Gill Atlas Hardisty AERE 2016davidhardisty.info/downloads/Gill Atlas Hardisty... · Gill Atlas...
Transcript of Gill Atlas Hardisty AERE 2016davidhardisty.info/downloads/Gill Atlas Hardisty... · Gill Atlas...
Matching costs to context: Status quo bias, temporal framing, and household energy decisions
Carrie Gill1, Stephen Atlas2, and David Hardisty3 1 PhD Candidate, Department of Environmental and Natural Resource Economics, University of Rhode Island
2 Assistant Professor of MarkeCng, College of Business AdministraCon, University of Rhode Island 3 Assistant Professor, MarkeCng and Behavioral Science Division, Sauder School of Business, University of BriCsh Columbia
US residenCal consumers used 10.2 quadrillion BTU of energy in 2009 • Amounts to nearly $230 billion spent on uClity bills • NegaCve environmental and health externaliCes
MoCvaCon: reduce energy consumpCon
Status quo bias may lead to inefficiency
Individuals tend to sCck with defaults and status quos
Individuals tend to keep contractor’s arbitrary light bulb choice
Dinner et al. 2011
Make costs salient – but how?
This cup of coffee costs $1.80.
This cup of coffee costs 180 pennies.
Framing is important
CogniCve fluency affects decision making
FLUENCY: The subjecCve experience of ease of processing informaCon Using familiar, easy to read, compare, understand descripCons leads to… • Product a^racCveness • IntenCons to exercise and cook • Choosing fuel-‐efficient cars
Difficulty in construcCng preferences (e.g. hard to disCnguish between a^ributes) may result in deferred choices
Lembregts and Pandelaere 2013; Song and Schwarz 2008; Camilleri and Larrick 2014; Novemsky et al 2007
ObjecCves
• IdenCfy which temporal frame of costs/savings reduces status quo bias in energy decisions (Study 1)
• InvesCgate the role of cogniCve fluency (Study 2)
• Explore the effect of providing context for evaluaCng costs in less-‐fluent frames (Study 3)
Online survey: N = 353 MTurkers, between-‐subjects experiment. We manipulated status quo behavior (inefficient, efficient) and how the costs or savings of switching behaviors was framed (per day, per month, per year).
Study 1: Status quo bias and cost frame
Status Quo Behavior
Inefficient
Efficient
Cost Frame
Per Day Per Month Per Year
Significant interacCon effect F (2, 347) = 4.33; p < 0.05
*** and ** indicate significance at the 1% and 5% levels, respecCvely, using the per family error rate.
Simple main effects of status quo at temporal frame:
Daily: F (1, 347) = 17.27***
Monthly: F (1, 347) = 0.01
Yearly: F (1, 347) = 7.01**
When doing the laundry, you can choose what water temperature your washing machine uses.
Some people choose to use warm or hot water because they think warmer water is most effec>ve for cleaning laundry.
However, warm and hot water use more energy than cold water.
Suppose you oBen use warm water, but you are considering whether to use only cold water for your laundry.
If you only use cold water, you will save 17 cents per day on energy costs. What do you think you would do? Please answer on the following scale where 10 means that you would definitely only use cold water and 0 means you would definitely use warm water.
Study 1: Example scenario and treatment
10 7 0 1 2 3
Definitely warm water
4 5 6 8 9
Definitely cold water
Significant interacCon effect F (2, 347) = 4.33; p < 0.05
*** and ** indicate significance at the 1% and 5% levels, respecCvely, using the per family error rate.
Simple main effects of status quo at temporal frame:
Daily: F (1, 347) = 17.27***
Monthly: F (1, 347) = 0.01
Yearly: F (1, 347) = 7.01**
When doing the laundry, you can choose what water temperature your washing machine uses.
Some people choose to use warm or hot water because they think warmer water is most effec>ve for cleaning laundry.
However, warm and hot water use more energy than cold water.
Suppose you oBen use warm water, but you are considering whether to use only cold water for your laundry.
If you only use warm water, you will pay 17 cents more per day on energy costs. What do you think you would do? Please answer on the following scale where 10 means that you would definitely only use cold water and 0 means you would definitely use warm water.
Study 1: Example scenario and treatment
10 7 0 1 2 3
Definitely warm water
4 5 6 8 9
Definitely cold water
Significant interacCon effect F (2, 347) = 4.33; p < 0.05
*** and ** indicate significance at the 1% and 5% levels, respecCvely, using the per family error rate.
Simple main effects of status quo at temporal frame:
Daily: F (1, 347) = 17.27***
Monthly: F (1, 347) = 0.01
Yearly: F (1, 347) = 7.01**
When doing the laundry, you can choose what water temperature your washing machine uses.
Some people choose to use warm or hot water because they think warmer water is most effec>ve for cleaning laundry.
However, warm and hot water use more energy than cold water.
Suppose you oBen use warm water, but you are considering whether to use only cold water for your laundry.
