# Store24 Case R2

Pages: 12 (3297 words) Published: April 23, 2015
﻿Date: 12/04/2014
From: Jason Chambers, Joost Rietdijk, Barry Dunham
To: Bob Gordon CEO, and Paul Doucette CFO
Subject: Store 24
Store 24 Case
Introduction:
Store 24 is a convenience store located in the New England area. They are currently operating in a highly competitive market and as a result are in need of further differentiation. The old differentiation method was called “Cause You Just Can’t Wait” (CYJCW) which focused on customer convenience with regards to locating products and getting in and out of the store quickly. The newest differentiation strategy is dubbed “Ban Boredom” and is tailored towards whom they believe is their current target market, urban youth and young adults aged 14 – 29. They feel this group is bored easily and that ensuring that their customers are entertained while at the store will be a beneficial differentiation strategy. Senior leadership wants to know if the Ban Boredom initiative is a good determinant of store financial performance.

Statistical Analyses
For the Store 24 case a dataset of 30 convenience stores was given. For each of these stores a future controllable contribution was given which is the margin less controllable expenses one period into the future, in this case the period ending April 2000. The data for the stores are from a previous period ending January 2000. It is expected that the data from the period be utilized via a multivariate of regression analysis to predict the stores Future Controllable Contribution (hereinafter “profit”). Univariate Analysis

Figure 1

To start the analysis we loaded the dataset and observed some descriptive statistics to better understand the basics of the dataset. From the descriptive statistics (Appendix A), we learn that the average of profit for each store is about \$32,554 and the standard deviation is \$11,580; this would imply that the difference in profit between the stores profit is quite large. Figure 1 above illustrates the store’s profit in a graph. It seems like there are two different clusters of data a cluster of stores with lower profit between \$20,000 and \$25,000 and another cluster of stores at the higher end between \$40,000 - \$45,000. This is confirmed by the histogram in figure 1, since the distribution of the data seems to represent a binomial distribution rather than a normal distribution required by linear regression. There are two options that could be used to address this issue. First, the continuous variable profit could be transformed into a binary indicator of high profit and low profit in order to adjust for violation of assumptions. Second, that the dataset could be split into two different regressions because according to the probability density plot, it seems that we have two normal distributions and the assumption of linearity does not seem to be violated within each group. Figure 2

Variable
Low Profit Mean
High Profit Mean
Difference (H-L)
t-stat
p-value
Future_Controllable_Contribution
21422.93
43686.27
22263.34
-24.6283
2.20E-16
CYJCW_Score
90.632
89.25533
-1.37667
0.696
0.4923
Ban Boredom_Score
106.6667
114.6607
7.994
-0.9403
0.3553
Crew_Skills
2.977333
3.717333
0.74
-6.5161
1.30E-06
Manager_Skills
3.292
3.304667
0.012667
-0.0655
0.9484
Population
11545.6
11669.07
123.47
-0.0498
0.9606
Per_Capita_Income
23849.07
22227.4
-1621.67
0.3579
0.7232
Number_of_Competitors
4.066667
4.533333
0.466666
-0.7182
0.4789

Next the variable Future Controllable Contribution was used to split the data set into a high profit group and a low profit group. All data from store #1598 on was used as the high profit group, and anything preceding store #1598 was inserted in the low profit group. The next step in the statistical analysis process was to perform a “difference of means test” to identify the variables that can be used to predict the profit of each store. Figure 2 above shows the output from this analysis. The results...