SlideShare a Scribd company logo
1 of 23
Download to read offline
The Effect Of Consumption Pattern On Variety-Seeking Behavior Of
Customer
Sagar Phull
Northeastern University
Author Note
Sagar Phull, MS Engineering Management, Department of Industrial Engineering,
Northeastern University
INDEX
CONTENTS PAGE
NO.
1. Introduction 1
2. The experiment design 2
3. Method 2
4. Procedure 3
4.1 Sequential choices condition 3
4.2 Simultaneous choices for sequential consumption condition 3
5. Factorial design 4
6. Factors and interactions analysis 5
7. Factors and interactions screening 7
8. Fitted Model 10
8.1.Regression Model 10
8.2.Check for the presence of higher order terms 11
8.3.Analysis of the fitted model 11
9. Multiple Linear Regression 12
10. Best Fit model 14
11. Test for assumptions 16
12. Detecting outliers 18
13. Conclusion 20
14. Reference List 20
Abstract
Possibly the most challenging concept in marketing deals with understanding buyer’s selection
pattern. Such knowledge is critical for marketers since having a strong understanding of buyer
behavior will help shed light on what is important to the customer and marketers can create marketing
programs that they believe will be of interest to customers. The evoked set refers to the number of
alternatives that are considered by consumers during the buying process. Sometimes also known as
consideration, this set tends to be small relative to the total number of options available. Marketing
organization try to increase the likelihood that their brand is part of the consumer's evoked set.
Consumers evaluate alternatives in terms of the functional and psychological benefits that they offer.
The marketing organization needs to understand what benefits consumers are seeking and therefore
which attributes are most important in terms of making a decision. One such attribute is the
consumption period of the product and specifically in the case of food products. Consumption of
products often is separated from the decision to buy those products. Hence, when making a purchase
decision, consumers must predict their preferences at the time of consumption. The decision is
complicated further if consumers want to avoid going to the store before each consumption occasion
and decide to buy several items in a category for a number of occasions. For example, in one shopping
trip a consumer might purchase a week's supply of yogurt. The study here examines the strategies
consumers use when making multiple purchases in a product category for future consumption. The
behavior of consumers who make multiple purchases in a product class for several consumption
occasions is compared with that of consumers who purchase one item at a time before each
consumption occasion.
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
1 | P a g e
1. Introduction
The proposition examined here is that consumers making multiple purchases for several
consumption occasions, who are uncertain about their preferences, tend to select a greater
variety of items than consumers making purchases sequentially. Variety seeking in this case
reflects the uncertainty about the items that would be most preferred, diminishes the risk
associated with changing preferences, and reduces the time and effort needed to reach a
decision. This proposition leads to two hypotheses that were tested, the findings are reported
and their implications are discussed.
H0: Overall preferences for alternatives are stronger predictors of choice among consumers
making choices sequentially than among consumers making simultaneous choices for
sequential consumption.
H1: Consumers who simultaneously choose multiple items in a category for sequential
consumption are more likely to choose different items than consumers who sequentially make
the same number of choices.
The hypothesis can be tested by examining the effect of purchase quantity, with the consumption
schedule held constant, on variety seeking behavior. Specifically, in one condition (referred to
hereafter as the sequential choices condition), a single choice from the same product set is made
for immediate consumption. This condition is contrasted with a second one (referred to hereafter
as the simultaneous choices for sequential consumption condition), in which consumers make
multiple choices simultaneously, expecting to consume the selected products sequentially, one
in each consumption period.
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
2 | P a g e
2. The experiment design
There are several product categories, varieties and brands fit for this study. In addition,
conducting this study in real environment with a large population would require considerable
amount of funding and resources. In order to conduct this study and save those resources, a
paper-and-pencil task was used to test H1 in several product categories on college going students.
3. Method
Subjects: The subjects were 67 undergraduate students enrolled in an introductory marketing
course. Participation in the experiment was a course requirement.
Product categories: Seven product categories were selected for the study. In all of these
categories, consumers often make multiple purchases within a short period of time. In addition,
these products typically are consumed completely in one consumption occasion. The latter
criterion allowed for an unconfounded test of the hypotheses. That is, if consumption of a
product (e.g., a car, a box of cereal) is temporally extended (i.e., extends over several
consumption periods), the difference between making multiple choices that will be received at
one time and making multiple choices that will be received over time is less clear.
Product alternatives (variety): Within each category, different product alternatives such as
different yogurt flavors or snacks were listed, and subjects were instructed to indicate the
option(s) they would select. H (high) variety indicates that three different alternatives were
selected within the product category. M (medium) variety indicates that one alternative was
chosen twice, plus an additional alternative within the product category. L (low) variety indicates
that same alternative was selected three times.
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
3 | P a g e
Conditions: Two conditions were used, subjects in both conditions were told to imagine they
were going to the supermarket with a shopping list that included eight items(Product categories):
yogurt, bread, soft drink, canned vegetable, milk, snack, fruit, and a can of soup (milk was
included to test for random choice behavior, as explained subsequently).
4. Procedure
4.1 Sequential choices condition
In the sequential choices condition, subjects were told to assume they were going to the
supermarket to do their daily shopping, and in each category they intended to buy only one item.
After making choices in all categories, subjects were told to assume they had consumed the
selected alternatives, and on the following day they again were going to the supermarket. The
shopping list and alternatives in each category were identical to those for the first shopping day.
4.2 Simultaneous choices for sequential consumption condition
In the simultaneous choices for sequential consumption condition, subjects were told to assume
they were going to the supermarket to do their shopping for the next three days; in each category
they intended to buy three items for the next three days such that only one item in each category
would be consumed on each day. Finally, they were instructed to enter next to each item the
number of units of that item (if greater than zero) that they would buy. Because of the
hypothetical nature of the choice task, a test for random choice behavior was included. One of
the eight categories, milk, was selected because subjects in all conditions were expected to select
the same type of milk consistently (skim, low fat, or regular). Indeed, between 94% and 97% of
the subjects in all three conditions consistently selected the same type of milk. Therefore, in all
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
4 | P a g e
conditions, subjects appear to have taken the choice task seriously, and any variety-seeking
behavior observed is likely to have been intentional rather than the result of random choice
behavior. Following table1 shows the observations made by following the above procedure:
