Writing and Critique

Validity of analysis, coherence of reporting, and demonstration of knowledge building

2. Exercises using SPSS

Validity of analysis, clarity of reporting, and demonstration of insight

3. Class Participation

Demonstration of class participation and learning

A+

**Department of Applied Social Sciences**

**SS5428** **Applied Social Statistical Analysis**

**Financial crisis attitude, trust in authorities, and risk perception among CityU College students **

Research Report & Exercise & Class participation report

May 4 2015

**Introduction**

Risk Perception refers to people's subjective assessment and judgment of risk, as well as attitudes and decision-making tendencies arising therefrom. It covers people's risk perception, comprehension, memory, evaluation, cognitive processes throughout the perception process. Under the state of emergency, individuals often make decisions based on intuition and experience, intuition bias due to the presence of these decisions often lead to individuals or groups of irrational behavior. Based on prospect theory, this paper analyzes the state of individual behavior under emergency decision-making model, meanwhile, it built in risk attitude, risk perception, and government trust government information supplied to the dimension mechanism framework. What’s more, government, through effective information supply group behavior can reduce the public space, reducing the uncertainty of emergency response.

The research model of risk perception has three common heuristic: the representative heuristic, availability heuristic, and anchoring heuristic. Besides, researchers gradually extended the difference of risk perception into twenty types. The reason that led to the difference of risk perception is divisible into two main factors: Internal factors include personality characteristics of the individual, knowledge and experience, achievement motivation, expectations of risk "loss" or "benefit” and other subjective factors. External factors include the nature of the risk, risk predictability, risk controllability and other objective factors. However, the risk perception occurs when individuals have judge of risk deviation or deviation tendency due to personally cognitive limitations, personal motivation and attitude toward risk, and trust in authorities.

**Literature review**

*Risk perception*

Factors affecting risk perception can be broadly grouped into two categories: The first category is idiosyncratic risk activities that also related to the detectability of the risk. Besides that the second category is personal characteristics such as knowledge, personal responsibility and socio-cultural and controllability of the government.

*Knowledge*

US researcher Wibecke Brun found that "the level of risk understanding" and "risk seriousness" explained most of the risk perception (Wibecke Brun, 1992). Meanwhile, Norwegian researchers confirmed that the knowledge of risk plays an important role to the perception of risk. In a study of natural hazards event entry, has been "positive - negative", "potential impact", "novelty" of the three-dimensional model consisting of which shows that people in different types of risk perception. Based on the previous finding, I hypothesis that:

H1: The higher level of risk knowledge, the lower the risk perception will be.

*Detectability*

There are two basic risk characteristics influence people's risk judgment and perception, and that is the degree of familiarity risk that is so called knowledge of the risk, on the other hand, the risk detectability can affect the individual risk cognition that may affect the risk perception. Some scholars also found risk perception has a negative correlation with the risk detectability. Researcher Roger E. Kasperson and Ortwin Renn point out that the society had the amplification to the risk, which is known as “The Social Amplification of Risk” (Roger E. Kasperson and Ortwin Renn, 1988). Based on the Social Amplification of Risk theory, the risk signal would be amplification by individuals and society, including communications, news media and detectability. Whereas, detectability of risk resulted in behavioral responses, which in turn, lead to a negative effect to the risk perception.

H2. The more risk detectability, the higher risk perception will be.

*Controllability of government*

Previous study, Huang et, al point out that the risk controllability of government is closely linked to the nature of risk (Huang et, al, 2010). The controllability of the government also related to the capability of the authorities, the capability played the role as the monitor to the personal risk perception. The uncertainty of the relationship between the risk perception and government capability also lead to the uncertainty of the controllability of government to the risk perception. However, a research which conducted to test the risk perception on Blue-Algae Bloom risk that in the Sihong village in china, it point out that the government construability to the risk has a significant relationship with the willingness of risk acceptances and have a positive effect on the risk perception.

H3. The higher controllability of risk by government, the lower risk perception will be.

*Trust in authorities*

Trust in authorities has an important position in risk communication, and even determine the success or failure of risk communication (Kittelsen, 2009). What’s more, risk communication refers to the expectations of the general public received information is correct or credible via communication with the government. The risk communication represented the trust from the general public to the authorities.

Previous studies mentioned the trust in authorities can have a significant relationship with the risk communication, which in turn can have a closely link with the individual risk perception. Earlier studies suggested that trust in authorities could convey though the risk communication the risk signal to the public and maintain social stability. In light of the previous study, the trust of authorities may have a positive affects to the risk perception.

H4. The more trust in the authorities, the less risk perception will be.

*Demographics*

In the past study of risk perception, demographic characteristics have been the indispensable analytical factors. In the study of demographic background, more commonly used the following factors such as gender, age, educational level, and major (Wiersema, M, 1992). For example, age represents a personal experience and risk tendencies, thus affecting the personal perception of the risks. What’s more, with the growth of age, work capacity reduction and the sensitivity of the risk also variation. The object of this study examine students of the City University of Hong Kong, I will examine the demographic characteristics impact on risk perception from character age, gender and major.

H5. “Age”, “Gender”, “Level”and “Major” are predictive in risk perception.

**Conceptual framework**

Attitude towards risk

- Knowledge
- Detectability
- Controllability of government

Distrust in authorities

Trust in authorities

Risk perception

**Methodology**

*Measurement*

A survey was conducted to undergraduate students of City University of Hong Kong. The questionnaire designed by investigator Jacky Chau-kiu Cheung and was designed to inquire about the students’ perception of the risk, attitudes towards the financial crisis, SARS-like epidemic and terrorist, government trust and so on. Other variables such as risk of SARS-like epidemic and risk of terrorist attack were also included in the questionnaire. Totally the questionnaire has 90 items contained the items on the perceptions and attitudes about risks and the main background information.

