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
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:
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.
“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
.369a
.136
.114
1.43356
ANOVAb
Model
Sum of Squares
df
Mean Square
F
Sig.
1
Regression
110.760
9
12.307
5.988
.000a
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
Coefficientsa
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
.348a
.121
.098
1.79451
ANOVAb
Model
Sum of Squares
df
Mean Square
F
Sig.
1
Regression
151.335
9
16.815
5.222
.000a
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
Coefficientsa
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
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
11.771a
Df
3
Asymp. Sig.
.008
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 Matrixa
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 Matrixa
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
.442a
.195
.187
1.39429
ANOVAa
Model
Sum of Squares
df
Mean Square
F
Sig.
1
Regression
183.545
4
45.886
23.604
.000b
Residual
756.232
389
1.944
Total
939.777
393
Coefficientsa
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
.442a
.196
.188
1.39387
a. Predictors: (Constant), female, new, age, three
ANOVAb
Model
Sig.
R Square Change
1
Subset Tests
three
.001a
.024
new, age
.000a
.046
female
.172a
.004
Regression
.000c
Residual
Total
Coefficientsa
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
Coefficientsa
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
Coefficientsa
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 Testsa
Effect
Value
F
Hypothesis df
Error df
Sig.
Intercept
Pillai's Trace
.149
30.874b
2.000
352.000
.000
Wilks' Lambda
.851
30.874b
2.000
352.000
.000
Hotelling's Trace
.175
30.874b
2.000
352.000
.000
Roy's Largest Root
.175
30.874b
2.000
352.000
.000
level
Pillai's Trace
.006
.508
4.000
706.000
.730
Wilks' Lambda
.994
.508b
4.000
704.000
.730
Hotelling's Trace
.006
.507
4.000
702.000
.731
Roy's Largest Root
.006
1.005c
2.000
353.000
.367
major
Pillai's Trace
.097
4.494
8.000
706.000
.000
Wilks' Lambda
.905
4.519b
8.000
704.000
.000
Hotelling's Trace
.104
4.543
8.000
702.000
.000
Roy's Largest Root
.082
7.213c
4.000
353.000
.000
age
Pillai's Trace
.007
1.240b
2.000
352.000
.291
Wilks' Lambda
.993
1.240b
2.000
352.000
.291
Hotelling's Trace
.007
1.240b
2.000
352.000
.291
Roy's Largest Root
.007
1.240b
2.000
352.000
.291
female
Pillai's Trace
.015
2.739b
2.000
352.000
.066
Wilks' Lambda
.985
2.739b
2.000
352.000
.066
Hotelling's Trace
.016
2.739b
2.000
352.000
.066
Roy's Largest Root
.016
2.739b
2.000
352.000
.066
Multivariate Testsa
Effect
Partial Eta Squared
Noncent. Parameter
Observed Powerd
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
84.867a
8
10.608
4.773
risk
51.388b
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 Powerc
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]
0a
.
.
.
.
[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]
0a
.
.
.
.
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]
0a
.
.
.
.
[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]
0a
.
.
.
.
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 Differenceb
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.