condition
simultaneous choices/ or
sequential consumption (1)
sequential choices (0)
Product No. variety outcomes H (3) M (2) L (1) H (3) M (2) L (1)
1 Yogurt 64 24 12 44 29 27
2 Bread/bagel/rolls 76 24 0 32 45 23
3
Canned
vegetables
53 41 6 35 24 41
4 Fruit 73 24 3 59 21 21
5 Snack 75 16 9 30 49 21
6 Soft drink/juice 46 30 24 29 35 35
7 Can of soup 44 44 12 38 47 15
Total 62 29 9 38 36 26
5. Factorial design
The study experiment basically utilizes 3 variables (factors) viz. conditions
(Simultaneous/sequential choice), Product categories (7 products) and Product variety (H, M and
L) and the response variable is number of subjects falling in each product-variety-condition
combination. To test the hypothesis, first we need to observe the effects of each of the 3
variables on the response variable and determine whether they actually effect the response.
table1
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
5 | P a g e
6. Factors and interactions analysis
The following table2 describes the above mentioned product-variety-condition combination and
number of subjects falling in each of those combinations.
subject condition variety product subject condition variety product
64
simultaneous
choice
High p1 44
sequential
choice
High p1
76
simultaneous
choice
High p2 32
sequential
choice
High p2
53
simultaneous
choice
High p3 35
sequential
choice
High p3
73
simultaneous
choice
High p4 59
sequential
choice
High p4
75
simultaneous
choice
High p5 30
sequential
choice
High p5
46
simultaneous
choice
High p6 29
sequential
choice
High p6
44
simultaneous
choice
High p7 38
sequential
choice
High p7
24
simultaneous
choice
Medium p1 29
sequential
choice
Medium p1
24
simultaneous
choice
Medium p2 45
sequential
choice
Medium p2
41
simultaneous
choice
Medium p3 24
sequential
choice
Medium p3
24
simultaneous
choice
Medium p4 21
sequential
choice
Medium p4
16
simultaneous
choice
Medium p5 49
sequential
choice
Medium p5
30
simultaneous
choice
Medium p6 35
sequential
choice
Medium p6
44
simultaneous
choice
Medium p7 47
sequential
choice
Medium p7
12
simultaneous
choice
Low p1 27
sequential
choice
Low p1
0
simultaneous
choice
Low p2 23
sequential
choice
Low p2
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
6 | P a g e
6
simultaneous
choice
Low p3 41
sequential
choice
Low p3
3
simultaneous
choice
Low p4 21
sequential
choice
Low p4
9
simultaneous
choice
Low p5 21
sequential
choice
Low p5
24
simultaneous
choice
Low p6 35
sequential
choice
Low p6
12
simultaneous
choice
Low p7 15
sequential
choice
Low p7
Here p1, p2, p3…., p7 correspond to 7 products listed in table 1. To determine the significance of
each factor and interactions (if exist) among those factors upon the response, an ANOVA test was
conducted. The following table3 shows the output of the ANOVA test.
As we can see from the ANOVA table, the variety*condition interaction and variety variable are
significant at 0.05 level of significance. Since the interaction between variety and condition
variable is significant, we can conclude that the condition variable’s significance has been masked
or leveled out by the presence of variety variable’s presence. A better understanding can be
achieved by observing the interaction plot. Following figure1 shows all the 3 factors plotted
against each other.
table2
table3
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
7 | P a g e
The graph1 from figure1 clearly shows that as the condition changes from simultaneous choice
to sequential choice, the number of subjects opting for higher variety decreases, whereas the
number of people opting for low variety increases, hence interaction between condition and
variety exists. Further the graph 5 shows that condition in independent of product and same can
also be said about variety from graph 6 as the slopes are almost parallel.
7. Factors and interactions screening
A better understanding of the interactions can be gained by coding the data of table1 according
to the levels of each variables. The x𝑖𝑖 corresponding to each factor level is mentioned in
parenthesis, example H (1) and simultaneous choice (1).Following formula is used to code each
factor input:
x′𝑖𝑖 =
x𝑖𝑖 −
x 𝐻𝐻 + x𝐿𝐿
2
x 𝐻𝐻 − x𝐿𝐿
2
figure1
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
8 | P a g e
Where x𝑖𝑖 is the original factor value, x 𝐻𝐻 is the highest value of that factor and x𝐿𝐿is the lowest
value. The coding results in the following table4.
subject condition variety product subject condition variety product
64 1 1 -3 44 -1 1 -3
76 1 1 -2 32 -1 1 -2
53 1 1 -1 35 -1 1 -1
73 1 1 0 59 -1 1 0
75 1 1 1 30 -1 1 1
46 1 1 2 29 -1 1 2
44 1 1 3 38 -1 1 3
24 1 0 -3 29 -1 0 -3
24 1 0 -2 45 -1 0 -2
41 1 0 -1 24 -1 0 -1
24 1 0 0 21 -1 0 0
16 1 0 1 49 -1 0 1
30 1 0 2 35 -1 0 2
44 1 0 3 47 -1 0 3
12 1 -1 -3 27 -1 -1 -3
0 1 -1 -2 23 -1 -1 -2
6 1 -1 -1 41 -1 -1 -1
3 1 -1 0 21 -1 -1 0
9 1 -1 1 21 -1 -1 1
24 1 -1 2 35 -1 -1 2
12 1 -1 3 15 -1 -1 3
Figure2 shows a surface plot of condition variety and subject plotted using the coded data.
It is also clear from the figure2 that the
number of subjects is the highest for
simultaneous choice condition (1) with High
variety (1) and the lowest for simultaneous
choice condition (1) with low variety (-1).
table4
figure2
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
9 | P a g e
The above discussion, tests, figures and graphs clearly show that the factors ‘variety’, ‘condition’
and interaction ‘variety*condition’ only are significant and should be considered for the
hypothesis testing. The tests actually superimpose the hypothesis that variety seeking behavior
of consumer is independent of the product category in a particular product family, in our case
food products.
To further verify this claim a normal quantile-quantile plot of various factors and their interaction
effects is plotted. The following figure3 shows the plot.
The graph clearly shows that product nor its interactions are significant and hence subject’s
response should be considered independent of the same.
figure3
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
10 | P a g e
8. Fitted Model
As a result, a model showing the relationship between number of subjects for a particular
condition-variety combination will be independent of product category.
8.1 Regression Model
A regression analysis of the coded data results in the following model and ANOVA test recorded
in the table5.
Linear regression model:
Subject = 1 + 10.036*condition*variety
The R2 value of the model being 0.7 means that the fitted model accounts for 70% of the subject’s
variety seeking behavior. A behavioral study, dealing in data impacted by variability in human
behavior, can consider a model with R2 as large as 70% to be a good fit for the data.
table5
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
11 | P a g e
8.2 Check for the presence of higher order terms
To cross-check if the model could be containing second order terms, a lack of fit test was
performed on the coded data, the results of the ANOVA test for lack-of-fit is shown in the
following table6
The ANOVA test in table6 clearly shows that the lack-of-fit is insignificant at 0.05 level of
significance and as a result higher order term of variety2 is insignificant and is not present in the
model.
8.3 Analysis of the fitted model
As a result the best fit model for the given study data is Subject = 1 + 10.036*condition*variety.
The model clearly supports the alternate hypothesis that as condition increases form sequential
choice (-1) to simultaneous choice (+1) along with variety from Low (-1) to High (-1) the number
of subjects increases. This clearly proves the alternate hypothesis that consumers who
simultaneously choose multiple items in a category for sequential consumption are more likely
table6
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
12 | P a g e
to choose different items (higher variety) than consumers who sequentially make the same
number of choices. A more formal multiple linear regression analysis for the uncoded data is
performed in the following sections of the project report along with tests for various assumptions
made for the data of the model.
9. Multiple Linear Regression
Concluding from the design of experiment of the project study, the significant factors for the
model are ‘variety’, ‘condition’ and interaction of ‘variety*condition’. The variables ‘variety’ and