*Sample Size*

The questionnaire was administered in class between **and ** and resulted in a sample size of 350 undergraduates. Majority of the respondents were students from the main programs of the Department of Applied Social Sciences were surveyed. They were primarily year 1, 2, 3 and 4 students enrolled in the criminology, psychology, counseling, sociology and social work programs. The education levels of participants are range from the bachelor's level, master's level and doctoral level.

**Data analysis**

Input data using the Statistical software.

For points:

- Factors analysis used to select the impact factors to the risk perception.
- Normality test of all the demographic variables.
- New components were computed after the “Factor analysis”, then test new normality.
- ANOVA used to test the difference in means among the different major groups.
- Correlation would be tested among all the variables in this research.

- Independent variables: “Knowledge”, “Detectability, “Controllability of government”, “Trust in authorities”, “demographic characteristics”.
- Dependent variable “perceived risk of financial crisis”, “SARS-like epidemic”, and “terrorist”.

- Univariate linear regression used in the three dependent variables.
- Multivariate linear regression model used in outcome variables.
- Logistic regression model used in the three DVS and use the IV as predictors respectively.

**Result **

**Factor Analysis: **Reliability

**Reliability Statistics**

Cronbach's Alpha

Cronbach's Alpha Based on Standardized Items

N of Items

.478

.481

3

IV: *Knowledge *

**Item-Total Statistics**

Scale Mean if Item Deleted

Scale Variance if Item Deleted

Corrected Item-Total Correlation

Squared Multiple Correlation

Cronbach's Alpha if Item Deleted

Knowledge about financial crisis

10.8020

8.337

.288

.088

.397

Knowledge about a SARS-like epidemic

10.2995

8.383

.332

.111

.325

Knowledge about terrorist attack

10.9086

7.987

.277

.080

.419

From the two table, the correlation of the three items are not obvious/correlated based on the Cronbach's Alpha is .478, and one item deleted Alpha is .397, .325, and .4.19. All of them are not at the significant correlation level >.60. Based on the factor analysis result, I will extract the financial result as my focus and specific my variables to “Knowledge about financial crisis”. Also for this reason, my other independent variables “Detectability” and “Controllability of government” also will be materialized to “Detectability about financial crisis” and “Controllability of government about financial crisis”. Moreover, will the total score of selected items as an index/scale score.

IV:* Distrust in authorities*

**Reliability Statistics**

Cronbach's Alpha

Cronbach's Alpha Based on Standardized Items

N of Items

.761

.759

3

**Item-Total Statistics**

Scale Mean if Item Deleted

Scale Variance if Item Deleted

Corrected Item-Total Correlation

Squared Multiple Correlation

Cronbach's Alpha if Item Deleted

Distrust of government policy

11.5288

11.657

.638

.444

.625

Distrust of government officials

11.3560

11.117

.655

.459

.602

Distrust of political parties

11.1990

13.808

.489

.240

.787

From the two tables, the correlation of the three items is well correlated based on the Cronbach's Alpha is .761, and one item deleted Alpha is .625, .602, and .787. All of them are at the significant correlation level >.60. That means I can use the three items total score to measure the independent variable “Distrust of authorities”.

IV: *Trust in authorities*

**Reliability Statistics**

Cronbach's Alpha

Cronbach's Alpha Based on Standardized Items

N of Items

.693

.694

3

**Item-Total Statistics**

Scale Mean if Item Deleted

Scale Variance if Item Deleted

Corrected Item-Total Correlation

Squared Multiple Correlation

Cronbach's Alpha if Item Deleted

Trust in the government of Hong Kong

11.6573

10.067

.530

.281

.573

Trust in legislative councilors

11.4246

11.040

.508

.259

.600

Trust in the laws of Hong Kong

10.2225

11.979

.490

.241

.624

From the two tables, the correlation of the three items is well correlated based on the Cronbach's Alpha is .693which is at the significant correlation level >.60, and one item deleted Alpha is .573, .600, and .624. That means I can use the three items total score to measure the independent variable “Trust in authorities”.

**Distribution**

**Frequencies**

**Statistics**

Distrust

N

Valid

382

Missing

12

Mean

17.0419

**Statistics**

Distrust

Knowledge

risk

age

major

level

detect

sex

N

Valid

382

394

394

371

383

381

394

383

Missing

12

0

0

23

11

13

0

11

Mean

17.0419

16.0051

5.8718

24.0997

2.70

1.4934

5.3342

1.6475

Skewness

-.156

-.084

-.180

2.616

-.159

.465

-.345

-.620

Std. Error of Skewness

.125

.123

.123

.127

.125

.125

.123

.125

Kurtosis

.725

.532

-.062

7.281

4.192

-.930

.315

-1.624

Std. Error of Kurtosis

.249

.245

.245

.253

.249

.249

.245

.249

From this chart and based on the sample is 394 and valid sample size is 382 which means the sample is large enough besides that the independent variables, demographic variables, and dependent variables retained a fair skewness (<3) and kurtosis (<10) level. It is standard normal distribution curve.

**T-Test**

**Independent Samples Test**

Levene's Test for Equality of Variances

t-test for Equality of Means

F

Sig.

t

df

Sig. (2-tailed)

Mean Difference

Std. Error Difference

95% Confidence Interval of the Difference

Lower

Upper

The risk of financial crisis to society

Equal variances assumed

.000

1.000

-.197

381

.844

-.03256

.16535

-.35767

.29256

Equal variances not assumed

-.195

267.513

.846

-.03256

.16701

-.36137

.29626

**Independent Samples Test**

Levene's Test for Equality of Variances

t-test for Equality of Means

F

Sig.

t

df

Sig. (2-tailed)

Mean Difference

Std. Error Difference

95% Confidence Interval of the Difference

Lower

Upper

The risk of financial crisis to you

Equal variances assumed

.672

.413

-.771

381

.441

-.15612

.20258

-.55445

.24220

Equal variances not assumed

-.756

260.245

.450

-.15612

.20657

-.56288

.25063

From the two tables T-test, it shows that the p value of risk of financial crisis to society is equal to 1.00 and risk of financial crisis to you is .413, which both of them are not at the significant level .05. It also means there is not a significant difference in means between two gender groups, group one represent male and group two represent female. That’s point that male and female has not significant different in perceived financial crisis to society or to self.

