chapter – 4 testing of hypothesis and model building
TRANSCRIPT
213
CHAPTER – 4
TESTING OF HYPOTHESIS AND MODEL BUILDING
TESTING OF HYPOTHESIS – ORGANISED RETAIL OUTLETS Testing of Hypothesis: 98
Inferential statistics is concerned with estimating the true value of population
parameters using sample statistics. A statistical hypothesis is a claim (assertion,
statements, belief or assumption) about an unknown population parameter value.
On the basis of sample findings, the hypothesized value of the population
parameter is either accepted or rejected. The process that enables a decision
maker to test the validity (or significance) of his claim by analysing the difference
between the value of sample statistic and the corresponding hypothesised
parameter value is called testing of hypothesis.
Hypothesis 1
One-Way ANOVA
There are three different groups of customers namely customers visit with
family, friends or alone (independent variable) to an outlet.
1. The time spent by the groups (dependent variable – continuous variable)
2. The purchase made by the groups (dependent variable – continuous
variable)
Null Hypotheses
1. Mean time spent by different groups of customers in the organised outlets
was same.
2. Mean amount of purchase made by different groups of customers in the
organised outlets was same
98 Sharma, J.K., Business Statistics, Pearson Education, Second Edition, 2007, pp327-329.
214
Table 4.1: Descriptive Statistics (Time Spent)
N Mean Std.
Deviation
Std.
Error
95% Confidence
Interval for Mean Minimum Maximum
Lower
Bound
Upper
Bound
Gr_tim
Family 150 43.8667 15.02734 1.22698 41.4421 46.2912 13.00 113.00
Friends 150 28.1000 10.17086 .83045 26.4590 29.7410 8.00 45.00
Alone 150 21.4133 9.27448 .75726 19.9170 22.9097 3.00 40.00
Total 450 31.1267 15.05371 .70964 29.7320 32.5213 3.00 113.00
Gr_purc
Family 150 931.50 343.741 28.066 876.04 986.96 200 2000
Friends 150 609.00 227.425 18.569 572.31 645.69 150 1100
Alone 150 462.83 238.577 19.480 424.34 501.33 100 1000
Total 450 667.78 337.170 15.894 636.54 699.01 100 2000
Table 4.2: Test of Homogeneity of Variances (Time Spent – Levene Statistics)
Levene Statistic df1 df2 Sig.
gr_tim 15.555 2 447 .000
gr_purc 4.903 2 447 .008
The significance value of 0.000 in the case of time spent by the groups and
0.008 in the case of purchases made by the groups indicate that the variances of
time spent and the purchases made by the groups do indeed differ significantly
(Table 4.1 and Table 4.2).
The ANOVA table also shows the significance value of 0.000 in both the
cases namely time spent by the groups and purchases made by the groups. This
shows that the mean time spent and the mean amount of purchases made by the
groups differ significantly (Table 4.3).
215
Table 4.3: ANOVA (Time Spent)
Sum of Squares
df Mean Square F Sig.
gr_tim
Between Groups
39872.573 2 19936.287 144.019
0.000 Within Groups 61877.207 447 138.428
Total 101749.780 449
gr_purc
Between Groups
1.725E7 2 8625484.722 114.094
0.000 Within Groups 3.379E7 447 75599.683
Total 5.104E7 449
Table 4.4: Contrast Coefficients (Time Spent) – I
Contrast gr_name
Family Friends Alone
1 0 1 -1
2 1 0 -1
Table 4.5: Contrast Tests (Time Spent) – II
The contrast tests show that the time spent and purchases made by the
groups namely customers visiting alone and customers visiting with friends were
significantly different (with their means) under both the conditions namely
Contrast Value of
Contrast Std. Error t df
Sig. (2-
tailed)
gr_tim
Assume equal
variances
1 6.6867 1.35857 4.922 447 .000
2 22.4533 1.35857 16.527 447 .000
Does not assume
equal variances
1 6.6867 1.12387 5.950 295.499 .000
2 22.4533 1.44184 15.573 248.127 .000
Gr_purc
Assume equal
variances
1 146.17 31.749 4.604 447 .000
2 468.67 31.749 14.762 447 .000
Does not assume
equal variances
1 146.17 26.912 5.431 297.320 .000
2 468.67 34.164 13.718 265.515 .000
216
‘assuming equal variances’ and ‘not assuming equal variances’. As shown by the
Levene’s test, ‘variances are different’ are considered (Table 4.4 to Table 4.6).
The contrast tests show that the time spent and purchases made by the
groups namely customers visiting alone and customers visiting with family were
significantly different (with their means) under both the conditions namely
‘assuming equal variances’ and ‘not assuming equal variances’. But as shown by
the Levene’s test, ‘the variances are different’ are considered in the context (Table
4.4 to Table 4.6).
Table 4.6: Multiple Comparisons – Least Significant Difference (LSD)
Method (Time Spent)
Depende
nt
Variable
(I)
gr_name
(J)
gr_name
Mean
Difference (I-J) Std. Error Sig.
95% Confidence Interval
Lower Bound Upper Bound
gr_tim
Family Friends 15.76667
* 1.35857 .000 13.0967 18.4366
Alone 22.45333* 1.35857 .000 19.7834 25.1233
Friends Family -15.76667
* 1.35857 .000 -18.4366 -13.0967
Alone 6.68667* 1.35857 .000 4.0167 9.3566
Alone Family -22.45333
* 1.35857 .000 -25.1233 -19.7834
Friends -6.68667* 1.35857 .000 -9.3566 -4.0167
gr_purc
Family Friends 322.500
* 31.749 .000 260.10 384.90
Alone 468.667* 31.749 .000 406.27 531.06
Friends Family -322.500
* 31.749 .000 -384.90 -260.10
Alone 146.167* 31.749 .000 83.77 208.56
Alone Family -468.667
* 31.749 .000 -531.06 -406.27
Friends -146.167* 31.749 .000 -208.56 -83.77
*. The mean difference is significant at the 0.05 level.
– In the above table 4.6, the asterisks marked variables show that the mean
of the given two pairs differ significantly.
– This shows that the mean time spent in the outlets by different groups
namely
217
family vis-a-vis friends and alone friends vis-a-vis family and alone alone vis-a-vis family and friends
are significantly different.
– Similarly, from the above table, it may be inferred that the mean amount of
purchases made by different groups namely
family vis-a-vis friends and alone friends vis-a-vis family and alone alone vis-a-vis family and friends
are significantly different.