‘condition’ are categorical variables as they divide the data into fixed categories. To represent
these variables in numerical form for uncoded regression model, we need to assign
numerical values to different levels of each factor. The following table7 gives the numerical
values for categorical variables
Assigning these values to each condition-variety and condition*variety interaction combination
for subjects and depicting response ‘subject’ by y we get the following dataset projected in table8
x w1 w2 xw1 xw2 y x w1 w2 xw1 xw2 y
1 1 0 1 0 64 0 1 0 0 0 59
1 0 1 0 1 24 0 0 1 0 0 21
1 0 0 0 0 12 0 0 0 0 0 21
0 1 0 0 0 44 1 1 0 1 0 75
0 0 1 0 0 29 1 0 1 0 1 16
0 0 0 0 0 27 1 0 0 0 0 9
1 1 0 1 0 76 0 1 0 0 0 30
variety outcomes w1 w2 Condition x
High (1) 1 0 simultaneous choices (1) 1
Medium (2) 0 1 sequential choices (0) 0
Low (3) 0 0
table7
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
13 | P a g e
1 0 1 0 1 24 0 0 1 0 0 49
1 0 0 0 0 0 0 0 0 0 0 21
0 1 0 0 0 32 1 1 0 1 0 46
0 0 1 0 0 45 1 0 1 0 1 30
0 0 0 0 0 23 1 0 0 0 0 24
1 1 0 1 0 53 0 1 0 0 0 29
1 0 1 0 1 41 0 0 1 0 0 35
1 0 0 0 0 6 0 0 0 0 0 35
0 1 0 0 0 35 1 1 0 1 0 44
0 0 1 0 0 24 1 0 1 0 1 44
0 0 0 0 0 41 1 0 0 0 0 12
1 1 0 1 0 73 0 1 0 0 0 38
1 0 1 0 1 24 0 0 1 0 0 47
1 0 0 0 0 3 0 0 0 0 0 15
Here, xw1 and xw2 are the condition*variety interaction terms. Performing multiple linear
regression for this dataset we get the following model and ANOVA test for the same depicted
in table9
table8
t a b l e 9
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
14 | P a g e
The fitted model is y = 26.13 - 16.714x + 12w1 + 9.5714w2 + 40.143xw1 + 10xw2. The ANOVA
test clearly shows that only terms ‘x’, ‘w1’, and ‘xw1’ are significant.
10. Best Fit model
To further verify the above mentioned claim and design the best fit model, Cp and PRESS
statistics were calculated and the following table10 displays models containing various terms
along with the values of their Cp and PRESS statistic.
PRESS is errors in prediction whereas Cp is a measure of model being under-fitted or over-fitted.
The lowest value of both PRESS and Cp is preferred; PRESS for obvious reason; while Cp ≈number
of model parameters which in our case would be four – ‘xw1’, ‘w2’, ‘w1’, x. After removing
‘xw2’ we get the following linear regression model along with ANOVA test in table11:
table10
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
15 | P a g e
y = 23.643 – 11.714x + 14.5w1 + 14.571w2 + 35.143xw1
There is no significant change in the R2 values of previous and current model and hence it is a
possibility that the model is still over-fitted. To further fit the model a stepwise multiple
regression analysis approach was adopted, the resulting model and ANOVA test is given in
the following table12.
table11
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
16 | P a g e
The stepwise multiple linear regression model is
y = 23.643 + 14.571w2 + 35.143xw1
The model shows a reduction of 1.6% in R2 from the original model but with only 2 variable and
1 constant term and hence is the best fit model for the given uncoded dataset. The model R2
value is 70.4% which is almost same as of coded model from design of experiment analysis.
11. Test for assumptions
To perform various analysis like ANOVA while conducting design of experiment and multiple
regression analysis, following assumptions were made,
1. The random error of disturbance, has constant variance also known as the homogeneous
variance assumption.
2. The random error of disturbance is normally distributed.
table12
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
17 | P a g e
Homogeneous variance is an important assumption made in regression analysis. Violations can
often be detected through the appearance of the residual plot. Following figure4 shows the graph
of residuals against fitted values of the model.
The residuals (although not completely) show a healthy random scatter around a value of zero
and hence the assumption of homogeneous variance holds well.
To gain some type of idea regarding the normal error assumption, a normal probability plot of
the residuals was generated. In this is the type of plot the horizontal axis represents the empirical
normal distribution function on a scale that produces a straight-line plot when plotted against
the residuals. Following Figure5 shows the normal probability plot of the residuals.
figure4
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
18 | P a g e
The normal probability plot does reflect the straight-line appearance that one would like to see
for the normality assumption to hold good.
12. Detecting outliers
The model errors, often shed light on the presence of “suspect” data points where the above
assumptions might be violated, an outlier is a data point where there is a deviation from the
above assumptions. An unusually high value of R-Student statistics highlights data points where
the error of fit is larger than what is expected by chance. The following table13 shows the
calculated values of studentized residuals (ri) and R-studentized residuals (ti) for the fitted model.
Obs. yi ŷi yi − ŷi ri ti
1 64 61.57143 2.428571 0.244767 0.241632
2 24 26.5 -2.5 -0.24688 -0.24372
figure5
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
19 | P a g e
3 12 11.92857 0.071429 0.007054 0.006958
4 44 38.14286 5.857143 0.59032 0.58505
5 29 38.21429 -9.21429 -0.90991 -0.90774
6 27 23.64286 3.357143 0.331518 0.327494
7 76 61.57143 14.42857 1.454203 1.477254
8 24 26.5 -2.5 -0.24688 -0.24372
9 0 11.92857 -11.9286 -1.17795 -1.18434
10 32 38.14286 -6.14286 -0.61912 -0.61388
11 45 38.21429 6.785714 0.67009 0.66502
12 23 23.64286 -0.64286 -0.06348 -0.06262
13 53 61.57143 -8.57143 -0.86388 -0.86085
14 41 26.5 14.5 1.431876 1.453231
15 6 11.92857 -5.92857 -0.58545 -0.58017
16 35 38.14286 -3.14286 -0.31676 -0.31287
17 24 38.21429 -14.2143 -1.40366 -1.42297
18 41 23.64286 17.35714 1.714019 1.762101
19 73 61.57143 11.42857 1.151844 1.157107
20 24 26.5 -2.5 -0.24688 -0.24372
21 3 11.92857 -8.92857 -0.8817 -0.87898
22 59 38.14286 20.85714 2.102115 2.209657
23 21 38.21429 -17.2143 -1.69991 -1.74636
24 21 23.64286 -2.64286 -0.26098 -0.25767
25 75 61.57143 13.42857 1.353417 1.369327
26 16 26.5 -10.5 -1.03688 -1.03796
27 9 11.92857 -2.92857 -0.2892 -0.28558
28 30 38.14286 -8.14286 -0.82069 -0.81699
29 49 38.21429 10.78571 1.06509 1.067084
30 21 23.64286 -2.64286 -0.26098 -0.25767
31 46 61.57143 -15.5714 -1.56939 -1.60228
32 30 26.5 3.5 0.345625 0.341474
33 24 11.92857 12.07143 1.192054 1.199086
34 29 38.14286 -9.14286 -0.92148 -0.91955
35 35 38.21429 -3.21429 -0.31741 -0.31352
36 35 23.64286 11.35714 1.121518 1.125556
37 44 61.57143 -17.5714 -1.77096 -1.82597
38 44 26.5 17.5 1.728126 1.777872
39 12 11.92857 0.071429 0.007054 0.006958
40 38 38.14286 -0.14286 -0.0144 -0.0142
41 47 38.21429 8.785714 0.86759 0.864625
42 15 23.64286 -8.64286 -0.85348 -0.85028
table13
The Effect Of Consumption Pattern On Variety-
Seeking Behavior Of Customer
20 | P a g e
As we can see from the table and figure5 there is no data point with a significantly high value of
ti and hence the dataset is free from outliers.
13. Conclusion
The final fitted model of y = 23.643 + 14.571w2 + 35.143xw1 is the best fit model which does not
violate any assumptions. The model also satisfies the alternate hypothesis, consider the situation
of simultaneous choice (x = 1) along with high variety (w1 = 1, w2 = 0) the value of
y = 23.643 + 35.143 = 58.786
Which is the average value of subjects at simultaneous choice and high variety.
Now consider the situation of sequential choice (x = 0) along with high variety (w1 = 1, w2 = 0)
the value of y = 23.643
Which being average value of subjects at sequential choice and high variety is certainly quiet
lower than the above y value of 58.786. The same can also be shown for simultaneous choice and
low variety being lower than sequential choice and low variety and hence the null hypothesis is
rejected in favor of alternate hypothesis.
14. Reference List
• Ronald E. Walpole, R. H. M., Sharon L. Myers, Keying Ye. (2012). Probability & Statistics
for Engineers & Scientists (9th ed.): Prentice Hall.
• Simonson, I. (1990). The Effect of Purchase Quantity and Timing on Variety-Seeking
Behavior. Journal of Marketing Research, 27(2), 150-162. doi:10.2307/3172842.

