PC Mall is a publicly traded company with annual revenues over $1 billion. The company sells a
variety of merchandise, focusing primarily on computers, software, electronics, and related items.
More than 100,000 different products from companies such as IBM, Microsoft, Compaq, Apple,
Hewlett-Packard, among others, are included in PC Mall’s portfolio and are sold through catalogs
and over the Internet. Its various catalogs include the PC Mall, MacMall, ClubMac, PC Mall Gov.,
and eCost.com brands. Online, the company operates websites for each of its catalog brands
(PcMall.com, MacMall.com, ClubMac.com, PcMallGov.com, and eCost.com) as well as OnSale.com,
a discount retailer and auction site. The majority of PC Mall revenue comes from the MacMall and
PC Mall brands.
Meeting customer needs in the best possible way is one of the key values that PC Mall’s
management believes in; it is one of the values that guide the firm’s strategic decisions. For example,
in order to achieve more rapid delivery of orders to its customers, PC Mall’s distribution center was
strategically located near FedEx’s main hub in Memphis, TN. In spite of all the efforts and activities
targeted at constant improvement of the customers’ satisfaction with the company, there were no
formal systems or processes that had been implemented for ongoing tracking of customer
satisfaction levels. The end goal for PC Mall’s management is to answer the decision problem of
“How can PC Mall improve its customer experiences and satisfaction?”
The lack of knowledge about customer satisfaction, and systems to measure this, were a major topic
of discussion at a recent cross-departmental meeting. Customer satisfaction was believed to affect
customer loyalty which was, in turn, directly related to the company’s long-term profitability. The
decision was made to engage a local research firm to help design such a process, in order to enable
PC Mall’s management to assess, on a quarterly basis, the state of the customers’ satisfaction and
loyalty with the company.
Based on this process, PC Mall’s management identify the following research problems:
1. How do current customers shop and what factors influence them to shop at PC Mall
websites? What do they think about PC Mall catalog distribution?
2. What are the ranges of customer satisfaction and the propensity for repeat purchase? How
likely is the customer base to recommend PC Mall to others?
3. How satisfied are customers with key benefit types of PC Mall services?
4. What is the relative impact of each benefit type on the overall satisfaction with PC Mall?
Appendix B provides you with the questionnaire used in an online survey that PC Mall conducted
among 50,000 consumers who had purchased in the past from PC Mall or MacMall, and whose email addresses were registered in the company’s database.
Nearly 5,000 responses were received during the two-week period allocated for on-line completion
of the questionnaires. Of them, 3,979 questionnaires were valid and could be used for the data
analysis. The invalid questionnaires, which were excluded from the analysis, were excessively
incomplete and/or contained contradictory answers. Of all the valid questionnaires, 232 non-users
were further excluded. Consumers who purchased at PC Mall and/or MacMall more than 2 years
ago were classified as “non-users”. It was decided that consumers with purchasing experience dated
from more than two years ago would not provide relevant and/or accurate answers.
This resulted in a final sample size of 3,744 respondents, which contained 644 PC Mall users
(17.2%), 2,377 MacMall users (63.5%), and 723 users of both PC and MacMall (19.3%). To obtain
the clearest inferences, PC Mall users, MacMall users, and PC/MacMall users were treated as three
separate user segments in the data analysis and interpretation.
To address PC Mall’s decision problem, below are specific outputs you are required to generate from
the data set PC_Mall.csv.
The instructions are as follows:
Conduct all statistical analyses using the R code provided to you along with the data
(PC_Mall.R). For the first few analyses, you only need to execute the code (i.e., press
Ctrl+R). You will then be asked to modify the code on your own for some later analyses.
Use R outputs as the basis to support and justify your interpretation of the data. Keep in
mind that a good summary of a data analysis reads much like a news report that is supported
by solid data-driven and statistical evidence.
Keep your result summary concise yet insightful. The summary of the outputs (excluding
the Appendix) should be no longer than two pages. If possible, please format in 11-point
Times New Roman at the smallest, with single spacing. Please also include the R outputs in
an Appendix, as instructed below. The easiest way to do this is to include screenshots of the
outputs.
Here is what your report needs to address:
1. Based on Questions 1.1–1.6, use the summary statistics to describe:
how many products customers have purchased from PCMall in the past year (Q1.1),
from which businesses (Q1.2), and through which methods they have made these
purchases (Q1.3);
what influenced them to make these purchases (Q1.4); and
how customers feel about the current frequency of catalog distribution, as well as
their preferred frequency (Q1.5 and 1.6).
Briefly summarize key results based on these summary statistics and create a pie chart for
Q1.1 and a bar chart for Q1.3. Note that Q1.2 will later be used as our main independent
variable to classify users into different segments. Include the pie and bar charts in the
Appendix, but do not include any other R outputs for this section.