**ANOVA**

**ANOVA**

Sum of Squares

df

Mean Square

F

Sig.

major

Between Groups

14.480

10

1.448

2.694

.003

Within Groups

199.990

372

.538

Total

214.470

382

level

Between Groups

5.201

10

.520

1.781

.062

Within Groups

108.032

370

.292

Total

113.234

380

age

Between Groups

83.184

9

9.243

.328

.966

Within Groups

10184.126

361

28.211

Total

10267.310

370

From this table, it showed only the major has a significant difference between groups based on the one way ANOVA test P value is .003 which is <.05, while the level and age do not show the significant difference between groups in perceived risk of financial crisis to society.

**ANOVA**

Sum of Squares

df

Mean Square

F

Sig.

major

Between Groups

1.891

10

.189

.331

.973

Within Groups

212.088

371

.572

Total

213.979

381

level

Between Groups

7.856

10

.786

2.812

.002

Within Groups

103.102

369

.279

Total

110.958

379

age

Between Groups

347.212

10

34.721

1.260

.252

Within Groups

9920.098

360

27.556

Total

10267.310

370

From this table, it showed only the level has a significant difference between groups based on the one way ANOVA test P value is .002 which is <.05, while the major and age do not show the significant difference between groups in perceived risk of financial crisis to self.

**Correlation**

**Correlations**

Knowledge about financial crisis

The detectabilityof financial crisis

The controllability of financial crisis by government

The risk of financial crisis to society

The risk of financial crisis to you

Distrust

Trust

Knowledge about financial crisis

Pearson Correlation

1

.440^{**}

.157^{**}

.147^{**}

.247^{**}

-.087

.048

Sig. (2-tailed)

.000

.002

.003

.000

.090

.348

N

394

394

394

394

393

382

391

The detectabilityof financial crisis

Pearson Correlation

.440^{**}

1

.209^{**}

.224^{**}

.246^{**}

-.046

.142^{**}

Sig. (2-tailed)

.000

.000

.000

.000

.370

.005

N

394

394

394

394

393

382

391

The controllability of financial crisis by government

Pearson Correlation

.157^{**}

.209^{**}

1

.144^{**}

.118^{*}

.029

.163^{**}

Sig. (2-tailed)

.002

.000

.004

.020

.574

.001

N

394

394

394

394

393

382

391

The risk of financial crisis to society

Pearson Correlation

.147^{**}

.224^{**}

.144^{**}

1

.422^{**}

.166^{**}

-.007

Sig. (2-tailed)

.003

.000

.004

.000

.001

.892

N

394

394

394

394

393

382

391

The risk of financial crisis to you

Pearson Correlation

.247^{**}

.246^{**}

.118^{*}

.422^{**}

1

.092

-.068

Sig. (2-tailed)

.000

.000

.020

.000

.072

.180

N

393

393

393

393

393

381

390

Distrust

Pearson Correlation

-.087

-.046

.029

.166^{**}

.092

1

-.317^{**}

Sig. (2-tailed)

.090

.370

.574

.001

.072

.000

N

382

382

382

382

381

382

382

Trust

Pearson Correlation

.048

.142^{**}

.163^{**}

-.007

-.068

-.317^{**}

1

Sig. (2-tailed)

.348

.005

.001

.892

.180

.000

N

391

391

391

391

390

382

391

Correlations between “Knowledge about financial crisis”, “The detectability of financial crisis”, “The controllability of financial crisis by government”, “The risk of financial crisis to society”, “The risk of financial crisis to you”, ‘Distrust in authorities”, and “Trust in authorities” were analyzed as result showed in Table.

“Knowledge about financial crisis”, “The detectability of financial crisis”, “The controllability of financial crisis by government”, and ‘Distrust in authorities” were positively correlated with perceived financial crisis risk to society (p < .01). While the “The risk of financial crisis to society’ and the “Trust in authorities” do not show a significant correlation with each other due to the P value is .892 which is much higher than the (p < .01). Besides, “The risk of financial crisis to self” and the “Distrust in authorities” also do not show a significant correlation with each other due to the P value is .072which is a little higher than the (p < .05).

“Knowledge about financial crisis” and “The detectability of financial crisis” showed a high correlation with each other at the Pearson Correlation coefficient is .440. Meanwhile, the two dependent variables also showed the high level of correlation with each other at the Pearson Correlation coefficient is .422.

- “Knowledge about financial crisis” has correlation with “The risk of financial crisis to society” and “The risk of financial crisis to you” respectively is .147** and .247** at a significant different level .001.
- “The detectability of financial crisis” has correlation with “The risk of financial crisis to society” and “The risk of financial crisis to you” respectively is .224** and .246** at a significant different level .001.

“The controllability of financial crisis by government” has correlation with “The risk of financial crisis to society” and “The risk of financial crisis to you” respectively is .144** and .118* at a significant different level .001 and .05 respectively.

**Regression 1 **

**Model Summary**

Model

R

R Square

Adjusted R Square

Std. Error of the Estimate

1

.369^{a}

.136

.114

1.43356

**ANOVA ^{b}**

Model

Sum of Squares

df

Mean Square

F

Sig.

1

Regression

110.760

9

12.307

5.988

.000^{a}

Residual

700.784

341

2.055

Total

811.544

350

a. Predictors: (Constant), level, sex, major, The detectabilityof financial crisis, The controllability of financial crisis by government, Distrust, Trust, Knowledge about financial crisis, age

b. Dependent Variable: The risk of financial crisis to society

**Coefficients ^{a}**

Model

Unstandardized Coefficients

Standardized Coefficients

t

Sig.