More Related Content

Viewers also liked

Nbc training report on railway bearing(spherical bearing)
Nbc training report on railway bearing(spherical bearing)Nbc training report on railway bearing(spherical bearing)
Nbc training report on railway bearing(spherical bearing)Ashutosh Singh
 
Railway training report
Railway training reportRailway training report
Railway training reportDeepak kango
 
TRAINING REPORT FULL
TRAINING REPORT FULLTRAINING REPORT FULL
TRAINING REPORT FULLanish malan
 
indian railway training report
indian railway training reportindian railway training report
indian railway training reportravi kant
 
Training Report on Bridge Construction
Training Report on Bridge ConstructionTraining Report on Bridge Construction
Training Report on Bridge ConstructionMAHAVIR MEENA
 
Training report done on Bridge Construction
Training report done on Bridge ConstructionTraining report done on Bridge Construction
Training report done on Bridge ConstructionSukhdeep Jat
 
ANALYSIS AND DESIGN OF RAILWAY STEEL AND COCRETE BRIDGE & NON DESTRUCTIVE TE...
ANALYSIS AND DESIGN OF RAILWAY STEEL AND COCRETE BRIDGE &  NON DESTRUCTIVE TE...ANALYSIS AND DESIGN OF RAILWAY STEEL AND COCRETE BRIDGE &  NON DESTRUCTIVE TE...
ANALYSIS AND DESIGN OF RAILWAY STEEL AND COCRETE BRIDGE & NON DESTRUCTIVE TE...GLA University
 
Ne railway gorakhpur summer training report
Ne railway gorakhpur summer training reportNe railway gorakhpur summer training report
Ne railway gorakhpur summer training reportKrishna Yadav
 
Building structure project 1 report
Building structure project 1 reportBuilding structure project 1 report
Building structure project 1 reportAdelinetingg
 
RDSO training report -NAVIN DIXIT
RDSO training report -NAVIN DIXITRDSO training report -NAVIN DIXIT
RDSO training report -NAVIN DIXITNavin Dixit
 
3D Analysis Of Truss Bridges
3D Analysis Of Truss Bridges3D Analysis Of Truss Bridges
3D Analysis Of Truss Bridgesblovelace
 
Truss Bridge Report
Truss Bridge ReportTruss Bridge Report
Truss Bridge ReportYvonne Chin
 
Railway Training Report
Railway Training ReportRailway Training Report
Railway Training ReportVishal Singh
 
project report on truss bridge
project report on truss bridgeproject report on truss bridge
project report on truss bridgerajdutt1111
 
Industrial Summer Training Report at Construction Site of CPWD
Industrial Summer Training Report at Construction Site of CPWD Industrial Summer Training Report at Construction Site of CPWD
Industrial Summer Training Report at Construction Site of CPWD Alok Mishra
 

Viewers also liked (16)

Nbc training report on railway bearing(spherical bearing)
Nbc training report on railway bearing(spherical bearing)Nbc training report on railway bearing(spherical bearing)
Nbc training report on railway bearing(spherical bearing)
 
Railway training report
Railway training reportRailway training report
Railway training report
 
TRAINING REPORT FULL
TRAINING REPORT FULLTRAINING REPORT FULL
TRAINING REPORT FULL
 
indian railway training report
indian railway training reportindian railway training report
indian railway training report
 
Training Report on Bridge Construction
Training Report on Bridge ConstructionTraining Report on Bridge Construction
Training Report on Bridge Construction
 
Training report done on Bridge Construction
Training report done on Bridge ConstructionTraining report done on Bridge Construction
Training report done on Bridge Construction
 
Bridge report
Bridge reportBridge report
Bridge report
 
ANALYSIS AND DESIGN OF RAILWAY STEEL AND COCRETE BRIDGE & NON DESTRUCTIVE TE...
ANALYSIS AND DESIGN OF RAILWAY STEEL AND COCRETE BRIDGE &  NON DESTRUCTIVE TE...ANALYSIS AND DESIGN OF RAILWAY STEEL AND COCRETE BRIDGE &  NON DESTRUCTIVE TE...
ANALYSIS AND DESIGN OF RAILWAY STEEL AND COCRETE BRIDGE & NON DESTRUCTIVE TE...
 
Ne railway gorakhpur summer training report
Ne railway gorakhpur summer training reportNe railway gorakhpur summer training report
Ne railway gorakhpur summer training report
 
Building structure project 1 report
Building structure project 1 reportBuilding structure project 1 report
Building structure project 1 report
 
RDSO training report -NAVIN DIXIT
RDSO training report -NAVIN DIXITRDSO training report -NAVIN DIXIT
RDSO training report -NAVIN DIXIT
 
3D Analysis Of Truss Bridges
3D Analysis Of Truss Bridges3D Analysis Of Truss Bridges
3D Analysis Of Truss Bridges
 
Truss Bridge Report
Truss Bridge ReportTruss Bridge Report
Truss Bridge Report
 
Railway Training Report
Railway Training ReportRailway Training Report
Railway Training Report
 
project report on truss bridge
project report on truss bridgeproject report on truss bridge
project report on truss bridge
 
Industrial Summer Training Report at Construction Site of CPWD
Industrial Summer Training Report at Construction Site of CPWD Industrial Summer Training Report at Construction Site of CPWD
Industrial Summer Training Report at Construction Site of CPWD
 

Similar to project

A study on consumer behavior of aavin milk in bhel township
A study on consumer behavior of aavin milk in bhel townshipA study on consumer behavior of aavin milk in bhel township
A study on consumer behavior of aavin milk in bhel townshipdagaashutosh
 
A study on consumer behavior of aavin milk in bhel township
A study on consumer behavior of aavin milk in bhel townshipA study on consumer behavior of aavin milk in bhel township
A study on consumer behavior of aavin milk in bhel townshipdagaashutosh
 
Rate of satiation and limited availability of goods
Rate of satiation and limited availability of goodsRate of satiation and limited availability of goods
Rate of satiation and limited availability of goodsKishore Muppaneni
 
