the excessive choice effect meets the market: experiments ......trey malone outline the excessive...
Post on 09-Sep-2020
2 Views
Preview:
TRANSCRIPT
The Excessive Choice Effect Meets the Market: Experiments on Craft Beer Choice
Trey MaloneDecember 2, 2016
Trey Malone
Outline
The Excessive Choice Effect Meets the Market: Experiments on Craft Beer Choice
Motivation → Study 1 → Study 2 → Discussion
Q & A
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 2
Trey Malone
What is the Excessive Choice Effect (ECE)?
Contrary to neoclassical utility theory, an excessive number of choices might lead to increased levels of regret and paralysis
E(U)
Number of options
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 3
Trey Malone
Antecedents
A large literature has provided inconclusive evidence on the existence of an ECE
Scheibehenne et al. (2010): Mixed results from 50 experiments (N ≈ 5,000)
Example: Scheibehenne (2008) unable to replicate the famous jam study
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 4
Trey Malone
Antecedents
More interesting is “why” and “so what”?
Regret minimization (Irons and Hepburn 2007)
Information asymmetry (Kamenica 2008)
Consumer heterogeneity (Schwartz et al. 2002)
Search costs and uncertainty (Norwood 2006)
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 5
Trey Malone
Antecedents
More interesting is “why” and “so what”?
Regret minimization (Irons and Hepburn 2007)
Information asymmetry (Kamenica 2008)
Consumer heterogeneity (Schwartz et al. 2002)
Search costs and uncertainty (Norwood 2006)
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 6
Trey Malone
Search costs: it’s easier to find the red ball when N=6 than when N=24
Search Costs and Uncertainty
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 7
Trey Malone
Uncertainty: it’s even harder to choose a ball when your preference for red is not very strong
Search Costs and Uncertainty
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 8
Trey Malone
Imagine buying beer at
(1) a liquor store or (2) a gas station
Key point: store has an incentive to keep the right number of options stocked
Search Costs and Uncertainty
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 9
Trey Malone
Research Questions
How does the ECE vary with consumers?
Do sellers encourage purchase by altering the choice architecture for buyers?
Can reduced search costs mitigate the ECE?
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 10
Trey Malone
Value-Added
Context and consumer heterogeneity play an integral part in the existence of an ECE
I empirically test the ECE’s potential causes
Reducing uncertainty and the cost of search can minimize ECE
By providing buyer-relevant information, sellers alter choice architecture to maximize sales
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 11
Trey Malone
Why Craft Beer?
Fulfills the preconditions of an ECE
“Sufficiently” large number of options
No beer listed is highly dominant
Vaguely defined preferences
Beer is an experience good
Quality can only be determined after purchase
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 12
Trey Malone
Why Craft Beer?
Proliferation of food options
Craft brewing has growing implications for agricultural producers
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 13
Trey Malone
Map of Breweries in the Continental United States, July 2015
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 14
Trey Malone
General Hypotheses
Increasing the number of beers at a bar will induce the ECE for some participants
Targeted information might reduce search costs and uncertainty, thereby reducing the probability of no purchase
Beer Advocate scores
Prominently listed “Special”
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 15
Trey Malone
Survey Experiment (1) Field Experiment (2)
Pros Cons Pros Cons
• Individualized data
• Stated preferences
• Revealedpreferences
• Loss of Control
• Control • Context
• No individualspecific data
• Within subjects
(more power)
• Between subjects
Two Studies
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 16
Trey Malone
ECE varies with:
1. Beer Advocate scores
2. Prominently listed “Special”
3. Novelty-seeking
(Manning, Bearden and Madden 1995)
Study 1: Hypotheses
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 17
Trey Malone
Study 1 Data: Consumption Frequency (N=1,697)
Beer Frequency
Craft BeerFrequency
Wine/LiquorFrequency
Never Less than Once a Month
Once a Month 2-3 Times a Month
Once a Week 2-3 Times a Week
Daily
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 18
Trey Malone
Study 1: Experimental Design
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 19
Trey Malone
Study 1: Experimental Design
5 Beers
Control
Special
Beer Advocate Scores
18 Beers
Control
Special
Beer Advocate Scores
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 20
Trey Malone
Study 1 Results: Heat Maps
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 21
142 of 1,697 participants (8.4%) selected a beer that was
