(Tiểu luận) survey and multivariate analysisind ividual portfolio

The manager wishes to know how trust varies due to different levelof age, purchase frequency and income.- Analyze Descriptive Statistics Crosstabs Move TRU to Rows Move Age to Columns O

Data “Repurchase intention”

1.1 The manager wishes to know which retailing website consumers make purchase most frequently and least frequently

- To answer, we use Frequencies in Descriptive Statistics, then select “Web” in the other column.

+ By Focusing on the Frequency column in the Frequency Table, we could see that ADAYROI and CHOTOT account for only 1% (as 1 would be in percentage)

We could conclude that LAZADA has the highest purchase frequency with 44%, while ADAYROI and CHOTOT have the lowest with only 1%.

1.2 In order to employ STP (segmenting – targeting – positioning), the manager wishes to know characteristics of the e-shopper in details(level of trust, gender, age, purchasing frequency, income and education level).

- With these numbers of Frequency, and the descriptive of:

Gender of participants: 1=Male, 2male

Age of participants: 1 = under 25; 2 = 25 – 30; 3 = above 30

Purchase frequency of participants: 1 = once per month; 2 = twice per month; 3 = three times per month; 4 = four times or above per month Monthly income of participants: 1 = under VND 5 million; 2 = VND 5 million - VND 10 million; 3 = above VND 10 million

Education level of participants : 1 = high school; 2 = college; 3 university or higher

Trust in online shopping: 1 = low trust, 2 = medium trust, 3 = high trust

=> Consumers are predominantly female, mostly under 25, have medium-to-high trust in e-commerce, monthly purchase frequency, university level education or higher, and 5 to 10 million VND income.

1.3 Following the same procedures as 1.1, we will have these tables in order to employ STP (segmenting – targeting – positioning).

- Variables must be in “Likert scale”.

- Model is Alpha click OK (Same goes for IQ and RP).

- The Cronbach Alpha for WOM is 932, IQ is 870 and RP is 820 respectively. Due to all cases being valid, it’s safe to assume that the internal consistency of Word of Mouth is excellent, while Repurchase Intention and Information Quality are good.

1.4 Before employing data analysis, the manager wishes to check the reliability of the scales measuring WOM, IQ and RP based on Cronbach alpha.

- Select WOM, IQ and RP into the right box after entering Descriptive Statistics inFrequencies in Analysis.

- Select Mean, Median, Mode, Range, Standard Deviation, Skewness, and Kurtosis from the “Statistics ” Click OK.

1.5 The manager wishes to identify basic statistical figures of WOM, IQ and RP including mean, median, mode, range, standard deviation, skewness and kurtosis.

- We can use one-sample t-test will be implemented to answer this question As a result, we obtain

1.6 The manager wishes to know how trust varies due to different level of age, purchase frequency and income.

Move TRU to Row(s) Move Age to Column(s) OK

Most under-25-year-old participants have high trust in e-shopping while the trust of the 25-30 tends to reach just medium.

The more the purchase frequency increases, the lower the trust is in online shopping.

A large number of people who have income within the range from 5-10M VND buy things online and have high trust in online shopping.

1.7 The manager thinks that WOM may influence heavily on consumer in online purchase, the manager wishes to test the hypothesis stating that mean of WOM in the population exceeds 5.

1.8 The managers think that male and female consumer may not similar in their repurchase intention Thus, the manager wishes to check whether repurchase intention is statistically different or not across gender.

- We will use the Independent samples t-test

- Analyze Compare means Select Independent samples t-test

Data “Consumer satisfaction”

2.1 Does satisfaction vary due to consumers’ gender?

+ Move SAT1, SAT2, SAT3, SAT4 to the Variables box

+ In the “Statistics” Check the “Scale if item deleted” this will provide additional information if an item is removed click OK

Look for Cronbach's alpha coefficient in the output, it is 0.796 which is higher than 0.7.

+ Transform Compute variable Set like the picture Ok

+ Analyze Compare Means Choose One-Way ANOVA

+ In the "One-Way ANOVA" dialog box, select the “satisfaction” variable as the dependent variable and the “gender” variable as the factor.

+ Click on the Post Hoc Select LSD OK to run the analysis

+ Result: sig 0.036 < 0.05 (the significant level)

There are significant differences in the satisfaction of customers considering their gender

2.2 Does satisfaction vary due to consumers’ age?

+ In the "One-Way ANOVA" dialog box, select the “satisfaction” variable as the dependent variable and the “age” variable as the factor.

+ Result: for Ages 2.0 and 3.0, sig 0.03 < 0.05 there are significant differences between the satisfaction of customers considering their age.

2.3 Does satisfaction vary due to consumers’ education?

+ In the "One-Way ANOVA" dialog box, select the “satisfaction” variable as the dependent variable and the “Edu” variable as the factor

Data “Consumer shopping well-being”

3.1 Does shopping well-being vary due to consumer income under the potential effect of utilitarian value?

+ Computing the first variable: SWB (Shopping well-being)

+ Name the “Target Variable” (Ex: SWBmean) Compute the 4 items of SWB in the “Numeric Expression” box.