2. Create a cross-tabulation table and conduct a hypothesis test to examine whether
purchase quantities (Q1.1) differ by PCMall user segment (Q1.2). Q1.1 is considered the
dependent variable in this analysis. Briefly summarize your key results. For this analysis, you
will need to include an R output in the Appendix.
3. Compute appropriate summary statistics to examine consumers’ overall satisfaction with
PCMall (Q8.1 and Q8.2) and their propensity to make recommendations (Q8.4). To obtain a
measure of overall satisfaction, please average each participant’s ratings for Q8.1 and Q8.2.
Additionally, please do the following:
Compute means for overall satisfaction across three user segments, and then
conduct an ANOVA test for whether the means for overall satisfaction are the same
or different across segments.
Repeat the same analysis as above, but this time for propensity to make
recommendations (simply replicate the same R code provided to you but with a
different dependent variable).
For propensity to recommend, test whether the means between any pair of user
segments are significantly different from each other at α=.05. Briefly summarize your
results. Please include R outputs only for all the hypotheses tests.
4. Prior to this survey, PCMall’s management had conducted in-depth interviews with 15
PCMall customers to determine attributes that are important to them. These attributes
had been grouped into six benefit types: support services, product information, pricing,
sales service, product, and web convenience (see Appendix A).
Compute the measure of customer satisfaction with each benefit type by averaging
the attribute ratings that correspond to each benefit type (e.g., the average ratings for
timeliness of delivery, the accuracy of delivery, the packaging of delivered products,
and order status information to derive a measure of “support services” satisfaction).
In this computation, ignore Q2.2, Q4.2, Q5.2, Q5.3, Q6.2, and Q7.1.
Compute appropriate summary statistics for the new variables that you computed to
capture customer satisfaction with the six benefit types. Notice that there are
separate sets of attribute ratings for PCMall and MacMall web convenience. So, we
need separate summary statistics for PCMall web convenience and MacMall web
convenience. Briefly summarize your results and report your key summary statistics.
Which benefit types are customers more (or less) satisfied with?
Perform a hypothesis test on whether the mean for customer satisfaction is different
from that of product and for pricing. Briefly summarize your results.
For this section, you do not need to include an R output.
5. Perform a multiple linear regression analysis with the overall satisfaction you computed in
3., with Q8.1 and Q8.2 as the dependent variables and satisfactions with the six benefit
types you computed in 4. as the independent variables. Be sure to run a separate regression
for each segment (PCMall vs. MacMall vs. both users). For simplicity, you are not required
to conduct a full-scale diagnostic test (e.g., a residual plot), as we discussed in our live class
(C2). Again, you are provided with a partial R code to obtain these regression results. Briefly
interpret the results you obtain for each user segment. Make sure you include the R output
for each segment’s regression results.
6. Obtain the mean satisfaction for each benefit type for the different user segments. Populate
each cell in the table below using these results, together with the regression results you
obtained in 5. above. In each cell, be sure to include the following details:
the mean satisfaction rating associated with each benefit type and each user segment;
and
the regression coefficient for each benefit type for each user segment.
For this analysis, you do not need to provide the R output.
Benefit Types
Support
Service
Product
Information
Pricing Sales
Service
Product PC Web
Convenience
Mac Web
Convenience
Segments
PC Mall
users
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
MacMAll
users
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
Both
PC/Mac
Mall
users
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
Mean
RegCoef
What key insights you can draw from this table? Please also insert this table in your
write-up (NOT in the Appendix).
7. Based on all the findings, please provide two specific recommendations to PCMall’s
management that are particularly related to the results shown in the summary table
below. Additionally, provide a brief justification for your recommendations (i.e., which
findings lead you to such a recommendation).
APPENDIX A:
ATTRIBUTES AFFECTING OVERALL SATISFACTION LEVELS
Benefit Type Identified Attributes
Support Services
1. Timeliness of delivery
2. Accuracy of delivery
3. Packaging of delivered product
4. Order status information
Product
Information
1. Product information quality
2. Catalog information
3. Product information accuracy
4. Information on compatibility of
products
5. Stock information
6. Product comparisons
Pricing
1. Competitiveness of prices
2. Timeliness of rebates
3. Shipping costs
4. Package deals
5. Consistency of pricing
Sales Service
1. Response to email inquiry
2. Phone service availability
3. Attitude of sales staff
4. Knowledgeable sales people
5. Ease of phone order
6. Technical support – post
purchase
7. Return policy
8. Exchange policy
Product
1. Product variety
2. Product selection w/in category
3. Stock availability
4. Latest products
Web Convenience
1. Web site design
2. Ease of navigation
3. Ease of check-out process on
web
4. Web page speed
5. Ease of product search
6. Ease of rebate search
7. Readability of web pages