Hypothesis 2 One-Way ANOVA
H0: Mean of the percentage of sales over different period of a month (namely 1-10, 11-20 and 21-30 days) are same. H1: Mean of the percentage of sales over different period of a month (namely 1-10, 11-20 and 21-30 days) are not same.
Table 4.7: Descriptives (Percentage of Sales)
Days N Mean Std.
Deviation
Std.
Error
95% Confidence
Interval for Mean Min Max
Lower
Bound
Upper
Bound
1-10 150 46.53 8.608 .703 45.14 47.92 30 75
11-20 150 27.22 7.175 .586 26.06 28.38 10 50
21-30 150 26.52 8.216 .671 25.19 27.85 10 40
Total 450 33.42 12.256 .578 32.29 34.56 10 75
Table 4.8: Test of Homogeneity of Variances (Percentage of Sales)
Levin
Statistic df1 df2 Sig.
3.708 2 447 .025
218
Levin Statistics shows the significance value of 0.025. This shows that the
variances of percentage of sales over different period of a month do differ
significantly (Table 4.7 and Table 4.8).
Table 4.9: ANOVA (Percentage of Sales)
Sum of Squares df Mean Square F Sig.
Between Groups 38675.204 2 19337.602 300.464
.000 Within Groups 28768.573 447 64.359 -
Total 67443.778 449 - -
The ANOVA table also shows the significance value of 0.000. This shows
that the mean of the percentage of sales over different period of a month differ
significantly (Table 4.9).
Table 4.10: Contrast Coefficients (Percentage of Sales)
Contrast Days of a Month
1-10 11-20 21-30
1 1 0 -1
2 0 -1 1
3 -1 1 0
Table 4.11: Contrast Tests (Percentage of Sales)
Percentage of Sales Con-
trast
Value of
Contrast
Std.
Error t df
Sig.
(2-tailed)
Assume equal
variances
1 20.01 .926 21.597 447 .000
2 -.70 .926 -.756 447 .450
3 -19.31 .926 -20.842 447 .000
Does not assume
equal variances
1 20.01 .972 20.591 297.353 .000
2 -.70 .891 -.786 292.690 .433
3 -19.31 .915 -21.101 288.628 .000
Under both the conditions namely ‘assuming equal variances’ and ‘not
assuming equal variances’ (which is applicable in the present context), the contrast
219
tests show that the mean of the percentage of sales over the first ten days (1st -
10th) differ significantly with next ten days (11th-20th) and also last ten days (21st-
30th) of a month (Table 4.10 to Table 4.12).
Under both the conditions namely ‘assuming equal variances’ and ‘not
assuming equal variances’ (which is applicable in the present context), the contrast
tests show that the mean of the percentage of sales over 11th – 20th does not differ
significantly with the last ten days 21st– 30th of a month (Table 4.10 to Table 4.12).
Table 4.12: Post Hoc Tests – Multiple Comparisons (Percentage of
Sales) Least Significant Difference (LSD) Method
(I)
Day_
Mon
(J)
Day_
Mon
Mean
Difference
(I-J)
Std. Error Sig.
95% Confidence Interval
Lower
Bound
Upper
Bound
1-10 11-20 19.307* .926 .000 17.49 21.13
21-30 20.007* .926 .000 18.19 21.83
11-20 1-10 -19.307* .926 .000 -21.13 -17.49
21-30 .700 .926 .450 -1.12 2.52
21-30 1-10 -20.007* .926 .000 -21.83 -18.19
11-20 -.700 .926 .450 -2.52 1.12
*. The mean difference is significant at the 0.05 level.
In the above table 4.12, the asterisks show that the mean of the given two
pairs differ significantly. It could be inferred from the table that
The mean of percentage of sales over the first ten days (1st – 10th) differ
significantly with 11th – 20th and also 21st – 30th of a month.
The mean of percentage of sales over the second ten days (11th – 20th)
differ significantly with 1st – 10th but not with the 21st – 30th of a month.
The mean of percentage of sales over the last ten days (21st – 30th) differ
significantly with 1st – 10th but not with 11th – 20th of a month.
220
CROSS-TABULATION AND CHI-SQUARE ANALYSIS Hypothesis 3
H0: Footfalls at an organised retail outlet is independent of the location of the outlet. H1: Footfalls at an organised retail outlet is not independent of the location of the outlet.
Table 4.13: Case Processing Summary (Independence of Location)
Cases
Valid Missing Total
N Percent N Percent N Percent
Location *
Footfalls 126 100.0% 0 .0% 126 100.0%
Table 4.14: Location * Footfalls Cross-Tabulation
Location Footfalls
101-150 151-200 201-250 Total
Main Road
Count 26 27 14 67
Expected Count 27.1 25.5 14.4 67.0
Residual -1.1 1.5 -.4
Middle of the
Street
Count 19 9 8 36
Expected Count 14.6 13.7 7.7 36.0
Residual 4.4 -4.7 .3
Shopping
Complex
Count 6 12 5 23
Expected Count 9.3 8.8 4.9 23.0
Residual -3.3 3.2 .1
Total Count 51 48 27 126
Expected Count 51.0 48.0 27.0 126.0
221
Table 4.15: Chi-Square Tests (Location)
Value df
Asymp. Sig.
(2-sided)
Pearson Chi-
Square 5.492a 4 .240
Likelihood Ratio 5.644 4 .227
N of Valid Cases 126
a. 1 cells (11.1%) have expected count less than 5. The minimum expected count is 4.93.
Table 4.16: Symmetric Measures (Location)
Value Approx. Sig.
Nominal by
Nominal
Phi .209 .240
Cramer's V .148 .240
N of Valid
Cases 126
Inference: The Chi-square analysis of the hypothesis is given in the table
4.13 to table 4.15. The table show χ20.05,4 value as 0.240. i.e. the footfalls at an
organised retail outlet is independent of the location of the outlet. The phi and
Cramer’s V measures of association (Table 4.16) are also small and do not
approach significance.
The test conducted showed that the footfall is not dependent on location
alone. There may be other influencing factors that affect the footfalls.
Hypothesis 4
H0: Number of customers is independent of the size of the organised retail outlets. H1: Number of customers is not independent of the size of the organised retail outlets.
