中原大學九十二學年度博士班入學招生考試─管理研究所，應用統計

中原大學九十二學年度博士班入學招生考試

92年6月11日　10:30~12:00　管理研究所	誠實是我們珍視的美德，我們喜愛「拒絕作弊，堅守正直」的你！
科目：應用統計

PART 1 (50%)
Notice: You must answer the question number 1 and choose either one of the question number 2 or 3.

1.	A new manufacturer of children's toys was planning to retail the firm's product through mail order. The firm's management was considering the use of direct-mail advertisements to stimulate sales of their products. Prior to committing themselves to advertising through direct mail, management conducted an experiment. A random sample of 1,000 families with at least one under 10-year-old child selected from Taipei city. The sample was divided into two groups, with each subject being randomly assigned to one of the two groups. Direct-mail advertisements were sent twice over a period of one month to respondents of one of the groups. Two weeks later, respondents of both groups were mailed the company's catalog of toys. Sales to each group were monitored. Comment on this experimental design and discuss the internal and external validity for the design. (25%)

2.	What is "data mining"? How does it differ from traditional data analysis? (25%)

3.	A cell-phone company wants to investigate the needs of its customers. Propose a scheme for the sample and discuss the benefits of your scheme. (25%)

PART 2 (50%)

Multivariate analytical techniques are being widely applied in industry, government, and university-related research centers. These analysis methods will predominate in the future and will result in drastic changes in the manner in which research workers think about problems and how they design their research. In fact, the purpose of scientific research include simplifying structure or reducing dimensions, classification or clustering, examining dependence relationship among variables, prediction, and model building or hypotheses testing. The following multivariate techniques will meet these purposes, please fill the blank and match them. (25%)

Analysis of Variance (ANOVA)

Multivariate Analysis of Variance (MANOVA)

Multiple Regression Analysis

Automatic Interaction Detector (AID)

Multiple Discriminant Analysis

Conjoint Analysis

Canonical Correlation Analysis

Factor Analysis

Cluster Analysis

Multidimensional Scaling

Structural Equation Modeling

Notice: Each method may meet several purposes, i.e. you can repeat fill the same method in different cell.

the purpose of scientific research	multivariate techniques（multiple choices）
simplifying structure or reduce dimensions
classification or clustering
examining dependence relationship among variables
prediction
model building or hypotheses testing

If we want to understand how customers' perceptions of TSMC (company name) affect their behavior and attitudes, 136 customers were asked to provide evaluation on 13 attributes across the entire spectrum of ways in which they might interact with and evaluate TSMC. The items and evaluation dimensions are shown as follow.

Evaluative Dimension	Firm Attribute
Firm and product factors	Product quality Invoice accuracy Technical support Introduction of new products Reliable delivery Customer service
Price-based factors	Product value Low-price supplier Negotiation position (policies)
Buying relationship	Mutuality of interests Integrity/honesty Flexibility Problem resolution

When the dimensions of evaluation that will act as the independent variables in this analysis have been established, the next step is to determine their relationships to purchase decisions and satisfaction. Because satisfaction is based on a customer's evaluation of past experience with TSMC, the appropriate causal relationship is for current satisfaction levels to be predicted by level of purchases from TSMC. Although a causal relationship can be posed depicting purchases arising from satisfaction, this method would require a greater degree of control over longitudinal aspects of data collection necessary to establish the temporal order. However, we know that purchase level reflects past purchase behavior, so that it can be a valid predictor of current satisfaction.

Please answer the following question:

(1)

Please draw your research model. (5%)

(2)

Based on your judgment, what kind of multivariate analysis method may solve this research question? Please make a detail description of your answer (include the name of this method, the reason of choosing this method, the assumption of this method, and analytical steps). (15%)

(3)

Based on your analytical method, what are the criteria by which the researchers should decide whether a model has achieved an acceptable level of fit? (5%)

--- END ---