B

Std. Error

Beta

1

(Constant)

2.971

.843

3.523

.000

Knowledge about financial crisis

.078

.048

.093

1.628

.104

The detectabilityof financial crisis

.137

.047

.169

2.933

.004

The controllability of financial crisis by government

.086

.047

.097

1.843

.066

Distrust

.053

.017

.173

3.115

.002

Trust

-.006

.018

-.019

-.337

.736

age

.016

.020

.054

.826

.409

sex

.079

.167

.025

.472

.638

major

.355

.128

.158

2.772

.006

level

.242

.174

.086

1.394

.164

In predicting the perceived risk of financial crisis to society using a univariate linear regression analysis, significant predictors are “The detectability of financial crisis” .004 (p < .01), ‘Distrust in authorities”.002 (p < .01) and “Major” .006(p < .01). All the significant predictors exhibited positive effects on perceived risk of financial crisis to society. The strongest effect of “perceived risk of financial crisis to society” is the independent variable “Major of study” (Beta= .355).

**Regression 2**

**Model Summary**

Model

R

R Square

Adjusted R Square

Std. Error of the Estimate

1

.348^{a}

.121

.098

1.79451

**ANOVA ^{b}**

Model

Sum of Squares

df

Mean Square

F

Sig.

1

Regression

151.335

9

16.815

5.222

.000^{a}

Residual

1098.107

341

3.220

Total

1249.442

350

a. Predictors: (Constant), level, sex, major, The detectabilityof financial crisis, The controllability of financial crisis by government, Distrust, Trust, Knowledge about financial crisis, age

b. Dependent Variable: The risk of financial crisis to you

**Coefficients ^{a}**

Model

Unstandardized Coefficients

Standardized Coefficients

t

Sig.

B

Std. Error

Beta

1

(Constant)

3.239

1.056

3.068

.002

Knowledge about financial crisis

.181

.060

.174

3.018

.003

The detectability of financial crisis

.177

.058

.175

3.021

.003

The controllability of financial crisis by government

.090

.058

.082

1.532

.126

Distrust

.034

.021

.090

1.608

.109

Trust

-.044

.023

-.108

-1.933

.054

age

.006

.025

.015

.235

.815

sex

.185

.210

.047

.882

.378

major

-.105

.160

-.038

-.655

.513

level

.226

.217

.065

1.041

.299

a. Dependent Variable: The risk of financial crisis to you

In predicting the “perceived risk of financial crisis to self” using a univariate linear regression analysis, significant predictors are “Knowledge about financial crisis” .003 (p < .01) and ‘The detectability of financial crisis”.003 (p < .01). Both of the significant predictors exhibited positive effects on the “perceived risk of financial crisis to self”, while both of them not show a strong effect of “perceived risk of financial crisis to self” based on the Beta is equal to .181and .177 respectively.

**Discussion**

Based on the SPSS analysis result, the following conclusions emerge.

First, “Knowledge about financial crisis” has a positive correlation with “perceived risk of financial crisis to self”. Which also can support the H1 is correct (H1: The higher level of risk knowledge, the lower the risk perception will be.) However, “Knowledge about financial crisis” has not showed the significant correlation with the “perceived risk of financial crisis to society”. It also means when the risk represented the risk to self, the H1 while be proved, while, when the risk represent the risk to society, the H1 would not be proved.

Second, “The detectability of financial crisis” has a positive correlation with “perceived risk of financial crisis to self” and “perceived risk of financial crisis to society”. Which also can support the H2 is correct (H2: The more risk detectability, the higher risk perception will be.)

Third, “The controllability of risk by government” has not showed a significant correlation with the “perceived risk of financial crisis to self” and “perceived risk of financial crisis to society”. It also means H3 cannot support by our data and the relationship between “The controllability of risk by government” and “risk perception” is not clear.

Fourth, “Distrust of the authorities” has a positive correlation with “perceived risk of financial crisis to society”. While the “Trust of the authorities” has not showed a significant correlation with “perceived risk of financial crisis to society”. Neither “distrust of the authorities” nor “Trust of the authorities” showed a significant correlation with the “perceived risk of financial crisis to self”. It also means the H4 cannot be proved based on our limited data. (H4: The more trust of the authorities, the less risk perception will be.)

Fifth: At the demographic aspect, only “Major of study” showed a significant correlation with the “perceived risk of financial crisis to society”. But not showed a significant correlation with the “perceived risk of financial crisis to self”. Besides that other demographic aspect do not showed the significant difference with two dependent variables. It also indicates that the major of study can impact the risk perception.

**Reference**

Roger E. Kasperson,' Ortwin Renn,' Paul Slovic,2 Halina S. Brown,' Jacque Emel,'Robert Goble,' Jeanne X. Kasperson,'~a~n d Samuel Ratick'. 1987.The Social Amplification of Risk A Conceptual Framework. Revised January 8, I988

Pilarski, R. (2009). Risk perception among women at risk for hereditary breast and ovarian cancer. Journal of Genetic Counseling, 18(4), 303–312.

Wibecke Brun, Cognitive components in risk perception: natural versus man made risks. Journal of Behavioral Decision Making, 1992,5: 117-132

Wiersema,M ,Bantel,K．Top management team demography and corporate strategic change[J]．Academy of Management Journal,1992,35(1):91－121

Norusis, Maruha J. 2005. SPSS 14.0 Advanced Statistical Procedures Companion. Upper Saddle River, NJ: Prentice Hall.

Hawkins, Daniel N., Paul R. Amato, and Valarie King. 2006. “Parent Adolescent Involvement: The Relative Influence of Parent Gender and Residence.” Journal of Marriage & Family 68(1):125-136

**Part 2: Summary of the class exercise**

**Chi-Square Test**

**Frequencies**

major

Category

Observed N

Expected N

Residual

1

1

35

37.5

-2.5

2

2

66

75.0

-9.0

3

3

268

262.7

5.3

4

4

10

3.8

6.2

Total

379

**Test Statistics**

major

Chi-Square

11.771^{a}

Df

3

Asymp. Sig.