Capstone spiral binding (2)for pdf
Capstone spiral binding (2)for pdfCapstone spiral binding (2)for pdf
Capstone spiral binding (2)for pdfROHANDEFINED
 
CUET MA Economics book .pdf [Sample PDF]
 CUET MA Economics book .pdf [Sample PDF] CUET MA Economics book .pdf [Sample PDF]
CUET MA Economics book .pdf [Sample PDF]DIwakar Rajput
 
A study on consumer behavior of saras milk in hzl township
A study on consumer behavior of saras milk in hzl townshipA study on consumer behavior of saras milk in hzl township
A study on consumer behavior of saras milk in hzl townshipdagaashutosh
 
1Recycling of Products as a Marketing ProblemA.docx
1Recycling of Products as a Marketing ProblemA.docx1Recycling of Products as a Marketing ProblemA.docx
1Recycling of Products as a Marketing ProblemA.docxeugeniadean34240
 
consumer behavior.pdf
consumer behavior.pdfconsumer behavior.pdf
consumer behavior.pdfjudithpatnaan
 
761 2014 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 41 .docx
761 2014 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 41 .docx761 2014 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 41 .docx
761 2014 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 41 .docxevonnehoggarth79783
 
Managing changing attitudes of consumer on buying preferences a strategic eva...
Managing changing attitudes of consumer on buying preferences a strategic eva...Managing changing attitudes of consumer on buying preferences a strategic eva...
Managing changing attitudes of consumer on buying preferences a strategic eva...Dr. Juturu Viswanath
 
Benefit Hierarchy Analysis
Benefit Hierarchy AnalysisBenefit Hierarchy Analysis
Benefit Hierarchy AnalysisEfim Shvartsburg
 
MKT412ResearchReportFinal
MKT412ResearchReportFinalMKT412ResearchReportFinal
MKT412ResearchReportFinalHannah Dion
 
Operations DecisionECO550 Assignment 2Lydia L. BrooksRunning.docx
Operations DecisionECO550 Assignment 2Lydia L. BrooksRunning.docxOperations DecisionECO550 Assignment 2Lydia L. BrooksRunning.docx
Operations DecisionECO550 Assignment 2Lydia L. BrooksRunning.docxhopeaustin33688
 
Agricultural Economics.pdf
Agricultural Economics.pdfAgricultural Economics.pdf
Agricultural Economics.pdfAngelina Johnson
 
Project report-soap-market
Project report-soap-marketProject report-soap-market
Project report-soap-marketSafal Verma
 
Consumer behavior all material Prepared by karventhan
Consumer behavior all material Prepared by karventhan Consumer behavior all material Prepared by karventhan
Consumer behavior all material Prepared by karventhan karventhanps
 
A Case study on Consumer preference for ready to eat food products
A Case study on Consumer preference for ready to eat food productsA Case study on Consumer preference for ready to eat food products
A Case study on Consumer preference for ready to eat food productsSyed Sadath
 

Similar to project (20)

A study on consumer behavior of aavin milk in bhel township
A study on consumer behavior of aavin milk in bhel townshipA study on consumer behavior of aavin milk in bhel township
A study on consumer behavior of aavin milk in bhel township
 
A study on consumer behavior of aavin milk in bhel township
A study on consumer behavior of aavin milk in bhel townshipA study on consumer behavior of aavin milk in bhel township
A study on consumer behavior of aavin milk in bhel township
 
Rate of satiation and limited availability of goods
Rate of satiation and limited availability of goodsRate of satiation and limited availability of goods
Rate of satiation and limited availability of goods
 
Capstone spiral binding (2)for pdf
Capstone spiral binding (2)for pdfCapstone spiral binding (2)for pdf
Capstone spiral binding (2)for pdf
 
CUET MA Economics book .pdf [Sample PDF]
 CUET MA Economics book .pdf [Sample PDF] CUET MA Economics book .pdf [Sample PDF]
CUET MA Economics book .pdf [Sample PDF]
 
A study on consumer behavior of saras milk in hzl township
A study on consumer behavior of saras milk in hzl townshipA study on consumer behavior of saras milk in hzl township
A study on consumer behavior of saras milk in hzl township
 
1Recycling of Products as a Marketing ProblemA.docx
1Recycling of Products as a Marketing ProblemA.docx1Recycling of Products as a Marketing ProblemA.docx
1Recycling of Products as a Marketing ProblemA.docx
 
consumer behavior.pdf
consumer behavior.pdfconsumer behavior.pdf
consumer behavior.pdf
 
Retailers out of stock
Retailers out of stockRetailers out of stock
Retailers out of stock
 
761 2014 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 41 .docx
761 2014 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 41 .docx761 2014 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 41 .docx
761 2014 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 41 .docx
 
Managing changing attitudes of consumer on buying preferences a strategic eva...
Managing changing attitudes of consumer on buying preferences a strategic eva...Managing changing attitudes of consumer on buying preferences a strategic eva...
Managing changing attitudes of consumer on buying preferences a strategic eva...
 
Benefit Hierarchy Analysis
Benefit Hierarchy AnalysisBenefit Hierarchy Analysis
Benefit Hierarchy Analysis
 
MKT412ResearchReportFinal
MKT412ResearchReportFinalMKT412ResearchReportFinal
MKT412ResearchReportFinal
 
Operations DecisionECO550 Assignment 2Lydia L. BrooksRunning.docx
Operations DecisionECO550 Assignment 2Lydia L. BrooksRunning.docxOperations DecisionECO550 Assignment 2Lydia L. BrooksRunning.docx
Operations DecisionECO550 Assignment 2Lydia L. BrooksRunning.docx
 
Agricultural Economics.pdf
Agricultural Economics.pdfAgricultural Economics.pdf
Agricultural Economics.pdf
 
Project report-soap-market
Project report-soap-marketProject report-soap-market
Project report-soap-market
 
Zachary Brown - Forecasting Consumer Response to GMOs
Zachary Brown - Forecasting Consumer Response to GMOsZachary Brown - Forecasting Consumer Response to GMOs
Zachary Brown - Forecasting Consumer Response to GMOs
 
Review of literature
Review of literatureReview of literature
Review of literature
 
Consumer behavior all material Prepared by karventhan
Consumer behavior all material Prepared by karventhan Consumer behavior all material Prepared by karventhan
Consumer behavior all material Prepared by karventhan
 