+ Choose functions and Special Variables -> Mean -> OK

+ The same goes for UV named “UVmean”

+ Analyze -> General Linear Model -> Univariate

+ Compute the variable in the right category -> click OK

The significance of UVmean is 0 (0.05)

The unitarian value affects SWB.

There is not enough evidence to prove the relationship between income and SWB.

3.2 Does shopping well-being vary due to consumer income under the potential effect of hedonic value?

+ Computing the next variable: HV (Hedonic Value)

+ Name the “Target Variable”: HVmean Compute the 4 items of HV in the

+ Choose functions and Special Variables -> Mean -> OK

+ Analyze -> General Linear Model -> Hedonic

+ Compute the variable in the right category -> click OK

The significance of HV is 0 ( UVmean and HVmean have the highest individual effects

=> Income may not have a significant impact.

3.4 Does shopping well-being vary due to consumer income and gender under the potential effect of utilitarian value and hedonic value?

(steps are exactly like question 3.1 & 3.2)

Gender and Income & Gender interaction (p-values > 0.05) are non- significant

=> UVmean and HVmean have the highest individual effects

=> Income may have a weak impact

=> Gender and Income & Gender interaction don’t significantly impact SWBmean.

4.1 Are there any relationships among EMP, SAL, TRA, EXP, MAS, SUP and HWL? Which of them are significant?

+ Move EMP to the “Dependant” box; SAL, TRA, EXP, MAS, SUP, and HWL to the “Independent” box.

+ Click on Statistics Check the “Collinearity diagnostics” box Continue

+ Click on Plots → put residual values (*ZRESID) in the Y box, and predicted values (*ZPRED) in the X box select “normal probability” Continue OK

+ In “Analysis” -> Select “Correlate” -> Select “Bivariate”.

+ Move all variables including EMP, SAL, TRA, EXP, MAS, SUP, and HWL to the

+ Select “Pearson”, “Two-tailed” and “Flag significant correlations” so that SPSS can mark the statistically significant correlations based on your chosen alpha level (usually 0.05) OK

- EMP: With SAL, TRA, EXP, MAS, and SUP, EMP both have a significant 0.05, so the HWL does not have a strong effect on EMP.

- SAL: For TRA, EXP, MAS, and SUP, SAL both have significance at 0 < 0.05, and r are 0.679, 0.661, 0.646, 0.179, which means that there are strongly positive effects between SAL and these factors However, the significance between SAL and HWL is 0.184, which indicates that the HWL and SAL don't have a significant impact on each other

- TRA: The significance between TRA and EXP, MAS, and SUP are both < 0.05, and r is 0.696, 0.707, 0.171, so it means that TRA and these factors have strongly positive effects on each other Meanwhile, the significance between TRA and HWL is 0.258 > 0.05, so there is no relation between TRA and HWL.

- EXP: With MAS and SUP, EXP both has significance at 0, and r are 0.853 and 0.197, so it means that there are significantly positive effects between EXP and these factors However, the significance for EXP and HWL is 0.065, which means that there is no strong relation between EXP and HWL.

- MAS: The significance between MAS and SUP, HWL are both < 0.05, and r are 0.194 and 0.098, so it indicates that MAS and SUP, HWL have a strongly positive effect on each other

- SUP: for HWL, SUP has significance at 0 < 0.05 and r at 0.76, so there is a significant positive between HWL and SUP

4.2 The EMP is potentially influenced by both SAL and MAS If this is the case, to which extent the variation of EMP is uniquely due to SAL? And, uniquely due to MAS?

+ Move the variable EMP into the box “Dependent”, MAS, and SAL into the box labeled Independent(s).

+ In the "Statistics" dialog box, check the box for "R squared change" under

"Model fit" to obtain the change in the R-square value

To know exactly the extent to which variation of EMP is uniquely due to SAL and MAS, let’s calculate it using the value of part (or semipartial) correlation

For SAL, square its part correlation 0.25 and then multiply with 100, we have the result of 6.25 This means that 6.25 of the variance in EMP is accounted for uniquely by SAL

We will do the same for MAS And then we have the result of 12.18, which means that 12.18% of the variance in EMP is uniquely due to MAS.

4.3 Whether increasing salary is a good solution for improving employees’ performance? To which extent employees’ salary predicts their performance?

+ Specify the variables: In the Linear Regression dialog box, move the SAL (salary) into the "Independent" box, and the EMP (performance) into the

+ Click "OK" to run the analysis.

Looking at the Unstandardized Coefficient column: we can see that EMP = 2.027 + 0.586SAL

=> If we increase the salary by 1 dollar, we will see an increase of 58.6% in employee performance Considerably good to increase the salary.