.008

- 1 cells (25.0%) have expected frequencies less than 5.
- The minimum expected cell frequency is 3.8.

A chi-square test of the major field of study P=0.008 indicated that the four fields were not equally likely among the students in a statistical sense which is (p < .05) significant different from the expected distribution. Accordingly, the chance of being a sociology student has a greatest chance and the chance of being a counseling student was least. This finding may just describe the population distribution.

**2.**** Reliability**

**Reliability Statistics**

Cronbach's Alpha

Cronbach's Alpha Based on Standardized Items

N of Items

.654

.680

3

**Item-Total Statistics**

Scale Mean if Item Deleted

Scale Variance if Item Deleted

Corrected Item-Total Correlation

Squared Multiple Correlation

Cronbach's Alpha if Item Deleted

a17

13.1527

13.538

.475

.247

.572

a27

13.3257

11.730

.524

.289

.489

a37

14.1527

8.415

.460

.213

.628

The chart showed the three independent variables of a17, a27, and a37 are satisfactory internal consistency because the Cronbach's Alpha is 0.654, but from the second chart, by excluding the items a17, a27, or a37 separately the Cronbach's Alpha changed to (.572, .489, or .628) which are <.654, that indicated that the reliability cannot be improved by excluding the items in this data set.

**3.**** ****Factor analysis**

**Total Variance Explained**

Factor

Initial Eigenvalues

Extraction Sums of Squared Loadings

Total

% of Variance

Cumulative %

Total

% of Variance

Cumulative %

1

3.327

41.594

41.594

2.577

32.207

32.207

2

1.117

13.963

55.557

.897

11.218

43.424

3

.776

9.697

65.253

4

.710

8.880

74.133

5

.685

8.564

82.698

6

.560

7.001

89.699

** Factor Matrix ^{a}**

Factor

1

2

a29

.432

.080

a30

.596

.166

a31

.822

-.366

a32

.596

.100

a33

.403

.571

a34

.599

.293

a35

.577

.194

a36

.386

.519

**Rotated Factor Matrix ^{a}**

Factor

1

2

a29

.365

.245

a30

.481

.389

a31

.900

-.009

a32

.507

.328

a33

.143

.685

a34

.433

.507

a35

.452

.407

a36

.148

.630

Extraction Method: Maximum Likelihood.

Rotation Method: Varimax with Kaiser Normalization.^{a}

a. Rotation converged in 3 iterations.

Extraction Method: Maximum Likelihood.^{a}

a. 2 factors extracted. 13 iterations required.

**Factor Transformation Matrix**

Factor

1

2

1

.918

.397

2

-.397

.918

**Goodness-of-fit Test**

Chi-Square

df

Sig.

46.511

13

.524

Extraction Method: Maximum Likelihood.

Rotation Method: Varimax with Kaiser Normalization

A factor analysis about eight items about financial risk suggested two factors. The two-factor solution represented a good fit to the data, based on a chi-square test, using the maximum likelihood extraction method (p = .524).

The factor analysis using the varimax with Kaiser Normalization rotation method and using a rotation converged in 3 iterations, the first factor consisted of (a29), (a30), (a31), and (a32); the second factor undergirded (a33), (a34), (a35), and (a36). The perception about (a31) the personal responsibility for terrorist attack was the most important item to reflect the first factor. It appears that the first factor is about the prominence of risk.

**4: Regression**

*Part1:*

It shows Regression Standardized Residual is decreasing variation with age, is a sign of heteroskedasticity.

It shows Regression Standardized Residual Variation with Control roughly equal across controllability levels, it supporting the homoscedasticity.

*Part 2:*

It is repeating the previous finding of heteroskedasticity, using the residualized age, which is independent of controllability to the dependent variable of risk.

compute risk.r=age*risk.r.

graph scatterplot= age with risk.r .

It shows after weighting the residual by age, it would produce homoskedasticity.

temporary.

compute both=mean(control,new).

regression /missing mean /statistics default

/dependent risk /enter both new age female.

It showing the Regression standardized residual of the dependent variable risk has a normal distribution.

*Part 3: *

**Model Summary**

Model

R

R Square

Adjusted R Square

Std. Error of the Estimate

1

.442^{a}

.195

.187

1.39429

- Predictors: (Constant), female, new, age, both

**ANOVA ^{a}**

Model

Sum of Squares

df

Mean Square

F

Sig.

1

Regression

183.545

4

45.886

23.604

.000^{b}

Residual

756.232

389

1.944

Total

939.777

393

- Predictors: (Constant), female, new, age, both

**Coefficients ^{a}**

Model

Unstandardized Coefficients

Standardized Coefficients

t

Sig.

B

Std. Error

Beta

1

(Constant)

2.544

.489

5.202

.000

both

.361

.106

.268

3.399

.001

new

.214

.090

.188

2.378

.018

age

.003

.014

.011

.237

.812

female

.199

.152

.061

1.304

.193

The independent variable of new, age, and female has a R square is (.195) which can have a strong confident to use the three predictors to predicts the dependent variable of risk. From the chart, the independent variable of new, age, and female (p<.05) has a significant effect on the dependent variable of risk.

The independent variables “Both” and “New” have significant effects on the dependent variable “Risk”. The regression coefficient of “Both” is .361 (p=.001) which indicates that for every one unit increase in “Both” will result in .361 units in “Risk”. Whereas, the regression coefficient of “New” is .214 (p=.018) which means that for every one unit increase in “New” will result in .214 units in “Risk”. The independent variables “Age” and “Female” do not have significant effects (P > .05) on the dependent variable “Risk”. It also means that “Age” and “Female” are not significant predictors of “Risk”.

*Part 4: *

**Model Summary**

Model

R

R Square

Adjusted R Square

Std. Error of the Estimate

1

.442^{a}

.196

.188

1.39387

a. Predictors: (Constant), female, new, age, three

**ANOVA ^{b}**

Model

Sig.