A Case study on Consumer preference for ready to eat food products
A Case study on Consumer preference for ready to eat food productsA Case study on Consumer preference for ready to eat food products
A Case study on Consumer preference for ready to eat food products
 

project

  • 1. The Effect Of Consumption Pattern On Variety-Seeking Behavior Of Customer Sagar Phull Northeastern University Author Note Sagar Phull, MS Engineering Management, Department of Industrial Engineering, Northeastern University
  • 2. INDEX CONTENTS PAGE NO. 1. Introduction 1 2. The experiment design 2 3. Method 2 4. Procedure 3 4.1 Sequential choices condition 3 4.2 Simultaneous choices for sequential consumption condition 3 5. Factorial design 4 6. Factors and interactions analysis 5 7. Factors and interactions screening 7 8. Fitted Model 10 8.1.Regression Model 10 8.2.Check for the presence of higher order terms 11 8.3.Analysis of the fitted model 11 9. Multiple Linear Regression 12 10. Best Fit model 14 11. Test for assumptions 16 12. Detecting outliers 18 13. Conclusion 20 14. Reference List 20
  • 3. Abstract Possibly the most challenging concept in marketing deals with understanding buyer’s selection pattern. Such knowledge is critical for marketers since having a strong understanding of buyer behavior will help shed light on what is important to the customer and marketers can create marketing programs that they believe will be of interest to customers. The evoked set refers to the number of alternatives that are considered by consumers during the buying process. Sometimes also known as consideration, this set tends to be small relative to the total number of options available. Marketing organization try to increase the likelihood that their brand is part of the consumer's evoked set. Consumers evaluate alternatives in terms of the functional and psychological benefits that they offer. The marketing organization needs to understand what benefits consumers are seeking and therefore which attributes are most important in terms of making a decision. One such attribute is the consumption period of the product and specifically in the case of food products. Consumption of products often is separated from the decision to buy those products. Hence, when making a purchase decision, consumers must predict their preferences at the time of consumption. The decision is complicated further if consumers want to avoid going to the store before each consumption occasion and decide to buy several items in a category for a number of occasions. For example, in one shopping trip a consumer might purchase a week's supply of yogurt. The study here examines the strategies consumers use when making multiple purchases in a product category for future consumption. The behavior of consumers who make multiple purchases in a product class for several consumption occasions is compared with that of consumers who purchase one item at a time before each consumption occasion.
  • 4. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 1 | P a g e 1. Introduction The proposition examined here is that consumers making multiple purchases for several consumption occasions, who are uncertain about their preferences, tend to select a greater variety of items than consumers making purchases sequentially. Variety seeking in this case reflects the uncertainty about the items that would be most preferred, diminishes the risk associated with changing preferences, and reduces the time and effort needed to reach a decision. This proposition leads to two hypotheses that were tested, the findings are reported and their implications are discussed. H0: Overall preferences for alternatives are stronger predictors of choice among consumers making choices sequentially than among consumers making simultaneous choices for sequential consumption. H1: Consumers who simultaneously choose multiple items in a category for sequential consumption are more likely to choose different items than consumers who sequentially make the same number of choices. The hypothesis can be tested by examining the effect of purchase quantity, with the consumption schedule held constant, on variety seeking behavior. Specifically, in one condition (referred to hereafter as the sequential choices condition), a single choice from the same product set is made for immediate consumption. This condition is contrasted with a second one (referred to hereafter as the simultaneous choices for sequential consumption condition), in which consumers make multiple choices simultaneously, expecting to consume the selected products sequentially, one in each consumption period.
  • 5. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 2 | P a g e 2. The experiment design There are several product categories, varieties and brands fit for this study. In addition, conducting this study in real environment with a large population would require considerable amount of funding and resources. In order to conduct this study and save those resources, a paper-and-pencil task was used to test H1 in several product categories on college going students. 3. Method Subjects: The subjects were 67 undergraduate students enrolled in an introductory marketing course. Participation in the experiment was a course requirement. Product categories: Seven product categories were selected for the study. In all of these categories, consumers often make multiple purchases within a short period of time. In addition, these products typically are consumed completely in one consumption occasion. The latter criterion allowed for an unconfounded test of the hypotheses. That is, if consumption of a product (e.g., a car, a box of cereal) is temporally extended (i.e., extends over several consumption periods), the difference between making multiple choices that will be received at one time and making multiple choices that will be received over time is less clear. Product alternatives (variety): Within each category, different product alternatives such as different yogurt flavors or snacks were listed, and subjects were instructed to indicate the option(s) they would select. H (high) variety indicates that three different alternatives were selected within the product category. M (medium) variety indicates that one alternative was chosen twice, plus an additional alternative within the product category. L (low) variety indicates that same alternative was selected three times.
  • 6. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 3 | P a g e Conditions: Two conditions were used, subjects in both conditions were told to imagine they were going to the supermarket with a shopping list that included eight items(Product categories): yogurt, bread, soft drink, canned vegetable, milk, snack, fruit, and a can of soup (milk was included to test for random choice behavior, as explained subsequently). 4. Procedure 4.1 Sequential choices condition In the sequential choices condition, subjects were told to assume they were going to the supermarket to do their daily shopping, and in each category they intended to buy only one item. After making choices in all categories, subjects were told to assume they had consumed the selected alternatives, and on the following day they again were going to the supermarket. The shopping list and alternatives in each category were identical to those for the first shopping day. 4.2 Simultaneous choices for sequential consumption condition In the simultaneous choices for sequential consumption condition, subjects were told to assume they were going to the supermarket to do their shopping for the next three days; in each category they intended to buy three items for the next three days such that only one item in each category would be consumed on each day. Finally, they were instructed to enter next to each item the number of units of that item (if greater than zero) that they would buy. Because of the hypothetical nature of the choice task, a test for random choice behavior was included. One of the eight categories, milk, was selected because subjects in all conditions were expected to select the same type of milk consistently (skim, low fat, or regular). Indeed, between 94% and 97% of the subjects in all three conditions consistently selected the same type of milk. Therefore, in all