revealed dominated in their earlier choice
Trey Malone
Study 1 Results: Heat Maps
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 22
Trey Malone
Study 1 Results: Heat Maps
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 23
Trey Malone
Study 1 Results: Logit(Prob. of None)
Variable
Standard Logit with
Clustered Standard
Errors
Random Effects
Logit
Random Effects
Logit with
Novelty-seeking
Interactions
Intercept -0.414* (0.058) -0.447* (0.081) 1.245* (0.287)
Specials -0.671* (0.120) -0.746* (0.115) -5.370* (0.424)
Scores -1.557* (0.129) -1.709* (0.128) -4.643* (0.447)
Number of options -0.036* (0.003) -0.040* (0.006) -0.011 (0.021)
Novelty-seeking --- --- -0.505* (0.084)
# of options x Specials 0.030* (0.007) 0.034* (0.009) 0.302* (0.032)
# of options x Scores 0.076* (0.008) 0.083* (0.009) 0.255* (0.033)
# of options x Novelty-seeking --- --- -0.010 (0.006)
Specials x Novelty-seeking --- --- 1.381* (0.120)
Scores x Novelty-seeking --- --- 0.895* (0.129)
# of options x Specials x
Novelty-seeking
--- --- -0.080* (0.009)
# of options x Scores x
Novelty-seeking
--- --- -0.053* (0.010)
Generalized Chi-square N/A 6905.53 7139.76
Number of choices 10182 10182 10182
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 24
Trey Malone
Study 1 Results: P 𝑦𝑖 = 0
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 25
Probability of not selecting a beer in the menu experiment(Note: upward sloping line indicates presence of an ECE)
0
0.2
0.4
0.6
0.8
1
N=5 N=18
Control Special Advocate Scores
Trey Malone
Study 1 Results: P( 𝑦𝑖 = 0 |CNS)
High Novelty-Seeking (CNS=5)Low Novelty-Seeking (CNS=1)
Probability of not selecting a beer in the menu experiment(Note: upward sloping line indicates presence of an ECE)
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 26
N=5 N=18
Control Special Advocate Scores
0
0.2
0.4
0.6
0.8
1
N=5 N=18
Trey Malone
Study 1: Implications
The ECE does not affect everyone equally
Sellers likely make decisions based on their desired type of buyer
Novelty-seeking buyers likely have lower concern for uncertainty
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 27
Trey Malone
In a field context where consumers are not likely to be novelty-seeking in their beer preferences, an ECE will exist
If the ECE does exist, targeted information will reduce its consequences by lowering search costs
Study 2: Hypotheses
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 28
Trey Malone
Study 2: Experimental Design
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 29
Field experiment conducted at a wine bar in Stillwater, OK
September to October, 2014
Two datasets collected to improve accuracy
Restaurant sales system
Server-collected receipts
Trey Malone
Study 2: Experimental Design
6 Beers
Control
Special
Beer Advocate Scores
Control
12 Beers
Control
Special
Beer Advocate Scores
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 30
Trey Malone
Study 2: Experimental Design
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 31
Trey Malone
Study 2 Results: Odds Ratios
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 32
TreatmentNumber of Options Odds Ratio - effect of
number of optionsN=6 N=12
Control 16.86% 13.64% 0.959
[0.907, 1.014]b
Special 20.14% 13.25% 0.920
[0.867, 0.976]
Beer Advocate Scores 19.93% 20.44% 1.005
[0.958, 1.055]
Odds Ratio - effects of special and scores
Control to Special 1.244
[0.922, 1.678]
0.968
[0.661, 1.417]
Control to Scores 1.227
[0.929, 1.620]
1.627
[1.154, 2.295]
Beer Sales as a Percentage of All Drink Sales in Field Experiment
Trey Malone
Study 2 Results: P 𝑦𝑖 = 0
Results for Experiment 2
0.75
0.80
0.85
0.90
0.95
N=6 N=12Number of Menu Options
ControlSpecial
BA Scores
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 33
Probability of not selecting a beer in the field experiment(Note: upward sloping line indicates presence of an ECE)
Trey Malone
Study 2: Implications
In this context, the ECE existed
Simply providing a prominently listed “Special” does not appear to reduce the ECE
Providing third-party quality information (i.e. Beer Advocate scores) eliminated the ECE
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 34
Trey Malone
Discussion: Value-Added
Market context and consumer heterogeneity matter
Reducing uncertainty and cost of search can minimize ECE
By providing buyer-relevant information, sellers alter choice architecture to maximize sales
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 35
Trey Malone
Discussion: Key Applied Contribution
To overcome the ECE, know your target market
If your consumers prefer novelty, give them novelty
“the trend for 2016 is not a trend, rather it’s diversity”– Bart Watson, Brewers’ Association
In some contexts, it might be helpful to “nudge” consumers via quality identification
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 36
Trey Malone
Discussion: Related Future Research
Are there benefits to categorization?Manipulate menu categories at a pub
Do “mavens” (i.e. bartenders or friends) reduce search costs?