R Square Change

1

Subset Tests

three

.001^{a}

.024

new, age

.000^{a}

.046

female

.172^{a}

.004

Regression

.000^{c}

Residual

Total

**Coefficients ^{a}**

Model

Unstandardized Coefficients

Standardized Coefficients

t

Sig.

B

Std. Error

Beta

1

(Constant)

5.683

.856

6.642

.000

three

.751

.219

.320

3.434

.001

new

.207

.091

.182

2.279

.023

age

-.044

.019

-.145

-2.284

.023

female

.208

.152

.064

1.370

.172

a. Dependent Variable: risk

Based on the first chart it indicates that the independent variables of “Three”, “New”, “Age”, and “Female” have R square equal to .196 which can have a strong confident to use the four predictors to predicts the dependent variable of risk.

The second chart it indicates the independent variable of “Three”, “New”, “Age” (p<.05) has a significant effect on the dependent variable of risk, while the independent variable “Female” do not have significant effects (P > .05) on the dependent variable “Risk.

However, Comprehensive front three charts, the independent variables “Three”, “New”, “Age”, have significant effects on the dependent variable “Risk”. The regression coefficient of Three”, is .751 (p=.001) which indicates that for every one unit increase in “Three” will result in .751 increase units in “Risk”. Whereas, the regression coefficient of “Age” is -.044 (p=.023) which means that for every one unit increase in “Age” will result in decrease .044 units in “Risk”.

The independent variable “Female” do not has significant effects (P > .05) on the dependent variable “Risk”. It also means that “Female” is not a significant predictor of “Risk”.

** **

**5. General Linear Model**

**Coefficients ^{a}**

Model

Unstandardized Coefficients

Standardized Coefficients

t

Sig.

B

Std. Error

Beta

1

(Constant)

2.811

.679

4.140

.000

three

.070

.220

.030

.319

.750

new

.376

.082

.331

4.609

.000

control

.164

.073

.152

2.241

.026

female

.202

.153

.062

1.320

.188

**Coefficients ^{a}**

Model

Unstandardized Coefficients

Standardized Coefficients

t

Sig.

B

Std. Error

Beta

1

(Constant)

2.479

.493

5.027

.000

control

.181

.053

.167

3.408

.001

new

.400

.056

.351

7.132

.000

age

.004

.014

.012

.255

.799

female

.206

.153

.063

1.348

.179

inter

.064

.063

.047

1.021

.308

temporary.

compute inter=zcontrol*znew.

regression /missing mean /statistics default tol

/dependent risk /enter control new age female inter

The general linear model that added the interaction between the control and new as a predictor turned out that this interaction did not generally make a significant difference (p>0.05) in the dual outcomes of perceived risks to self and to society. Nevertheless, two particular significant interactions were that involved, the controllability of terrorist attack by government, and the perceived novelty that indicated significantly perceived risk to society.

**6. Two-stage Least Squares Analysis**

2sls risk with control new /control with risk /new with yr control /instrument age female yr.

**ANOVA**

Sum of Squares

df

Mean Square

F

Equation 1

Regression

16.286

2

8.143

2.952

Residual

976.635

354

2.759

Total

992.921

356

Equation 2

Regression

16.593

1

16.593

5.329

Residual

1105.295

355

3.114

Total

1121.888

356

Equation 3

Regression

2.534

2

1.267

.702

Residual

638.443

354

1.804

Total

640.977

356

**ANOVA**

Sig.

Equation 1

Regression

.054

Residual

Total

Equation 2

Regression

.022

Residual

Total

Equation 3

Regression

.496

Residual

Total

**Coefficients**

Unstandardized Coefficients

Beta

t

B

Std. Error

Equation 1

(Constant)

-.922

3.403

-.271

new

.772

.421

.711

1.835

control

.468

.673

.411

.695

Equation 2

(Constant)

-.269

2.529

-.106

risk

.991

.429

.995

2.309

Equation 3

(Constant)

5.188

2.276

2.279

control

.062

.387

.065

.159

yr

-.081

.096

-.055

-.846

**Coefficients**

Sig.

Equation 1

(Constant)

.787

new

.007

control

.487

Equation 2

(Constant)

.915

risk

.022

Equation 3

(Constant)

.023

control

.874

yr

.398

In a nonrecursive, reciprocal-effect model, perceived novelty and perceived risks were causes and outcomes reciprocally, with the year of study, age, and female used as instruments to enable the two-stage least squares estimation. Regression analysis with two-stage least squares estimation showed both the reciprocal effects were significant. Accordingly, the effect of perceived novelty on perceived risk was very strong (β = .771), and the effect of perceived risk on perceived novelty was very strong too (β = .995). However, the Equation 3 is not significant.

**7: General Linear Model**

glm risk.soc risk by level major with age female

/print parameter opower etasq homogeneity

/plot=profile(level major)

/emmean table(level) compare adj(bonferroni)

/emmean table(major) compare adj(bonferroni)

/design level major age female.

**Multivariate Tests ^{a}**

Effect

Value

F

Hypothesis df

Error df

Sig.