222
Table 4.17: Case Processing Summary (Size)
Cases
Valid Missing Total
N Percent N Percent N Percent
FD_Area *
FD_NCust 147 100.0% 0 .0% 147 100.0%
Table 4.18: FD_Area * FD_NCust Cross-Tabulation
FD_Area FD_NCust
1000 – 1500 1501 – 2000 2001 – 2500 2501 – 3000 Total
1000 – 1500
Count 45 0 0 0 45
Expected Count 13.8 22.0 4.0 5.2 45.0
Residual 31.2 -22.0 -4.0 -5.2
1501 – 2000
Count 0 72 0 0 72
Expected Count 22.0 35.3 6.4 8.3 72.0
Residual -22.0 36.7 -6.4 -8.3
2001 – 2500
Count 0 0 13 0 13
Expected Count 4.0 6.4 1.1 1.5 13.0
Residual -4.0 -6.4 11.9 -1.5
2501 – 3000
Count 0 0 0 17 17
Expected Count 5.2 8.3 1.5 2.0 17.0
Residual -5.2 -8.3 -1.5 15.0
Total Count 45 72 13 17 147
Expected Count 45.0 72.0 13.0 17.0 147.0
Table 4.19: Chi-Square Tests (Size)
Value df Asymp. Sig. (2-sided)
Pearson Chi-Square 4.410E2 9 .000
Likelihood Ratio 345.730 9 .000
N of Valid Cases 147
a. 6 cells (37.5%) have expected count less than 5. The minimum expected count is 1.15.
223
Table 4.20: Symmetric Measures (Size)
Value Approx. Sig.
Nominal by
Nominal
Phi 1.732 .000
Cramer's V 1.000 .000
N of Valid Cases 147
Inference: The Chi-square analysis (Table 4.17 to Table 4.19) shows that
the χ20.05,9 value is .000, i.e. null hypothesis could not be accepted. It may be
concluded that the number of customers for an organised retail outlet is not
independent of the size of the outlet. Also, the phi and Cramer’s V measures of
association (Table 4.20) show that the number of customers and size of the
organised outlets are strongly associated.
Hypothesis 5 H0: Number of footfalls at an organised retail outlet is independent of the size of the
outlet. H1: Number of footfalls at an organised retail outlet is not independent of the size of
the outlet.
Table 4.21: Case Processing Summary (Footfalls)
Cases
Valid Missing Total
N Percent N Percent N Percent
FD_Area *
Footfalls 132 100.0% 0 .0% 132 100.0%
224
Table 4.22: FD_Area * Footfalls Cross tabulation (Footfalls)
FD Area Footfalls
101-150 151-200 201-250 Total
1000 – 1500
Count 22 14 7 43
Expected Count 17.3 16.9 8.8 43.0
Residual 4.7 -2.9 -1.8
1501 – 2000
Count 29 31 7 67
Expected Count 26.9 26.4 13.7 67.0
Residual 2.1 4.6 -6.7
2001 – 2500
Count 1 6 4 11
Expected Count 4.4 4.3 2.2 11.0
Residual -3.4 1.7 1.8
2501 – 3000
Count 1 1 9 11
Expected Count 4.4 4.3 2.2 11.0
Residual -3.4 -3.3 6.8
Total Count 53 52 27 132
Expected Count 53.0 52.0 27.0 132.0
Table 4.23: Chi-Square Tests (Footfalls)
Value df
Asymp. Sig.
(2-sided)
Pearson Chi-Square 36.525a 6 .000
Likelihood Ratio 31.642 6 .000
N of Valid Cases 132
a. 6 cells (50.0%) have expected count less than 5. The minimum expected count is 2.25.
Table 4.24: Symmetric Measures (Footfalls)
Value Approx. Sig.
Nominal by Nominal
Phi .526 .000
Cramer's V .372 .000
N of Valid
Cases 132
225
Inference: The Chi-square analysis (Table 4.21 to Table 4.23) shows that
the χ20.05,6 value is 0.000, i.e. null hypothesis could not be accepted. It is concluded
that the number of footfalls at an organised retail outlet is not independent of the
size of the outlet. Also, the phi and Cramer’s V measures of association
(Table 4.24) confirm the same.
CORRELATION ANALYSIS Hypothesis 6
H0: Total Number of Customers depends on Catchment area of an organised retail
outlet. H1: Total Number of Customers does not depend on Catchment area of an
organised retail outlet.
Table 4.25: Correlations (Catchment Area and No. of Customers) – I
Catch_Area No_of_Customers
Catch_Area
Pearson Correlation 1 .179*
Sig. (2-tailed) .028
N 150 150
No_of_Customers
Pearson Correlation .179* 1
Sig. (2-tailed) .028
N 150 150
*. Correlation is significant at the 0.05 level (2-tailed).
Table 4.26: Correlations (Catchment Area and No. of Customers) - II
Spearman's rho Catch_Area No_of_Customers
Catch_Area
Correlation Coefficient 1.000 .160
Sig. (2-tailed) . .051
N 150 150
No of Customers
Correlation Coefficient .160 1.000
Sig. (2-tailed) .051 .
N 150 150
226
Inference: Both the Pearson’s and Spearman’s tests (Table 4.25 and Table
4.26) show that the number of customers for an outlet and catchment area are
correlated.
Hypothesis 7
H0: Footfalls per day depend on Catchment area of an organised retail outlet
H1: Footfalls per day do not depend on Catchment area of an organised retail outlet.
Table 4.27: Correlations (Footfalls Vs Catchment Area) – I
foot_fall_mid Catch_Area
Foot_fall_mid
Pearson Correlation 1 .240**
Sig. (2-tailed) .003
Sum of Squares and Cross-products
681420.833 1625.667
Covariance 4573.294 10.911
N 150 150
Catch_Area
Pearson Correlation .240** 1
Sig. (2-tailed) .003
Sum of Squares and Cross-products
1625.667 67.173
Covariance 10.911 .451
N 150 150
**. Correlation is significant at the 0.01 level (2-tailed).
Table 4.28: Correlations (Footfalls Vs Catchment Area) – II
foot_fall_mid Catch_Area
Spearman's rho
foot_fall_mid
Correlation
Coefficient 1.000 .306**
Sig. (2-tailed) . .000
N 150 150
Catch_Area
Correlation
Coefficient .306** 1.000
Sig. (2-tailed) .000 .
N 150 150
**. Correlation is significant at the 0.01 level (2-tailed).
227
Inference: Both the Pearson’s and Spearman’s tests show (Table 4.27 and
Table 4.28) that the footfalls at an outlet and catchment area are correlated.