Measure changes in sales/satisfaction when bartender makes recommendations
Veblen effects?Invent a beer, and ask bar patrons if they are more (less) likely to buy it at varying prices
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 37
Trey Malone
Q&A
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 38
The Excessive Choice Effect Meets the Market: Experiments on Craft Beer Choice
Trey MaloneOctober 17, 2016
Trey Malone
Study 1 Data: Description
Gender Ethnicity Education Age IncomeMale 54.15% White/
Caucasian79.08% Less than High
School1.00% 21 to 24
years7.54% Less than
$29,99915.67%
Female 45.85% African American
7.37% High School/GED 15.62% 25 to 34 years
25.22% $30,000-$59,999
31.17%
Hispanic 7.25% Some College 21.98% 35 to 44 years
16.85% $60,000-$89,999
23.63%
Asian 4.77% 2-year College Degree
11.61% 45 to 54 years
20.39% $90,000-$119,999
12.73%
Native American
0.24% 4-year College Degree
32.41% 55 to 64 years
17.86% $120,000-$149,999
8.66%
Pacific Islander
0.35% Masters Degree 13.73% 65 to 74 years
10.67% $150,000-$179,999
3.36%
Other 0.94% Doctoral Degree 1.36% 75 to 84 years
1.18% More than $180,000
4.77%
Professional Degree (JD, MD)
2.30% 85 years or over
0.29%
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 40
Trey Malone
AppendixConsumer Novelty-Seeking Scale
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 41
CNS scale statement Mean
I often seek out information about new product and brands.3.427
(1.316)
I like to go to places where I will be exposed to information about new products and
brands.
3.456
(1.252)
I like magazines that introduce new brands.3.104
(1.551)
I frequently look for new products and services.3.433
(1.310)
I seek out situations in which I will be exposed to new and different sources of product
information.
3.354
(1.259)
I am continually seeking new product experiences3.426
(1.339)
When I go shopping, I find myself spending very little time checking out new products
and brands.
2.853
(1.554)
I take advantage of the first available opportunity to find out about new and different
products
3.282
(1.334)
Novelty Seeking Score3.292
(0.198)
Trey Malone
Study 1 Results: Odds Ratios
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 42
TreatmentNumber of Options Odds Ratio – effect of number
of optionsN=5 N=18
Control 64.44% 74.34% 1.037
[1.030, 1.043]b
Special 75.28% 76.62% 1.006
[0.992, 1.020]
Beer Advocate Scores 72.46% 77.80% 1.022
[1.016, 1.028]
Odds Ratio - effect of specials and scores
Control to Special 1.680
[1.421, 1.987]
1.131
[1.067, 1.200]
Control to Scores 1.452
[1.343, 1.571]
1.209
[1.129, 1.296]
Percentage of Times Beer Was Chosen in Survey-Based Choice Experiment
Trey Malone
Study 1 Data: Identification as a craft beer drinker
Not at all like me12%
Not much like me14%
Somewhat like me27%
Quite a lot like me24%
Just like me23%
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 43
Trey Malone
AppendixBeer options for the field experiment
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 44
Trey Malone
AppendixZanotti’s menu with Beer Advocate scores
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 45
Trey Malone
Embedded Short Trap Question (Above) and Long Trap Question (Below)
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 46