  • 7. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 4 | P a g e conditions, subjects appear to have taken the choice task seriously, and any variety-seeking behavior observed is likely to have been intentional rather than the result of random choice behavior. Following table1 shows the observations made by following the above procedure: condition simultaneous choices/ or sequential consumption (1) sequential choices (0) Product No. variety outcomes H (3) M (2) L (1) H (3) M (2) L (1) 1 Yogurt 64 24 12 44 29 27 2 Bread/bagel/rolls 76 24 0 32 45 23 3 Canned vegetables 53 41 6 35 24 41 4 Fruit 73 24 3 59 21 21 5 Snack 75 16 9 30 49 21 6 Soft drink/juice 46 30 24 29 35 35 7 Can of soup 44 44 12 38 47 15 Total 62 29 9 38 36 26 5. Factorial design The study experiment basically utilizes 3 variables (factors) viz. conditions (Simultaneous/sequential choice), Product categories (7 products) and Product variety (H, M and L) and the response variable is number of subjects falling in each product-variety-condition combination. To test the hypothesis, first we need to observe the effects of each of the 3 variables on the response variable and determine whether they actually effect the response. table1
  • 8. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 5 | P a g e 6. Factors and interactions analysis The following table2 describes the above mentioned product-variety-condition combination and number of subjects falling in each of those combinations. subject condition variety product subject condition variety product 64 simultaneous choice High p1 44 sequential choice High p1 76 simultaneous choice High p2 32 sequential choice High p2 53 simultaneous choice High p3 35 sequential choice High p3 73 simultaneous choice High p4 59 sequential choice High p4 75 simultaneous choice High p5 30 sequential choice High p5 46 simultaneous choice High p6 29 sequential choice High p6 44 simultaneous choice High p7 38 sequential choice High p7 24 simultaneous choice Medium p1 29 sequential choice Medium p1 24 simultaneous choice Medium p2 45 sequential choice Medium p2 41 simultaneous choice Medium p3 24 sequential choice Medium p3 24 simultaneous choice Medium p4 21 sequential choice Medium p4 16 simultaneous choice Medium p5 49 sequential choice Medium p5 30 simultaneous choice Medium p6 35 sequential choice Medium p6 44 simultaneous choice Medium p7 47 sequential choice Medium p7 12 simultaneous choice Low p1 27 sequential choice Low p1 0 simultaneous choice Low p2 23 sequential choice Low p2
  • 9. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 6 | P a g e 6 simultaneous choice Low p3 41 sequential choice Low p3 3 simultaneous choice Low p4 21 sequential choice Low p4 9 simultaneous choice Low p5 21 sequential choice Low p5 24 simultaneous choice Low p6 35 sequential choice Low p6 12 simultaneous choice Low p7 15 sequential choice Low p7 Here p1, p2, p3…., p7 correspond to 7 products listed in table 1. To determine the significance of each factor and interactions (if exist) among those factors upon the response, an ANOVA test was conducted. The following table3 shows the output of the ANOVA test. As we can see from the ANOVA table, the variety*condition interaction and variety variable are significant at 0.05 level of significance. Since the interaction between variety and condition variable is significant, we can conclude that the condition variable’s significance has been masked or leveled out by the presence of variety variable’s presence. A better understanding can be achieved by observing the interaction plot. Following figure1 shows all the 3 factors plotted against each other. table2 table3
  • 10. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 7 | P a g e The graph1 from figure1 clearly shows that as the condition changes from simultaneous choice to sequential choice, the number of subjects opting for higher variety decreases, whereas the number of people opting for low variety increases, hence interaction between condition and variety exists. Further the graph 5 shows that condition in independent of product and same can also be said about variety from graph 6 as the slopes are almost parallel. 7. Factors and interactions screening A better understanding of the interactions can be gained by coding the data of table1 according to the levels of each variables. The x𝑖𝑖 corresponding to each factor level is mentioned in parenthesis, example H (1) and simultaneous choice (1).Following formula is used to code each factor input: x′𝑖𝑖 = x𝑖𝑖 − x 𝐻𝐻 + x𝐿𝐿 2 x 𝐻𝐻 − x𝐿𝐿 2 figure1
  • 11. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 8 | P a g e Where x𝑖𝑖 is the original factor value, x 𝐻𝐻 is the highest value of that factor and x𝐿𝐿is the lowest value. The coding results in the following table4. subject condition variety product subject condition variety product 64 1 1 -3 44 -1 1 -3 76 1 1 -2 32 -1 1 -2 53 1 1 -1 35 -1 1 -1 73 1 1 0 59 -1 1 0 75 1 1 1 30 -1 1 1 46 1 1 2 29 -1 1 2 44 1 1 3 38 -1 1 3 24 1 0 -3 29 -1 0 -3 24 1 0 -2 45 -1 0 -2 41 1 0 -1 24 -1 0 -1 24 1 0 0 21 -1 0 0 16 1 0 1 49 -1 0 1 30 1 0 2 35 -1 0 2 44 1 0 3 47 -1 0 3 12 1 -1 -3 27 -1 -1 -3 0 1 -1 -2 23 -1 -1 -2 6 1 -1 -1 41 -1 -1 -1 3 1 -1 0 21 -1 -1 0 9 1 -1 1 21 -1 -1 1 24 1 -1 2 35 -1 -1 2 12 1 -1 3 15 -1 -1 3 Figure2 shows a surface plot of condition variety and subject plotted using the coded data. It is also clear from the figure2 that the number of subjects is the highest for simultaneous choice condition (1) with High variety (1) and the lowest for simultaneous choice condition (1) with low variety (-1). table4 figure2
  • 12. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 9 | P a g e The above discussion, tests, figures and graphs clearly show that the factors ‘variety’, ‘condition’ and interaction ‘variety*condition’ only are significant and should be considered for the hypothesis testing. The tests actually superimpose the hypothesis that variety seeking behavior of consumer is independent of the product category in a particular product family, in our case food products. To further verify this claim a normal quantile-quantile plot of various factors and their interaction effects is plotted. The following figure3 shows the plot. The graph clearly shows that product nor its interactions are significant and hence subject’s response should be considered independent of the same. figure3
  • 13. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 10 | P a g e 8. Fitted Model As a result, a model showing the relationship between number of subjects for a particular condition-variety combination will be independent of product category. 8.1 Regression Model A regression analysis of the coded data results in the following model and ANOVA test recorded in the table5. Linear regression model: Subject = 1 + 10.036*condition*variety The R2 value of the model being 0.7 means that the fitted model accounts for 70% of the subject’s variety seeking behavior. A behavioral study, dealing in data impacted by variability in human behavior, can consider a model with R2 as large as 70% to be a good fit for the data. table5
  • 14. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 11 | P a g e 8.2 Check for the presence of higher order terms To cross-check if the model could be containing second order terms, a lack of fit test was performed on the coded data, the results of the ANOVA test for lack-of-fit is shown in the following table6 The ANOVA test in table6 clearly shows that the lack-of-fit is insignificant at 0.05 level of significance and as a result higher order term of variety2 is insignificant and is not present in the model. 8.3 Analysis of the fitted model As a result the best fit model for the given study data is Subject = 1 + 10.036*condition*variety. The model clearly supports the alternate hypothesis that as condition increases form sequential choice (-1) to simultaneous choice (+1) along with variety from Low (-1) to High (-1) the number of subjects increases. This clearly proves the alternate hypothesis that consumers who simultaneously choose multiple items in a category for sequential consumption are more likely table6