Hypothesis 8 H0: The sales of the outlets affected by the competition is not more than 10%
H1: The sales of the outlets affected by the competition is more than 10%
One-sample t-test is used to test the hypothesis. The result is shown in
Table 4.29 and table 4.30. The p-value (p > .05) is not significant. i.e., statistically
the null hypothesis cannot be rejected. It can be inferred that the competition for
the retail outlets has affected the sales not more than 10%
Table 4.29: One-Sample Statistics (Competition)
N Mean Std.
Deviation
Std. Error
Mean
Compet_Sal_Mid 150 10.0167 6.00263 .49011
Table 4.30: One-Sample Test (Competition)
Test Value = 10
T Df Sig. (2-
tailed)
Mean
Difference
95% Confidence Interval
of the Difference
Lower Upper
Compet_Sal_Mid .034 149 .973 .01667 -.9518 .9851
TESTING OF HYPOTHESIS –KIRANA STORES Hypothesis 9 Correlation Analysis of Store Size and Number of Customers
To know whether there exists a relationship between the store size and
number of customers, correlation analysis was undertaken. The result is as follows.
228
H0: The number of customers of a store is dependent on size of the store H1: The number of customers of a store is not dependent on size of the store
Table 4.31: Correlation Between the Size and
the Number of Customers (kirana Stores)
Size NumCust
Size
Pearson Correlation 1.000 .407**
Sig. (2-tailed) .000
N 200.000 200
NumCust
Pearson Correlation .407** 1.000
Sig. (2-tailed) .000
N 200 200.000
**Correlation is significant at the 0.01 level (2-tailed).
Inference: The correlation analysis (Table 4.31) shows that there exists a
strong positive relationship between the size of a store and the number of
customers. The Pearson Correlation is r=0.407, p<0.01 (Two-tailed).
Hypothesis 10
Ho: Number of relatives is independent of number of staff working in a kirana store. H1: Number of relatives depends on number of staff working in a kirana store.
Table 4.32: Correlations (Relatives Vs No. of Staff)
Spearman's rho
and Pearson's
Staff Relatives
Staff
Correlation
Coefficient 1.000 .380**
Sig. (2-tailed) . .000
N 200 200
Relatives
Correlation
Coefficient .380** 1.000
Sig. (2-tailed) .000 .
N 200 200
**. Correlation is significant at the 0.01 level (2-tailed).
229
The testing of hypothesis shows the p value <0.05 (Table 4.32). Therefore,
statistically null hypothesis, that the number of relatives is independent of number
of staff working in a kirana store could not be accepted. That is, the number of
relatives working in a kirana store depends on the number of staff. i.e. more the
number of staff, more the number of relatives working in a store.
Hypothesis 11
Multiple Correlation
H0: Average sales is dependent on size of the store, number of customers of a store, number of footfalls per day, average time spent by a customer. H1: Average sales is not dependent on size of the store, number of customers of a store, number of footfalls per day, average time spent by a customer.
The analysis (Table 4.33 to Table 4.35) show that the average sales is
dependent on number of footfalls per day, average sales per day and the time
spent by a customer.
Table 4.33: Variables Entered / Removedb (Multiple Correlation)
Model Variables Entered Variables Removed Method
1 Avg_time_spent, Aver_cust, Avg_sales, Size, NumCusta
. Enter
a. All requested variables entered. b. Dependent Variable: Avg_sale_store.
Table 4.34: Coefficientsa (Multiple Correlation)
Model
Unstandardized Coefficients
Standardized Coefficients
t Sig.
95% Confidence Interval for B
B Std. Error
Beta Lower Bound
Upper Bound
1
(Constant) -3596.498 238.631 -15.071 .000 -4067.142 -3125.853
Size -.287 .360 -.023 -.798 .426 -.996 .422
NumCust .401 .987 .012 .406 .685 -1.546 2.347
Aver_cust 48.262 3.057 .433 15.789 .000 42.233 54.290
Avg_sales 70.632 2.343 .719 30.142 .000 66.010 75.253
Avg_time_spent
49.916 21.550 .065 2.316 .022 7.414 92.418
a. Dependent Variable: Avg_sale_store
230
Table 4.35: Coefficient Correlationsa (Multiple Correlation)
Model Avg_time_
spent Aver_cust Avg_sales Size
Num Cust
1
Correlations
Avg_time_spent 1.000 -.073 -.095 -.534 -.027
Aver_cust -.073 1.000 -.100 .207 -.532
Avg_sales -.095 -.100 1.000 .027 -.192
Size -.534 .207 .027 1.000 -.345
NumCust -.027 -.532 -.192 -.345 1.000
Covariances
Avg_time_spent 464.397 -4.824 -4.801 -4.135 -.581
Aver_cust -4.824 9.344 -.716 .228 -1.605
Avg_sales -4.801 -.716 5.491 .023 -.445
Size -4.135 .228 .023 .129 -.123
NumCust -.581 -1.605 -.445 -.123 .974
a. Dependent Variable: Avg_sale_store Comparison of traditional kirana stores and modern organised retail outlets
When sample data do not meet the basic assumptions that underlie the
parametric procedure (e.g. normality or homogeneity of variance), nonparametric
methods are used. Kolomogorov-Smirnov test is used to find whether the given
distribution is normally distributed or not. Wilcoxon- Mann-Whitney independent
two sample test is used to find the independence of the two populations.99
i. Comparison of Sales per Sq. foot (Rs.)
The investigator was interested in comparing the sales per square foot of
organised retail outlets and unorganised kirana stores. Table 4.36 and Figure 4.1
show the information regarding sales per square foot. Further, non-parametric test
was used to test the difference. The result is as follows.
99 Carver, Robert H. and Nash, Jane G, “Data Analysis with SPSS”, Cengage Learning, India Edition, 2009
231
Table 4.36: Comparison of Sales per Square foot
Source: Field Survey.
Figure 4.1: Sales per Sq. foot (in Rs.)
Source: Field Survey.
Sales (Rs.) Kirana Stores Organised Outlets
No. of Outlets Percentage No. Of Outlets Percentage <250 69 34.5 17 11.3
250 – 500 35 17.5 22 14.7 500 – 750 22 11.0 30 20.0
750 – 1000 15 7.5 48 32.0
1,000 – 1,250 28 14.0 22 14.7 1,250 – 1,500 16 8.0 7 4.7 1,500 – 1,750 5 2.5 2 1.3
1,750 – 2,000 6 3.0 1 0.7 >2,000 4 2.0 1 0.7
Total 200 100.0 150 100.0
232
Hypothesis 12
One sample Kolomogorov-Smirnov test shows that the metric sales per
square foot of kirana stores were not normally distributed (p < 0.05). But the sales
per square foot of organised retail outlets were normally distributed (p > .05)
(Table 4.37).