Intercept

Pillai's Trace

.149

30.874^{b}

2.000

352.000

.000

Wilks' Lambda

.851

30.874^{b}

2.000

352.000

.000

Hotelling's Trace

.175

30.874^{b}

2.000

352.000

.000

Roy's Largest Root

.175

30.874^{b}

2.000

352.000

.000

level

Pillai's Trace

.006

.508

4.000

706.000

.730

Wilks' Lambda

.994

.508^{b}

4.000

704.000

.730

Hotelling's Trace

.006

.507

4.000

702.000

.731

Roy's Largest Root

.006

1.005^{c}

2.000

353.000

.367

major

Pillai's Trace

.097

4.494

8.000

706.000

.000

Wilks' Lambda

.905

4.519^{b}

8.000

704.000

.000

Hotelling's Trace

.104

4.543

8.000

702.000

.000

Roy's Largest Root

.082

7.213^{c}

4.000

353.000

.000

age

Pillai's Trace

.007

1.240^{b}

2.000

352.000

.291

Wilks' Lambda

.993

1.240^{b}

2.000

352.000

.291

Hotelling's Trace

.007

1.240^{b}

2.000

352.000

.291

Roy's Largest Root

.007

1.240^{b}

2.000

352.000

.291

female

Pillai's Trace

.015

2.739^{b}

2.000

352.000

.066

Wilks' Lambda

.985

2.739^{b}

2.000

352.000

.066

Hotelling's Trace

.016

2.739^{b}

2.000

352.000

.066

Roy's Largest Root

.016

2.739^{b}

2.000

352.000

.066

**Multivariate Tests ^{a}**

Effect

Partial Eta Squared

Noncent. Parameter

Observed Power^{d}

Intercept

Pillai's Trace

.149

61.747

1.000

Wilks' Lambda

.149

61.747

1.000

Hotelling's Trace

.149

61.747

1.000

Roy's Largest Root

.149

61.747

1.000

level

Pillai's Trace

.003

2.033

.173

Wilks' Lambda

.003

2.030

.173

Hotelling's Trace

.003

2.027

.172

Roy's Largest Root

.006

2.010

.225

major

Pillai's Trace

.048

35.950

.997

Wilks' Lambda

.049

36.148

.997

Hotelling's Trace

.049

36.345

.997

Roy's Largest Root

.076

28.852

.996

age

Pilla i's Trace

.007

2.480

.270

Wilks' Lambda

.007

2.480

.270

Hotelling's Trace

.007

2.480

.270

Roy's Largest Root

.007

2.480

.270

female

Pillai's Trace

.015

5.477

.539

Wilks' Lambda

.015

5.477

.539

Hotelling's Trace

.015

5.477

.539

Roy's Largest Root

.015

5.477

.539

**Tests of Between-Subjects Effects**

Source

Dependent Variable

Type III Sum of Squares

df

Mean Square

F

Corrected Model

risk.soc

84.867^{a}

8

10.608

4.773

risk

51.388^{b}

8

6.424

2.813

Intercept

risk.soc

128.357

1

128.357

57.755

risk

97.992

1

97.992

42.907

level

risk.soc

3.538

2

1.769

.796

risk

.330

2

.165

.072

major

risk.soc

63.988

4

15.997

7.198

risk

36.091

4

9.023

3.951

age

risk.soc

4.992

1

4.992

2.246

risk

.879

1

.879

.385

female

risk.soc

3.046

1

3.046

1.370

risk

12.059

1

12.059

5.280

Error

risk.soc

784.523

353

2.222

risk

806.187

353

2.284

Total

risk.soc

17629.583

362

risk

13288.667

362

Corrected Total

risk.soc

869.390

361

risk

857.575

361

**Tests of Between-Subjects Effects**

Source

Dependent Variable

Sig.

Partial Eta Squared

Noncent. Parameter

Observed Power^{c}

Corrected Model

risk.soc

.000

.098

38.186

.998

risk

.005

.060

22.501

.942

Intercept

risk.soc

.000

.141

57.755

1.000

risk

.000

.108

42.907

1.000

level

risk.soc

.452

.004

1.592

.186

risk

.930

.000

.145

.061

major

risk.soc

.000

.075

28.792

.996

risk

.004

.043

15.803

.904

age

risk.soc

.135

.006

2.246

.321

risk

.535

.001

.385

.095

female

risk.soc

.243

.004

1.370

.215

risk

.022

.015

5.280

.630

Error

risk.soc

risk

Total

risk.soc

risk

Corrected Total

risk.soc

risk

In multivariate analysis of variance, perceived risk to oneself and perceived risk to society were outcomes predicted by the level of study, major, age, and female. Multivariate tests showed that only the major made a significance difference in the two outcomes, with an observed statistical power of .997. This power indicated that the risk of the error of wrongly acceptance of the null hypothesis of no difference was close to zero (Type I error = .003). At the same time, the effect size in terms of partial eta (η) was .048. In addition, univariate tests revealed that the major made a significant difference in both perceived risks to society and to self. The observed statistical power was .996 and .904 respectively.

**Parameter Estimates**

Dependent Variable

Parameter

B

Std. Error

t

Sig.

95% Confidence Interval

Lower Bound

risk.soc

Intercept

7.543

1.249

6.042

.000

5.088

[level=1.00]

.165

.699

.236

.814

-1.210

[level=2.00]

.384

.687

.559

.577

-.967

[level=3.00]

0^{a}

.

.

.

.

[major=1]

-2.855

1.298

-2.199

.028

-5.407

[major=2]

-2.741

1.267

-2.164

.031

-5.232

[major=3]

-1.750

1.258

-1.392

.165

-4.224

[major=4]

-1.994

1.403

-1.421

.156

-4.754

[major=5]

0^{a}

.

.

.

.

age

.037

.025

1.499

.135

-.012

female

.200

.171

1.171

.243

-.136

risk

Intercept

4.880

1.266

3.856

.000

2.391

[level=1.00]

.147

.708

.207

.836

-1.246

[level=2.00]

.198

.696

.285

.776

-1.171

[level=3.00]

0^{a}

.

.

.

.

[major=1]

.091

1.316

.069

.945

-2.497

[major=2]

-.520

1.284

-.405

.686

-3.046

[major=3]

.345

1.275

.271

.787

-2.162

[major=4]

.699

1.423

.491

.624

-2.100

[major=5]

0^{a}

.

.

.

.

age

.015

.025

.621

.535

-.034

female

.399

.174

2.298

.022

.057

Compared with the major of criminology [major=5], the social work and psychology majors showed marginally significantly lower perceived risk. Moreover, the multivariate test indicated that the effects of all predictors on all outcomes were not significant as a whole, for example, with a psychology major Type II error of 5.407 and Type I error of .302.