Approximately 38.7% of the variance in the dependent variable (performance ratings) can be explained by the independent variable(s) (salary) included in the regression model

Employees' salary predicts their performance ratings to a moderate extent and it is good to some extent to increase the salary and salary can predict 43.7% of the performance.

Performance-based bonuses, salary increases, offer competitive salaries to attract high-performing individuals to recruit.

4.4 Whether providing more supports from managers is a good solution for improving employees’ performance? Identify the extent to which employees’ performance is explained by managers’ support?

+ Move MAS (manager support) into the Independent box and EMP (performance) into the Dependant(s) box OK

- Approximately only 0.9% of the variance in employees' performance can be explained by managers' support in the regression model This suggests that managers' support has a very limited explanatory power in predicting employees' performance.

=> Only a 7.7% increase in performance if we raise the support.

=> Not a good solution as other factors can provide a better boost in performance.

4.5 Do salary, training, experience, managers’ support, management style and heavy workload predict employees’ performance? To which extent employees’ performance is explained by all these factors? Among the influencing factors, which is the most and the least important factor in determining employees’ performance?

In this case, the R-value of 0.745 indicates a good level of prediction This means that the independent variables can predict the outcome variable (EMP), and predict it pretty well

R square (coefficient of determination) is 0.555, which means that all of our independent variables can explain 55.5% of the variance in our dependent variable EMP.

To know exactly the importance level of each predictor variable, we will look at its standardized coefficients (beta value) The predictor with the largest absolute beta value is considered the most important factor, and vice versa So, as shown in the table, the most important predictor is TRA with 0.269, and the least important factor is HWL with 0.021

In conclusion, salary, training, experience, managers’ support, management style, and heavy workload do predict employees’ performance, and they can explain 55.5% of the variance in EMP Also, the most and least important factors in determining EMP are TRA and HWL respectively.

5.1 Which factors significantly explain for the differences between online shopping adopting group and online shopping refusing group?

Cronbach Alpha is 0.684 < 0.7 We can see that SHE is 0.813 in the Cronbach Alpha if item deleted

Income, education, perceived value, and computer skills are the factors that significantly explain the differences between consumers who adopt and consumers who refuse to make online purchases ( they may both belong to the Refusing group.

6.1 Among PRI, PEU, DES, BRA, Age and Gender, which factors help to predict whether target consumers will buy the new product?

- Decision rule: Sig < 0.05 => Help to predict

- Sig > 0.05 => Not help to predict

Only PEU (Perceived usefulness of the new product) and BRA (Brand reputation of the company) help to predict whether target consumers will buy the new product.

6.2 How strongly the significant factors predict purchase intention of target consumers?

- 84 don’t have the intention to purchase

- 44 have the intention to purchase

Conclusion: The significant factors DON’T strongly predict purchase intention of target consumers

6.3 Assume that there are two potential consumers having following characteristics; identify the probability that they will purchase the new product.

Consumer PEU PRI DES BRA Gender Age

- Build the Formula to identify Probability from Logit Model

Questions: How many underlying factors that the managers should pay attention to? Which are they?

8.1 Classify the students into 3 groups, label these groups and identify the number of students in each group.

- We must classify students into 3 groups have a certain number of groups to classify => Use K-Means Cluster

+ On the toolbar, choose “Analyze” Descriptive Statistics Descriptive + Inputting all the variables into the box

+ Tick “Save standardized values as variables” OK

+ Select “Analyze” from the toolbar Classify K-Means Clustering + Input the standardized variables (ZUR, ZFA, ZPC, ZCF, ZFACI, ZCD, ZPI) into the “Variables” box.

+ Type “3” in the “Number of clusters” box.

+ Click the “Options” button Tick “Initial cluster centers” and “ANOVA table” Click “Continue” OK

From the ANOVA table: The sig of ZUR, ZAR, ZPC, ZFACI, Z CD, ZPI is lower than 0.05, and only the sig of ZCF is higher than 0.5 -> CF isn’t a significant variable

Number of students in each cluster: Cluster 1 has 3 students, Cluster 2 has 27 students, and Cluster 3 has 20 students.

Cluster 1 can be labeled as “Low Standard” and has 3 students.

Cluster 2 can be labeled as “High Standard” and has 27 students.

Cluster 3 can be labeled as “Medium Standard” and has 20 students.

8.2 Identify the optimal number of groups that best separate the students

+ Input UR, FA, PC, CF, FACI, CD, PI into the “Continuous Variables” box

- Analyze Hierarchical cluster Plots Method Save

=> The optimal number of clusters is 2

Questions: Analyze the data and interpret the results to describe the competitive landscape of the universities.

Questions: Analyze the data and interpret the results to describe the competitive landscape of the universities Further, compare UEH to IU in terms of university reputation and career development.

11 Data “ProductDesign_plan” and “ProductDesign_prefs”

Questions: Which is the most potential product? The manager wishes to diversify the new product to better meet market demand Help him to identify the second and third potential options.

Data “Online shopping”