  • 15. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 12 | P a g e to choose different items (higher variety) than consumers who sequentially make the same number of choices. A more formal multiple linear regression analysis for the uncoded data is performed in the following sections of the project report along with tests for various assumptions made for the data of the model. 9. Multiple Linear Regression Concluding from the design of experiment of the project study, the significant factors for the model are ‘variety’, ‘condition’ and interaction of ‘variety*condition’. The variables ‘variety’ and ‘condition’ are categorical variables as they divide the data into fixed categories. To represent these variables in numerical form for uncoded regression model, we need to assign numerical values to different levels of each factor. The following table7 gives the numerical values for categorical variables Assigning these values to each condition-variety and condition*variety interaction combination for subjects and depicting response ‘subject’ by y we get the following dataset projected in table8 x w1 w2 xw1 xw2 y x w1 w2 xw1 xw2 y 1 1 0 1 0 64 0 1 0 0 0 59 1 0 1 0 1 24 0 0 1 0 0 21 1 0 0 0 0 12 0 0 0 0 0 21 0 1 0 0 0 44 1 1 0 1 0 75 0 0 1 0 0 29 1 0 1 0 1 16 0 0 0 0 0 27 1 0 0 0 0 9 1 1 0 1 0 76 0 1 0 0 0 30 variety outcomes w1 w2 Condition x High (1) 1 0 simultaneous choices (1) 1 Medium (2) 0 1 sequential choices (0) 0 Low (3) 0 0 table7
  • 16. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 13 | P a g e 1 0 1 0 1 24 0 0 1 0 0 49 1 0 0 0 0 0 0 0 0 0 0 21 0 1 0 0 0 32 1 1 0 1 0 46 0 0 1 0 0 45 1 0 1 0 1 30 0 0 0 0 0 23 1 0 0 0 0 24 1 1 0 1 0 53 0 1 0 0 0 29 1 0 1 0 1 41 0 0 1 0 0 35 1 0 0 0 0 6 0 0 0 0 0 35 0 1 0 0 0 35 1 1 0 1 0 44 0 0 1 0 0 24 1 0 1 0 1 44 0 0 0 0 0 41 1 0 0 0 0 12 1 1 0 1 0 73 0 1 0 0 0 38 1 0 1 0 1 24 0 0 1 0 0 47 1 0 0 0 0 3 0 0 0 0 0 15 Here, xw1 and xw2 are the condition*variety interaction terms. Performing multiple linear regression for this dataset we get the following model and ANOVA test for the same depicted in table9 table8 t a b l e 9
  • 17. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 14 | P a g e The fitted model is y = 26.13 - 16.714x + 12w1 + 9.5714w2 + 40.143xw1 + 10xw2. The ANOVA test clearly shows that only terms ‘x’, ‘w1’, and ‘xw1’ are significant. 10. Best Fit model To further verify the above mentioned claim and design the best fit model, Cp and PRESS statistics were calculated and the following table10 displays models containing various terms along with the values of their Cp and PRESS statistic. PRESS is errors in prediction whereas Cp is a measure of model being under-fitted or over-fitted. The lowest value of both PRESS and Cp is preferred; PRESS for obvious reason; while Cp ≈number of model parameters which in our case would be four – ‘xw1’, ‘w2’, ‘w1’, x. After removing ‘xw2’ we get the following linear regression model along with ANOVA test in table11: table10
  • 18. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 15 | P a g e y = 23.643 – 11.714x + 14.5w1 + 14.571w2 + 35.143xw1 There is no significant change in the R2 values of previous and current model and hence it is a possibility that the model is still over-fitted. To further fit the model a stepwise multiple regression analysis approach was adopted, the resulting model and ANOVA test is given in the following table12. table11
  • 19. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 16 | P a g e The stepwise multiple linear regression model is y = 23.643 + 14.571w2 + 35.143xw1 The model shows a reduction of 1.6% in R2 from the original model but with only 2 variable and 1 constant term and hence is the best fit model for the given uncoded dataset. The model R2 value is 70.4% which is almost same as of coded model from design of experiment analysis. 11. Test for assumptions To perform various analysis like ANOVA while conducting design of experiment and multiple regression analysis, following assumptions were made, 1. The random error of disturbance, has constant variance also known as the homogeneous variance assumption. 2. The random error of disturbance is normally distributed. table12
  • 20. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 17 | P a g e Homogeneous variance is an important assumption made in regression analysis. Violations can often be detected through the appearance of the residual plot. Following figure4 shows the graph of residuals against fitted values of the model. The residuals (although not completely) show a healthy random scatter around a value of zero and hence the assumption of homogeneous variance holds well. To gain some type of idea regarding the normal error assumption, a normal probability plot of the residuals was generated. In this is the type of plot the horizontal axis represents the empirical normal distribution function on a scale that produces a straight-line plot when plotted against the residuals. Following Figure5 shows the normal probability plot of the residuals. figure4
  • 21. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 18 | P a g e The normal probability plot does reflect the straight-line appearance that one would like to see for the normality assumption to hold good. 12. Detecting outliers The model errors, often shed light on the presence of “suspect” data points where the above assumptions might be violated, an outlier is a data point where there is a deviation from the above assumptions. An unusually high value of R-Student statistics highlights data points where the error of fit is larger than what is expected by chance. The following table13 shows the calculated values of studentized residuals (ri) and R-studentized residuals (ti) for the fitted model. Obs. yi ŷi yi − ŷi ri ti 1 64 61.57143 2.428571 0.244767 0.241632 2 24 26.5 -2.5 -0.24688 -0.24372 figure5
  • 22. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 19 | P a g e 3 12 11.92857 0.071429 0.007054 0.006958 4 44 38.14286 5.857143 0.59032 0.58505 5 29 38.21429 -9.21429 -0.90991 -0.90774 6 27 23.64286 3.357143 0.331518 0.327494 7 76 61.57143 14.42857 1.454203 1.477254 8 24 26.5 -2.5 -0.24688 -0.24372 9 0 11.92857 -11.9286 -1.17795 -1.18434 10 32 38.14286 -6.14286 -0.61912 -0.61388 11 45 38.21429 6.785714 0.67009 0.66502 12 23 23.64286 -0.64286 -0.06348 -0.06262 13 53 61.57143 -8.57143 -0.86388 -0.86085 14 41 26.5 14.5 1.431876 1.453231 15 6 11.92857 -5.92857 -0.58545 -0.58017 16 35 38.14286 -3.14286 -0.31676 -0.31287 17 24 38.21429 -14.2143 -1.40366 -1.42297 18 41 23.64286 17.35714 1.714019 1.762101 19 73 61.57143 11.42857 1.151844 1.157107 20 24 26.5 -2.5 -0.24688 -0.24372 21 3 11.92857 -8.92857 -0.8817 -0.87898 22 59 38.14286 20.85714 2.102115 2.209657 23 21 38.21429 -17.2143 -1.69991 -1.74636 24 21 23.64286 -2.64286 -0.26098 -0.25767 25 75 61.57143 13.42857 1.353417 1.369327 26 16 26.5 -10.5 -1.03688 -1.03796 27 9 11.92857 -2.92857 -0.2892 -0.28558 28 30 38.14286 -8.14286 -0.82069 -0.81699 29 49 38.21429 10.78571 1.06509 1.067084 30 21 23.64286 -2.64286 -0.26098 -0.25767 31 46 61.57143 -15.5714 -1.56939 -1.60228 32 30 26.5 3.5 0.345625 0.341474 33 24 11.92857 12.07143 1.192054 1.199086 34 29 38.14286 -9.14286 -0.92148 -0.91955 35 35 38.21429 -3.21429 -0.31741 -0.31352 36 35 23.64286 11.35714 1.121518 1.125556 37 44 61.57143 -17.5714 -1.77096 -1.82597 38 44 26.5 17.5 1.728126 1.777872 39 12 11.92857 0.071429 0.007054 0.006958 40 38 38.14286 -0.14286 -0.0144 -0.0142 41 47 38.21429 8.785714 0.86759 0.864625 42 15 23.64286 -8.64286 -0.85348 -0.85028 table13
  • 23. The Effect Of Consumption Pattern On Variety- Seeking Behavior Of Customer 20 | P a g e As we can see from the table and figure5 there is no data point with a significantly high value of ti and hence the dataset is free from outliers. 13. Conclusion The final fitted model of y = 23.643 + 14.571w2 + 35.143xw1 is the best fit model which does not violate any assumptions. The model also satisfies the alternate hypothesis, consider the situation of simultaneous choice (x = 1) along with high variety (w1 = 1, w2 = 0) the value of y = 23.643 + 35.143 = 58.786 Which is the average value of subjects at simultaneous choice and high variety. Now consider the situation of sequential choice (x = 0) along with high variety (w1 = 1, w2 = 0) the value of y = 23.643 Which being average value of subjects at sequential choice and high variety is certainly quiet lower than the above y value of 58.786. The same can also be shown for simultaneous choice and low variety being lower than sequential choice and low variety and hence the null hypothesis is rejected in favor of alternate hypothesis. 14. Reference List • Ronald E. Walpole, R. H. M., Sharon L. Myers, Keying Ye. (2012). Probability & Statistics for Engineers & Scientists (9th ed.): Prentice Hall. • Simonson, I. (1990). The Effect of Purchase Quantity and Timing on Variety-Seeking Behavior. Journal of Marketing Research, 27(2), 150-162. doi:10.2307/3172842.