Table 4.37: One-Sample Kolmogorov-Smirnov Test
(Sales per Square Foot)
Sales per Sq. foot
Kirana Stores Organised Retail Outlets
N 200 150
Normal Parametersa Mean 663.9800 758.8117
Std. Deviation 599.30898 363.54617
Most Extreme
Differences
Absolute .157 .062
Positive .157 .062
Negative -.143 -.050
Kolmogorov-Smirnov Z 2.221 .755
Asymp. Sig. (2-tailed) 0.000 0.620
a. Test distribution is Normal.
Wilcoxon-Mann-Whitney independent two sample test is used to find the
independence of the two populations.
Table 4.38: Mann-Whitney Test (Sales per Square foot)
Ranks
Kira_or_Organ N Mean
Rank
Sum of
Ranks
Sales_Sq_ft.
Organised Retail
Outlets 150 196.97 29545.50
Kirana Stores 200 159.40 31879.50
Total 350
233
Table 4.39: Test Statistics (Sales per Square Foot)
Test Statisticsa
Sales_Sq_Mt
Mann-Whitney U 11779.500
Wilcoxon W 31879.500
Z -3.438
Asymp. Sig. (2-tailed) 0.001
a. Grouping Variable: Kira_or_Organ
The result (Table 4.38 and Table 4.39) suggests that there exists
statistically significant difference between the underlying distributions of the sales
per square foot of organised retail outlets and the kirana stores (z = -3.438, p <
0.05). Mann-Whitney Test is more conservative than the parametric test such as
t-test. The sales per square foot of an organised retail outlets is greater than that of
kirana store.
The two sample independent t-test (Table 4.40 and Table 4.41) does
suggest that the sales per square foot is not statistically different (t=1.833, p>0.05).
Table 4.40: Independent Two-Sample T-test (Sales per Square foot)
Group Statistics
Kira_or_Organ N Mean Std. Deviation Std. Error Mean
Organised Retail
Outlets 150 758.8117 363.54617 29.68342
Kirana Stores 200 663.9800 599.30898 42.37754
234
Table 4.41: Equality of Variances (Sales per Square foot)
Levene's Test for Equality of Variances
t-test for Equality of Means
Sales_Sq_ft F Sig. t Df Sig. (2-
tailed)
Mean Differen
ce
Std. Error
Difference
95% Confidence Interval of the
Difference
Lower Upper
Equal variances assumed
36.66 0.00
0 1.715 348 0.087 94.832 55.285 -13.902 203.566
Equal variances not assumed
1.833 334.60 0.068 94.832 51.739 -6.9437 196.607
ii. Comparison of Sales per Employee (Rs. in Lakhs)
Comparison of the sales per employee of organised retail outlets and
unorganised kirana stores was done. Table 4.42 and Figure 4.2 show the
information regarding sales per employee. Also, non-parametric test was used to
compare the same. The result is as follows.
Table 4.42: Comparison of Sales per Employee (Rs. in Lakhs)
Organises Retail Outlets Kirana Stores
Amount (Rs.)
No. of Outlets
Percentage Amount
(Rs.) No. of Outlets
Percentage
<0.3 3 2 <0.1 21 10.5
0.3 - 0.6 18 12 0.1 – 0.2 30 15
0.6 - 0.9 32 21 0.2 – 0.3 30 15
0.9 - 1.2 21 14 0.3 – 0.4 22 11
1.2 - 1.5 36 24 0.4 – 0.5 20 10
1.5 - 1.8 24 16 0.5 – 0.6 47 23.5
1.8 -2.1 6 4 0.6 – 0.7 0 0
2.1 - 2.4 2 1 0.7 – 0.8 22 11
2.4 - 2.7 0 0 0.8 – 0.9 2 1
2.7 - 3.0 4 3 0.9 – 1.0 2 1
3.0 - 3.3 1 1 >1.0 4 2
3.3 - 3.6 1 1 Total 200 100
> 3.6 2 1
Source: Field Survey. Total 150 100
Source: Field Survey.
235
Figure 4.2: Sales per Employee (Rs. in Lakhs)
Hypothesis 13
One sample Kolomogorov-smirnov test shows that the metric, sales per
employee of kirana stores (p<0.04) and organised retail outlets (p<0.03) were not
normally distributed (Table 4.43).
Table 4.43: One-Sample Kolmogorov-Smirnov Test (Sales per Employee)
Sales per Employee
Kirana Stores Organised Retail Outlets
N 200 150
Normal Parametersa
Mean 58303 124299
Std. Deviation 40358 68654
Most Extreme Differences
Absolute 0.099 0.118
Positive 0.099 0.118
Negative -0.075 -0.076
Kolmogorov-Smirnov Z 1.4 1.448 Asymp. Sig. (2-tailed) 0.04 0.03
a. Test distribution is Normal.
236
So, Wilcoxon-Mann-Whitney independent two sample test is used to find the
independence of the two populations.
Table 4.44: Mann-Whitney Test (Sales per Employee)
Kira_or_Organ N Mean Rank Sum of Ranks
Sales_Emp
Organised Retail Outlets 150 240.89 36134.00
Kirana Stores 200 126.46 25291.00
Total 350
Table 4.45: Test Statisticsa (Sales per Employee)
Sales_Emp
Mann-Whitney U 5191.000
Wilcoxon W 25291.000
Z -10.474
Asymp. Sig. (2-tailed) .000
a. Grouping Variable: Kira_or_Organ.
The result (Table 4.44 and Table 4.45) suggests that there is a
statistically significant difference between the underlying distributions of sales
per employee of organised retail outlets and the kirana stores (z = -10.474, p
< 0.05). The two sample independent t-test also confirms the same (t =
10.492, p<0.05) (Table 4.46 and Table 4.47).
Table 4.46: Group Statistics (Sales per Employee)
Group Statistics
Kira_or_Organ N Mean Std.
Deviation
Std. Error
Mean
Sales_Emp
Organised Retail
Outlets 150 124298.7654 68654.31821 5605.60161
Kirana Stores 200 58303.0200 40358.49528 2853.77657
237
Table 4.47: Equality of Means (Sales per Employee)
Levene's Test for Equality
of Variances t-test for Equality of Means
Sales_ Emp
F Sig. t df Sig. (2-
tailed)
Mean Difference
Std. Error Difference
95% Confidence Interval of the
Difference
Lower Upper
Equal variances assumed
16.45 0.00 11.25 348 0.000 65995.75 5866.09 54458.29 77533.2
Equal variances not assumed
10.492 224.93 0.00 65995.75 6290.22 53600.46 78391.03
iii. Comparison of Monthly Sales (Rs. in Lakhs)
Comparison of monthly sales of organised retail outlets and unorganised
kirana stores was done. Table 4.48 and Figure 4.3 show the information regarding
monthly sales. Also, non-parametric test was used to compare the same. The
result is as follows.