If you only use warm water, you will pay $5.25 more per month on energy costs. What do you think you would do? Please answer on the following scale where 10 means that you would definitely only use cold water and 0 means you would definitely use warm water.
Study 1: Example scenario and treatment
10 7 0 1 2 3
Definitely warm water
4 5 6 8 9
Definitely cold water
Significant interacCon effect F (2, 347) = 4.33; p < 0.05
*** and ** indicate significance at the 1% and 5% levels, respecCvely, using the per family error rate.
Simple main effects of status quo at temporal frame:
Daily: F (1, 347) = 17.27***
Monthly: F (1, 347) = 0.01
Yearly: F (1, 347) = 7.01**
When doing the laundry, you can choose what water temperature your washing machine uses.
Some people choose to use warm or hot water because they think warmer water is most effec>ve for cleaning laundry.
However, warm and hot water use more energy than cold water.
Suppose you oBen use warm water, but you are considering whether to use only cold water for your laundry.
If you only use warm water, you will pay $63 more per year on energy costs. What do you think you would do? Please answer on the following scale where 10 means that you would definitely only use cold water and 0 means you would definitely use warm water.
Study 1: Example scenario and treatment
10 7 0 1 2 3
Definitely warm water
4 5 6 8 9
Definitely cold water
Significant interacCon effect F (2, 347) = 4.33; p < 0.05
*** and ** indicate significance at the 1% and 5% levels, respecCvely, using the per family error rate.
Simple main effects of status quo at temporal frame:
Daily: F (1, 347) = 17.27***
Monthly: F (1, 347) = 0.01
Yearly: F (1, 347) = 7.01**
Many homes have a second fridge or freezer. They are typically used for addi>onal food storage or convenience, and are oBen located in a garage or basement.
Suppose you own a second fridge and are deciding what to do with it?
If it’s plugged in, you can use it to keep things cold, but must pay for its energy costs.
If it’s not plugged in, you can store it in the house, give it away or sell it.
Suppose you are considering whether to disconnect a second fridge in your house.
If you disconnect the second fridge, you will save $10.00 per month on energy costs. What do you think you would do? Please answer on the following scale where 10 means that you would definitely disconnect the fridge and 0 means you would definitely connect the fridge.
Study 1: Example scenario and treatment
10 7 0 1 2 3
Definitely connect
4 5 6 8 9
Definitely disconnect
7.02
5.19
6.85 6.827.56
6.42
01
23
45
67
89
10B
ehav
ior I
nten
tion
[0: i
neffi
cien
t - 1
0: e
ffici
ent]
Daily Costs Monthly Costs Yearly Costs
Efficient Status Quo Inefficient Status Quo
N=353, +/- 1 standard error shown
Study 1: Status quo bias minimized in monthly frame
Study 1: ANOVA and regression results
df [1] [2] [3] [4]Status Quo 1 F = 15.93*** F = 20.24***Efficient omitted omittedInefficient -1.827 (0.439)*** -1.820 (0.427)***
Cost Frame 2 F = 4.67*** F = 5.35***Per Day omitted omittedPer Month -0.166 (0.438) -0.172 (0.424)Per Year 0.541 (0.441) 0.672 (0.430)
Interaction 2 F = 4.33** F = 3.59**Ineff. x Month 1.793 (0.616)*** 1.581 (0.598)***Ineff. x Year 0.684 (0.616) 0.589 (0.599)
ControlsLoss Aversion Coef. 1 F = 1.90 0.0510 (0.0370)Construal Level 1 F = 1.09 -0.572 (0.548)Cognitivie Reflection 1 F = 1.52 0.531 (0.431)Environmental Concern 1 F = 24.77*** 0.729 (0.146)***Constant 7.018 (0.317)*** 6.981 (0.484)***Obs 353 353 353 353R-squared 0.090 0.159 0.09 0.159
Analysis of Variance OLS RegressionDV = Behavior Intention [0: inefficient - 10: efficient]
Notes: *, **, and *** represent statistical significance at probabilities < 0.1, 0.05, and 0.01, respectively. F statistics shown for analysis of variance (columns 1 and 2). Regression coefficients with standard errors in parentheses shown for regression analysis (columns 3 and 4).
7.02
5.19
6.85 6.827.56
6.42
01
23
45
67
89
10B
ehav
ior I
nten
tion
[0: i
neffi
cien
t - 1
0: e
ffici
ent]
Daily Costs Monthly Costs Yearly Costs
Efficient Status Quo Inefficient Status Quo
N=353, +/- 1 standard error shown
Study 1: Status quo bias minimized in monthly frame
*** and ** indicate significance at the 1% and 5% levels, respectively, using the per family error rate.
Simple Main Effect of Status Quo at Cost Frame
*** not sig. **
à Monthly framing minimizes status quo bias
7.02
5.19
6.85 6.827.56
6.42
01
23
45
67
89
10B
ehav
ior I
nten
tion
[0: i
neffi
cien
t - 1
0: e
ffici
ent]
Daily Costs Monthly Costs Yearly Costs
Efficient Status Quo Inefficient Status Quo
N=353, +/- 1 standard error shown
Could monthly energy bills provide an implicit context to help evaluate costs?