**Pairwise Comparisons**

Dependent Variable

(I) major

(J) major

Mean Difference (I-J)

Std. Error

Sig.^{b}

95% Confidence Interval for Difference^{b}

Lower Bound

Upper Bound

risk.soc

1

2

-.114

.414

1.000

-1.283

1.055

3

-1.104^{*}

.381

.040

-2.181

-.028

4

-.860

.592

1.000

-2.532

.811

5

-2.855

1.298

.285

-6.521

.812

2

1

.114

.414

1.000

-1.055

1.283

3

-.990^{*}

.223

.000

-1.619

-.362

4

-.746

.627

1.000

-2.518

1.025

5

-2.741

1.267

.312

-6.319

.838

3

1

1.104^{*}

.381

.040

.028

2.181

2

.990^{*}

.223

.000

.362

1.619

4

.244

.607

1.000

-1.470

1.958

5

-1.750

1.258

1.000

-5.303

1.802

4

1

.860

.592

1.000

-.811

2.532

2

.746

.627

1.000

-1.025

2.518

3

-.244

.607

1.000

-1.958

1.470

5

-1.994

1.403

1.000

-5.959

1.970

5

1

2.855

1.298

.285

-.812

6.521

2

2.741

1.267

.312

-.838

6.319

3

1.750

1.258

1.000

-1.802

5.303

4

1.994

1.403

1.000

-1.970

5.959

risk

1

2

.611

.420

1.000

-.574

1.796

3

-.255

.386

1.000

-1.346

.836

4

-.608

.600

1.000

-2.302

1.087

5

.091

1.316

1.000

-3.626

3.807

2

1

-.611

.420

1.000

-1.796

.574

3

-.866^{*}

.226

.001

-1.503

-.228

4

-1.219

.636

.560

-3.015

.577

5

-.520

1.284

1.000

-4.148

3.107

3

1

.255

.386

1.000

-.836

1.346

2

.866^{*}

.226

.001

.228

1.503

4

-.353

.615

1.000

-2.090

1.384

5

.345

1.275

1.000

-3.256

3.947

4

1

.608

.600

1.000

-1.087

2.302

2

1.219

.636

.560

-.577

3.015

3

.353

.615

1.000

-1.384

2.090

5

.699

1.423

1.000

-3.320

4.717

5

1

-.091

1.316

1.000

-3.807

3.626

2

.520

1.284

1.000

-3.107

4.148

3

-.345

1.275

1.000

-3.947

3.256

4

-.699

1.423

1.000

-4.717

3.320

Based on estimated marginal means

*. The mean difference is significant at the .050 level.

b. Adjustment for multiple comparisons: Bonferroni.

Pairwise comparison of adjusted means with Bonferroni adjustment showed that the perceived risk to society was significantly lower in the sociology student than in the social work student (mean difference = -.990*, p = .000). Similarly, the perceived risk to society was significantly lower in the sociology student than in the psychology student (mean difference = -1.104*, p = .040). Besides that, Pairwise comparison of adjusted means with Bonferroni adjustment showed that the perceived risk to oneself was significantly lower in the sociology student than in the social work student (mean difference = -.866*, p = .001).

**Part 3: Class participation report**

In semester B, the course of Applied Social Statistical Analysis, taught me how to use the statistical reasoning to understand / analysis the social research. Besides that I also controlled the skills of using SPSS to analysis the basic statistical methods and explain the findings. This course is the most meaningful course to me, because I am a nurse in hospital and in my future work, I should use different kinds of statistic method to analysis the medical data on patients, medical staff, and medicine experiment data. I attend every class of Applied Social Statistical Analysis on time, not only because of its meaningful to be but also this course is a little hard for me to understand. I did not have any statistic foundation expect the last semester leaning of statistic learning, but thanks to Dr Jacky, he gave us lots of recommend on the course related readings and Chinese book on SPSS analysis, I had a clear idea of this course and the know how to choose the appropriate statistical methods to do the statistical analysis on sociological practices.

Though this course, at the beginning of this course I learned the data grooming method to simplifying statistical data, data cleaning method to remove the illogical and inconsistent values, data transformation method to Skew correction, recoding data, imputation missing value and weighting in order to make the data more reliability and validity. On the second phase, I mastered the knowledge of statistic testing, such as, parametric testing and nonparametric testing. Fellow that, the date reduction method followed, it contains factor analysis which contains varimax rotated Variance, exploring underlying factors and Construct validation and reliability assessment, what’s more, all the factors based on the theoretical grounds, interpretability, identified y by alternative sets of indicators, high explaining power and the distinct loadings. The clustering method to do data reduction is also important, clustering method can classifying /group variables in order to simplifying data besides that it also can identify new variables that can represent the configuration/combination variables. Based on the data handling method, I learned the data analysis method that is linear modeling. Linear modeling contains the linear regression, multivariate analysis, Path analysis through linear regression analysis and MANOVA. Regression analysis is a statistical method to determine interdependent relationships between two or more quantitative variables. It has a very wide range, according to the number of independent variables the regression analysis can be involved into regression and multiple regression analysis. Besides that according to the number of dependent variables, it can be divided into a regression analysis and multiple regression analysis; according to the relationship between the independent variables and the dependent variable, it can be divided into linear regression and nonlinear regression analysis. This section is the most important and difficult part in this course, while is a comprehensive statistical data analysis method assemble. Follow on the nonlinear modeling was taught and it is a very useful and productive statistical method. When I start doing the SPSS individual, some excited emotion came to my mind, the learning efforts of Applied Social Statistical Analysis and results obtained are obvious. I can use the data to analysis my variables and my analysis the findings.

Review

Correspondence between reviewed literature and hypotheses needs to be clearer.

Analysis

Elaborate interpretation must be present. Sociological insight is preferable.

Discussion

It needs to be lengthier, support by literature review.

Overall

Interpretation of findings is utterly necessary.

Exercise

It can point out learning gained from the exercise. Literature review is tremendously helpful.

Performance

Learning needs to be deeper, concrete, and pertinent to sociology. It can be more focused.