Table 4.48: Comparison of Monthly Sales (Rs. in Lakhs)
Organises Retail Outlets Kirana Stores
Amount (Rs.)
No. of Outlets
Percentage Amount
(Rs.) No. of Outlets
Percen-tage
<5 19 12.7 <0.25 21 10.5
5-10 23 15.3 0.25 - 0.50 14 7.0
10-15 43 28.7 0.50 - 0.75 25 12.5
15-20 40 26.7 0.75 - 1.0 0 0.0
20-25 6 4 1.0 - 1.25 37 18.5
25-30 9 6 1.25 - 1.50 0 0.0
30-35 6 4 1.50 - 1.75 48 24.0
35-40 3 2 1.75 - 2.0 0 0.0
>40 1 0.7 2.0 - 2.25 7 3.5
Total 150 100 2.25 - 2.50 25 12.5
Source: Field Survey.
2.5 -2.75 3 1.5
2.75 - 3.0 15 7.5
>3.0 5 2.5
Total 200 100.0
238
Figure 4.3 : Monthly Sales (Rs. in Lakhs)
Hypothesis 14
One sample Kolomogorov-smirnov test shows that the metric monthly sales
of organised retail outlets and the kirana stores were not normally distributed
(Table 4.49).
Table 4.49: One-Sample Kolmogorov-Smirnov Test (Monthly Sales)
Monthly Sales
Organised Retail Outlets kirana Stores
N 150 200
Normal Parametersa Mean 1.4777E6 146838.7500
Std. Deviation
8.93519E5 92592.35827
Most Extreme Differences
Absolute .146 .142
Positive .146 .142
Negative -.088 -.078
Kolmogorov-Smirnov Z 1.786 2.004
Asymp. Sig. (2-tailed) .003 .001
a. Test distribution is Normal.
239
The Wilcoxon-Mann-Whitney test is used to test the hypothesis
H0: The two samples come from same populations
H1: The two samples come from different populations
Table 4.50: Mann-Whitney Test (Monthly Sales)
Ranks
Kira_or_Organ N Mean
Rank
Sum of
Ranks
Month_Sales
Organised Retail
Outlets 150 275.09 41263.00
Kirana Stores 200 100.81 20162.00
Total 350
Table 4.51: Test Statistics (Monthly Sales)
Test Statisticsa
Month_Sales
Mann-Whitney U 62.000
Wilcoxon W 20162.000
Z -15.972
Asymp. Sig. (2-tailed) .000
a. Grouping Variable: Kira_or_Organ
The result (Table 4.50 and Table 4.51) suggests that there is statistically
significant difference between the underlying distributions of the monthly sales of
organised retail outlets and the kirana stores (z = -15.972, p < 0.05).
The two sample independent t-test also proves the same (t = 18.17, p<0.05)
(Table 4.52 and Table 4.53).
240
Table 4.52: Independent t-Test (Monthly Sales)
Group Statistics
Kira_or_Organ N Mean Std.
Deviation
Std. Error
Mean
Month_Sales
Organised Retail
Outlets 150 1.4777E6 8.93519E5 72955.54046
Kirana Stores 200 146838.7500 92592.35827 6547.26844
Table 4.53: Test for Equality of Means (Monthly Sales)
Monthly_Sales
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
Equal variances assumed
147.387 0.000 20.925 348 0.000 1.33E+06 63602.33 1.21E+06 1.46E+06
Equal variances not assumed
18.17 151.402 0.00 1.33E+06 73248.74 1.19E+06 1.48E+06
CLUSTER ANALYSIS
Cluster analysis100 is a collection of statistical methods, which identifies
groups of samples that behave similarly or show similar characteristics. In
common parlance it is also called look-a-like groups. The simplest mechanism is
to partition the samples using measurements that capture similarity or distance
between samples. In this way, clusters and groups are interchangeable words.
Often in market research studies, cluster analysis is also referred to as a
segmentation method.
100 George, Darren and Mallery, Paul, “SPSS for Windows – Step by Step”, Pearson Education, 8th Edition, 2009.
241
Analysis 1: An attempt is made to segment the organised outlets on the
basis of select parameters such as area (sq. ft.), number of customers, footfalls,
catchment area (sq. kms), monthly sales, expenditure and PBDIT. The cluster
analysis resulted in 3 distinct clusters. The first cluster of outlets is of small
category, the second one is of medium category and the third cluster of outlets is of
large category. The percentage of outlets in each cluster is 16%, 33% and 51%
respectively. They are as follows (Table 4.54).
Table 4.54: Clusters based on performance
Cluster 1 2 3
Area (Sq. ft.) 1,000-1,500 1,500-2,500 2,000-3,000
No. of Customers 1,000-2,000 1,000-2,000 1,500-3,000
Footfalls 100-200 100-300 200-400
Catchment Area (Sq. Kms) 2-3 2-4 2-4
Monthly Sales (Rs. in Lakhs)
1-15 5-20 5-40
Expenditure as percentage of Sales
51-80 70-90 61-90
PBDIT (in Rs.) Upto 3 lakhs Upto 4 Lakh 2 – 7 lakhs
No. of SKUs 1,500-3,000 2,000-3,000 2,000-5,000
No. of Outlets 24 49 77
Percentage 16 33 51
Analysis 2: Cluster analysis is used to segment the outlets on the basis of
time spent and amount purchased by customer groups. The analysis resulted in 3
clusters and they are as follows (Table 4.55). The first cluster consisted of outlets
where the customers in different groups spent less time and made less purchases.
This group constituted about one-fifth (19%) of the outlets surveyed. The second
cluster, a major cluster, constituted about half (50%) of the outlets. In this kind of
outlets, the customers spent more time and made more purchase. In the third kind
of outlets, the time spent and purchases made was maximum among the clusters.