Trey Malone
No FeedbackEmbedded Short
Trap QuestionLong Trap Question
Number of participants 1000 1017% incorrect 20.0 37.2
With FeedbackEmbedded Short
Trap QuestionLong Trap Question
Number of participants 559 591% incorrect 21.8 25.4% revised correctly 9.5 16.6
Effectiveness of Trap Questions at Identifying Inattentive Respondents
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 47
Trey Malone
WTP for…
Short Trap Question Long Trap Question
Correct
Respondents
Incorrect
Respondents
Correct
Respondents
Incorrect
RespondentsHamburger $4.16
[$3.98, $4.34]
$5.32
[$4.53, $6.11]
$3.77
[$3.60, $3.93]
$5.91
[$5.14, $6.68]
Steak $7.23
[$6.86, $7.60]
$10.46
[$5.88, $15.04]
$6.11
[$5.77, $6.45]
$12.07
[$8.55, $15.58]
Pork Chop $3.85
[$3.64, $4.05]
$5.08
[$2.68, $7.47]
$3.33
[$3.16, $3.50]
$5.80
[$4.44, $7.16]
Deli Ham $2.41
[$2.21, $2.63]
$3.81
[$2.05, $5.58]
$1.91
[$1.73, $2.08]
$4.62
[$3.34, $5.89]
Chicken Breast $5.43
[$5.18, $5.68]
$7.77
[$5.81, $9.74]
$4.56
[$4.36, $4.75]
$6.98
[$6.2, $7.76]
Chicken Wings $2.14
[$1.92, $2.36]
$2.57
[$0.40, $4.75]
$1.64
[$1.47, $1.80]
$4.57
[$3.25, $5.89]
Beans and Rice $2.76
[$2.41, $3.11]
$1.91
[$0.75, $3.08]
$1.94
[$1.66, $2.22]
$3.27
[$2.42, $4.12]
Pasta $3.67
[$3.40, $3.94]
$2.40
[-$1.35, $6.14]
$2.63
[$2.36, $2.91]
$3.87
[$3.18, $4.56]
Comparison of Willingness-to-Pay (WTP) Estimates for Participants who Correctly and Incorrectly Responded to the Trap Question
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 48
Trey Malone
WTP for…
Short Trap Question Long Trap Question
Correct
Respondents
Incorrect
RespondentsCorrect
Respondents
Incorrect
RespondentsMiller Lite $5.70
[$5.00, 6.40]
$10.31
[$5.70, $14.90]$3.75
[$3.33, $4.17]
$7.17
[$4.93, $9.41]
Corona$7.85
[$6.91, $8.80]
$13.93
[$-8.54, $36.41]$5.95
[$5.51, $6.40]
$30.13
[$-967.89,
$1028.15]
Samuel
Adams$10.09
[$8.71, 11.48]
$9.35
[$5.52, $13.15]$6.44
[$5.94, $6.94]
$18.64
[$-159.7, $196.98]
Budweiser $7.98
[$6.33, $9.62]
$8.05
[$5.32, $10.76]$4.32
[$3.99, $4.66]
$12.03
[$-13.99, $38.05]
Oskar Blues $3.31
[$2.75, $3.86]
$2.75
[$-85.43, 90.82]$1.64
[$0.85, $2.44]
$2.91
[$-5.01, $10.84]
Marshall$5.17
[$4.54, $5.79]
$0.83
[$-49.28, 50.97]$3.04
[$2.58, $3.51]
$5.70
[$-114.08, $125.48]
Comparison of Willingness-to-Pay (WTP) Estimates for Participants who Correctly and Incorrectly Responded to the Trap Questions with Feedback
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 49
Trey Malone
Per Capita Consumption (2011, in gallons)
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 50
Trey Malone
Self Distribution Legislation (2014)
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 51
Trey Malone
State Excise Taxes (2012, $ per barrel)
Source: Brewers’ Almanac
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 52
Trey Malone
Parameter Estimates for Total Number of Breweries per County
Parameter Poisson ZIP ZINB
Intercept -0.189 (0.420) -1.477* (0.426) -2.001* (0.654)
Political contributions 0.003* (0.001) 0.005* (0.001) 0.012* (0.002)
Beer taxes per barrel 0.007* (0.003) 0.002 (0.003) 0.003 (0.006)
Availability score 0.295* (0.064) 0.331* (0.067) 0.336* (0.100)
Self-distribution 0.228* (0.062) 0.205* (0.067) 0.289* (0.093)
Growler legislation 0.056 (0.035) -0.013 (0.037) 0.164* (0.067)