242
Table 4.55: Clusters Based on Customer Groups
Cluster
Customers visit with Family
Customers visit with Friends
Customers visit Alone Outlets
Time Spent
Amt Purchased
Time Spent
Amt Purchased
Time Spent
Amt Purchased
Number Percent
age
1 21-30 500-1,000 11-20 250-500 <10 <250 29 19
2 31-40 750-1,500 21-30 500-750 11-20 250-500 77 51
3 41-50 1,000-1,500 31-40 500-1,000 21-30 500-1,000 44 29
Total 150 100
MATHEMATICAL MODELLING
A mathematical model is the set of equations which describe the behaviour
of the system. The majority of interacting systems in the real world are far too
complicated to model in their entirety. But even if a model describes just a part of
the reality it can be very useful for analysis and design — if it describes the
dominating dynamic properties of the system.101 Hence the first level of
compromise is to identify the most important parts of the system. These will be
included in the model, the rest will be excluded.
The second level of compromise concerns the amount of mathematical
manipulation which is worthwhile. Although mathematics has the potential to prove
general results, these results depend critically on the form of equations used. Small
changes in the structure of equations may require enormous changes in the
mathematical methods. Using computers to handle the model equations may never
lead to elegant results, but it is much more robust against alterations.102
101 Bender, E.A., An introduction to mathematical modelling. Wiley, 1978. 102 Cross, M. and Moscardini, A.O., Learning the art of mathematical modelling, Ellis Horwood Ltd. Chichester, 1985.
243
ORGANISED RETAIL OUTLETS – REGRESSION MODELS Model 1 – Multiple Regression for Determining the Monthly Sales.
Monthly Sales of an organised retail outlet depends on number of
customers, footfalls and catchment area.
Table 4.56: Correlations (Number of Customers, Footfalls and Catchment Area)
Area_ Sqm
Catch_ Area
foot_fall_ mid
SKU_ NUMBER
Mon_ Sales
No_of_ Customers
Area_Sqm
Pearson Correlation
1 .199* .367** .398** .508** .577**
Sig. (2-tailed) 0.014 0 0 0 0
N 150 150 150 150 150 150
Catch_Area
Pearson Correlation
.199* 1 .240** .254** .264** .179*
Sig. (2-tailed) 0.014 0.003 0.002 0.001 0.028
N 150 150 150 150 150 150
foot_fall_mid
Pearson Correlation
.367** .240** 1 .509** .395** .579**
Sig. (2-tailed) 0 0.003 0 0 0
N 150 150 150 150 150 150
SKU_NUMBER
Pearson Correlation
.398** .254** .509** 1 .475** .593**
Sig. (2-tailed) 0 0.002 0 0 0
N 150 150 150 150 150 150
Mon_Sales
Pearson Correlation
.508** .264** .395** .475** 1 .834**
Sig. (2-tailed) 0 0.001 0 0 0
N 150 150 150 150 150 150
No_of_Customers
Pearson Correlation
.577** .179* .579** .593** .834** 1
Sig. (2-tailed) 0 0.028 0 0 0
N 150 150 150 150 150 150
*. Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed).
244
Table 4.57: Correlations - Spearman's rho (Number of Customers, Footfalls and Catchment Area)
Area_ Sqm
Catch_ Area
foot_fall_ mid
SKU_ NUMBER
Mon_ Sales
No_of_ Customers
Area_Sqm
Correlation Coefficient
1 .285** .408** .377** .290** .405**
Sig. (2-tailed) . 0 0 0 0 0
N 150 150 150 150 150 150
Catch_Area
Correlation Coefficient
.285** 1 .306** .262** .182* 0.16
Sig. (2-tailed) 0 . 0 0.001 0.026 0.051
N 150 150 150 150 150 150
foot_fall_mid
Correlation Coefficient
.408** .306** 1 .434** .242** .453**
Sig. (2-tailed) 0 0 . 0 0.003 0
N 150 150 150 150 150 150
SKU_NUMBER
Correlation Coefficient
.377** .262** .434** 1 .342** .375**
Sig. (2-tailed) 0 0.001 0 . 0 0
N 150 150 150 150 150 150
Mon_Sales
Correlation Coefficient
.290** .182* .242** .342** 1 .812**
Sig. (2-tailed) 0 0.026 0.003 0 . 0
N 150 150 150 150 150 150
No_of_Customers
Correlation Coefficient
.405** 0.16 .453** .375** .812** 1
Sig. (2-tailed) 0 0.051 0 0 0 .
N 150 150 150 150 150 150
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
245
Table 4.58: Model 1 (Number of Customers, Footfalls and
Catchment Area) Model Summaryd
Model R R
Square
Adjusted R
Square
Std. Error of the
Estimate
1 .834a 0.696 0.693 4.95E+05
2 .842b 0.709 0.705 4.85E+05
3 .852c 0.726 0.72 4.73E+05 a. Predictors: (Constant), No_of_Customers b. Predictors: (Constant), No_of_Customers, Catch_Area c. Predictors: (Constant), No_of_Customers, Catch_Area, foot_fall_mid d. Dependent Variable: Mon_Sales
Table 4.59: Coefficientsa (Number of Customers, Footfalls and Catchment Area)
Model
Unstandardized Coefficients Std.
Error
Standardized Coefficients
t Sig. B Beta
1
(Constant) -974328.6 139334 -6.993 0
No_of_Customers 1264.218 68.752 0.834 18.388 0
2
(Constant) -1.31E+06 188232 -6.973 0
No_of_Customers 1232.133 68.546 0.813 17.975 0
Catch_Area 157249.67 60176 0.118 2.613 0.01 3
(Constant) -1.26E+06 184218 -6.844 0
No_of_Customers 1366.941 80.696 0.902 16.939 0
Catch_Area 187397.37 59498 0.141 3.15 0.002
foot_fall_mid -2121.63 712.88 -0.161 -2.976 0.003
a. Dependent Variable: Mon_Sales
246
Table 4.60: Excluded Variablesd (Number of Customers,
Footfalls and Catchment Area)
Model Beta In T Sig. Partial
Correlation Collinearity Statistics
1 Tolerance
Area_Sqm .041a .736 .463 .061 .667
Catch_Area .118a 2.613 .010 .211 .968
foot_fall_mid -.132a -2.404 .017 -.195 .665
SKU_NUMBER -.030a -.540 .590 -.044 .648
2
Area_Sqm .024b .442 .659 .037 .658
foot_fall_mid -.161b -2.976 .003 -.239 .645
SKU_NUMBER -.060b -1.060 .291 -.087 .626
3
Area_Sqm .029c .540 .590 .045 .657
SKU_NUMBER -.024c -.420 .675 -.035 .593
a. Predictors in the Model: (Constant), No_of_Customers
b. Predictors in the Model: (Constant), No_of_Customers, Catch_Area
c. Predictors in the Model: (Constant), No_of_Customers, Catch_Area, foot_fall_mid
d. Dependent Variable: Mon_Sales
The regression analysis using ‘forward entry’ method shows that the sales of
an outlet depend on number of customers, footfalls and catchment area (Table
4.56 to Table 4.60).