Per capita beer consumption 0.025* (0.007) 0.012 (0.007) 0.015 (0.011)
Median household income -0.249* (0.016) -0.068* (0.015) -0.124* (0.035)
High Creative class 0.758* (0.051) 1.056* (0.054) 0.909* (0.087)
Percent of persons with a college degree0.051* (0.002) 0.039* (0.002) 0.064* (0.005)
Rural Urban Continuum code -0.312* (0.010)
High amenities 0.219* (0.053)
δ 0.741* (0.050)
Regional fixed effects? Yes Yes Yes
Religion effects? Yes Yes Yes
Logit link function
Intercept -3.051* (0.250) -7.343* (1.010)
Rural Urban Continuum code 0.480* (0.037) 0.917* (0.121)
High amenities -1.022* (0.201) -1.463* (0.452)
AIC 8008 8732 6952
Number of
observations in
each regression
= 3,108
Total number of
breweries
= 4,400
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 53
Trey Malone
Brewing Up Entrepreneurship: Government Intervention in Craft Beer
0
0.5
1
1.5
2
2.5
3
3.5
Banned Legal
Total Brewpub Micro
Expected Number of Breweries, Microbreweries, and
Brewpubs within a County when the State Legalizes
On-Premises Growler Sales
0
0.5
1
1.5
2
2.5
3
3.5
Banned Legal
Total Brewpub Micro
Expected Number of Breweries, Microbreweries, and
Brewpubs within a County when the State Legalizes
Self-Distribution
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 54
Trey Malone
If You Brew it, Will They Come? Factor Analysis
Miller Lite
Corona
Sam Adams
Budweiser
Oskar BluesMarshall
0
0.5
1
0 0.5 1
Cla
ssic
Novel
Factor Analysis of Taste Perceptions
Miller Lite
Corona
Sam Adams
Budwesier
Oskar Blues
Marshall
-0.5
0
0.5
1
0 0.5 1
Nic
he
Established
Factor Analysis of Beer Familiarity
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 55
Trey Malone
Parameter Estimates for Brand Equity Models
ParameterNo Equity
Effects
Brand Equity Main
Effects
Brand Equity with
Interactions
Price -0.266* (0.007) -0.360* (0.009) -0.066 (0.139)
Taste 1.384* (0.018) 1.729* (0.168)
Familiarity 0.648* (0.024) 0.960* (0.235)
Price x Familiarity -0.093 (0.054)
Price x Taste -0.097 (0.039)
Familiarity x Taste -0.110 (0.063)
Price x Familiarity x Taste 0.032* (0.015)
Miller Lite 1.414* (0.048) -4.460* (0.087) -5.485* (0.597)
Budweiser 1.595* (0.047) -4.588* (0.088) -5.613* (0.597)
Corona 2.023* (0.045) -4.348* (0.087) -5.375* (0.596)
Samuel Adams Lager 2.177* (0.045) -4.328* (0.088) -5.355* (0.596)
Oskar Blues Pilsner 0.511* (0.056) -4.408* (0.081) -5.439* (0.595)
Marshall Pilsner 1.049* (0.050) -4.030* (0.078) -5.058* (0.595)
-2 Log likelihood 48045 33864 33855
AIC 48058.7 33882.1 33881.2
Number of observations 13576 13576 13576
Number of participants 1697 1697 1697
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 56
Trey Malone
Own Elasticities for Price, Taste, and Familiarity
Own-Price Own-Familiarity Own-Taste
Miller Lite -0.879
[-0.92, -0.839]a
2.741
[2.673, 2.81]
1.580
[1.466, 1.695]
Budweiser -0.826
[-0.864, -0.788]
2.719
[2.651, 2.787]
1.583
[1.468, 1.697]
Corona -1.815
[-1.898, -1.731]
2.946
[2.872, 3.019]
1.927
[1.787, 2.067]
Sam Adams -1.767
[-1.848, -1.685]
2.714
[2.646, 2.782]
1.951
[1.81, 2.093]
Oskar Blues -2.097
[-2.194, -2.001]
1.819
[1.773, 1.864]
1.835
[1.702, 1.968]
Marshall -2.053
[-2.147, -1.958]
1.765
[1.72, 1.809]
1.867
[1.731, 2.002]
December 2, 2016 Motivation | Study 1 | Study 2 | Discussion 57
top related