The model was able to explain 72% of the variance in ‘monthly sales’ with
the three independent variables namely number of customers, footfalls and
catchment area.
Monthly Sales = 1377 x (Number of customers) + 1,87,397 x (Catchment Area) – 2,122 x (footfalls) -12,61,000
247
Model 2 – Determining PBDIT using multiple regression
PBDIT for an organised outlet is estimated using multiple linear regression
and the model is as follows.
Table 4.61: Correlation (Estimation of PBDIT)
Sales Per_Exp PBDIT
Sales
Pearson Correlation 1 0.122 .918**
Sig. (2-tailed) 0.137 0
N 150 150 150
Per_Exp
Pearson Correlation 0.122 1 -.237**
Sig. (2-tailed) 0.137 0.003
N 150 150 150
PBDIT
Pearson Correlation .918** -.237** 1
Sig. (2-tailed) 0 0.003
N 150 150 150
**. Correlation is significant at the 0.01 level (2-tailed).
Table 4.62: Model Summaryc (Estimation of PBDIT)
Model R R Square Adjusted R Square Std. Error of the Estimate
1 .918a .842 .841 83208.489
2 .983b .966 .966 38573.059
a. Predictors: (Constant), Sales, b. Predictors: (Constant), Sales, Per_Exp c. Dependent Variable: PBDIT.
248
Table 4.63: ANOVAc (Estimation of PBDIT)
Model Sum of Squares Df Mean Square F Sig.
1
Regression 5.479E12 1 5.479E12 791.349 .000a
Residual 1.025E12 148 6.924E9
Total 6.504E12 149
2
Regression 6.285E12 2 3.143E12 2.112E3 .000b
Residual 2.187E11 147 1.488E9
Total 6.504E12 149
a. Predictors: (Constant), Sales, b. Predictors: (Constant), Sales, Per_Exp, c. Dependent Variable: PBDIT.
Table 4.64: Coefficientsa (Estimation of PBDIT)
Model
Unstandardized
Coefficients
Standardized
Coefficients t Sig.
B Std. Error Beta
1 (Constant) 46042.109 13162.698 3.498 .001
Sales .215 .008 .918 28.131 .000
2
(Constant) 862288.39
5 35597.418 24.223 .000
Sales .225 .004 .961 63.070 .000
Per_Exp -11092.597 476.601 -.355 -23.274 .000
a. Dependent Variable: PBDIT.
In this model, 97% of the variance in PBDIT is explained by monthly sales
and operating expenses in percentage (Table 4.61 to Table 4.64).
The model is
PBDIT = 0.225 (Sales) – 11,093 (Percentage Expenditure) + 8,62,288
249
Model 3 – Multiple linear regression estimation for determining the sales of a
kirana store.
The analysis showed that the averages sales for a kirana store is dependent
on number of footfalls per day, average sales per day and the time spent by a
customer.
To confirm this proposition, the regression analysis using stepwise and
backward regression methods were used. The analysis corroborates the earlier
conclusion that the average sales of a kirana store is dependent on number of
footfalls per day, average sales per day and the time spent by a customer (Table
4.65 to Table 4.69).
The multiple linear regression equation for determining the average sales of
a kirana store is
Table 4.65: Variables Entered/Removedb (Average Sales of a kirana Store)
Model Variables Entered Variables Removed Method
1 Avg_time_spent, Aver_cust, Avg_sales, Size, NumCusta
. Enter
a. All requested variables entered. b. Dependent Variable: Avg_sale_store.
Table 4.66: Model Summary (Average Sales of a kirana Store)
Model R R
Square Adjusted R
Square Std. Error of the Estimate
1 .951a .904 .901 969.928
a. Predictors: (Constant), Avg_time_spent, Aver_cust, Avg_sales, Size, NumCust
Averages Sales = 48.976 x Aver_cust + 70.746 x Avg_sales + 41.416 x Avg_time_spent -3584.657
250
Table 4.67: ANOVAb (Average Sales of a kirana Store)
Model Sum of Squares df Mean Square F Sig.
1
Regression 1.713E9 5 3.426E8 364.207 .000a
Residual 1.825E8 194 940759.419
Total 1.896E9 199
a. Predictors: (Constant), Avg_time_spent, Aver_cust, Avg_sales, Size, NumCust b. Dependent Variable: Avg_sale_store.
Table 4.68: Coefficientsa (Average Sales of a kirana Store)
Model
Unstandardized Coefficients
Stand-ardized
Coefficients t Sig.
95% Confidence Interval for B
B Std. Error
Beta Lower Bound
Upper Bound
1
(Constant) -3596.498 238.631 -15.071 .000 -4067.142 -3125.853
Size -.287 .360 -.023 -.798 .426 -.996 .422
NumCust .401 .987 .012 .406 .685 -1.546 2.347
Aver_cust 48.262 3.057 .433 15.789 .000 42.233 54.290
Avg_sales 70.632 2.343 .719 30.142 .000 66.010 75.253
Avg_time_spent
49.916 21.550 .065 2.316 .022 7.414 92.418
a. Dependent Variable: Avg_sale_store.
Table 4.69: Coefficient Correlationsa (Average Sales of a kirana Store)
Model Avg_time_
spent Aver_ cust
Avg_ sales
Size NumCust
1
Correlations
Avg_time_spent 1.000 -.073 -.095 -.534 -.027
Aver_cust -.073 1.000 -.100 .207 -.532
Avg_sales -.095 -.100 1.000 .027 -.192
Size -.534 .207 .027 1.000 -.345
NumCust -.027 -.532 -.192 -.345 1.000
Covariances
Avg_time_spent 464.397 -4.824 -4.801 -4.135 -.581
Aver_cust -4.824 9.344 -.716 .228 -1.605
Avg_sales -4.801 -.716 5.491 .023 -.445
Size -4.135 .228 .023 .129 -.123
NumCust -.581 -1.605 -.445 -.123 .974
a. Dependent Variable: Avg_sale_store .
In this model 90% of the variance in average sale/day of a kirana store is
explained by footfalls per day, number of customers and average time spent by a
customer.
��������������������