Study 1: Status quo bias minimized in monthly frame
Study 2: CogniCve fluency across frames
Online survey: N = 1,199 MTurkers, between-‐subjects experiment. Same scenario set-‐up as Study 1. We elicited fluency using five measures and and compared across cost frames.
Fluency measures on 7-‐point Likert scale:
• Es>ma>ng how the energy decision would financially impact me was… • Understanding what the energy decision meant was… • The descrip>on of the financial impact seemed… • The descrip>on of the energy decision seemed… • How involved were you in the energy decision?
-0.1
0.0
0.1
Flue
ncy
Inde
x
Daily Costs Monthly Costs Yearly Costs
N=1,199 +/- 1 SE
Study 2: Fluency is highest in the monthly frame
Cronbach’s α = 0.790
df [1] [2] [3] [4]Cost Frame 2 F = 3.48** F = 3.99**Per Day omitted omittedPer Month 0.120 (0.052)** 0.127 (0.051)**Per Year 0.002 (0.052) 0.003 (0.051)
ControlsCare about saving money 1 F = 17.08*** 0.118 (0.028)***Care about environment 1 F = 7.82*** 0.065 (0.023)***Constant -0.041 (0.037) -0.833 (0.138)***Obs 1,199 1,199 1,199 1,199R-squared 0.006 0.035 0.006 0.035
DV = Cognitive Fluency [larger = more fluent]Analysis of Variance OLS Regression
Notes: *, **, and *** represent statistical significance at probabilities < 0.1, 0.05, and 0.01, respectively. F statistics shown for analysis of variance (columns 1 and 2). Regression coefficients with standard errors in parentheses shown for regression analysis (columns 3 and 4).
Study 2: Fluency is highest in the monthly frame
-0.1
0.0
0.1
Flue
ncy
Inde
x
Daily Costs Monthly Costs Yearly Costs
N=1,199 +/- 1 SE
à Individuals are most fluent with monthly costs
Study 2: Fluency is highest in the monthly frame
Your energy spending over 1 day�
3.75 Energy'Meter'
$� per day�
ENERGY COMPANY
Study 3: Provide context for unfamiliar frames
Online survey: N = 132 Qualtrics recruits, between-‐subjects experiment We provide an explicit context for total energy spending (ler fig.) and costs of engaging in energy-‐inefficient behaviors (right fig.)
Daily cost of inefficiency Daily spending context
Study 3: Provide context for unfamiliar frames
Online survey: N = 132 Qualtrics recruits, between-‐subjects experiment We provide an explicit context for total energy spending (ler fig.) and costs of engaging in energy-‐inefficient behaviors (right fig.)
Yearly cost of inefficiency Yearly spending context
Your energy spending over 1 year�
1,370 Energy'Meter'
$� per year�
ENERGY COMPANY
7.70
6.73
7.588.01
01
23
45
67
89
10B
ehav
ior I
nten
tion:
[0: i
neffi
cien
t - 1
0: e
ffici
ent]
Daily Spending Yearly Spending
Daily Costs Yearly Costs
N=132, +/- 1 standard error shown
Study 3: Provide context for unfamiliar frames
Study 3: Provide context for unfamiliar frames
df [1] [2] [3] [4]Per Year Cost Frame 1 F = 0.82 F = 2.38 -0.124 (0.423) -1.22 (0.420)***Per Year Spending Frame 1 F = 3.70* F = 2.61 -0.971 (0.427)** -0.294 (0.412)Interaction 1 F = 5.47** F = 6.97*** 1.400 (0.599)** 1.533 (0.581)***
ControlsConstrual Level 1 F = 0.29 -0.062 (0.115)Math Ability 1 F = 0.00 0.009 (0.953)Environmental Concern 1 F = 9.15*** -0.575 (0.190)***Care about Saving Money 1 F = 1.26 0.293 (0.261)Constant 7.705 (0.290)*** 6.024 (1.870)***Obs 132 132 132 132R-squared 0.070 0.160 0.070 0.160
DV = Behavior Intention [0: inefficient - 10: efficient]Analysis of Variance OLS Regression
Notes: *, **, and *** represent statistical significance at probabilities < 0.1, 0.05, and 0.01, respectively. F statistics shown for analysis of variance (columns 1 and 2). Regression coefficients with standard errors in parentheses shown for regression analysis (columns 3 and 4).
7.70
6.73
7.588.01
01
23
45
67
89
10B
ehav
ior I
nten
tion:
[0: i
neffi
cien
t - 1
0: e
ffici
ent]
Daily Spending Yearly Spending
Daily Costs Yearly Costs
N=132, +/- 1 standard error shown
Study 3: Provide context for unfamiliar frames
Simple Main Effect of Cost Frame at Spending Frame
** not sig.
** indicates significance at the 5% level using the per family error rate.
7.70
6.73
7.588.01
01
23
45
67
89
10B
ehav
ior I
nten
tion:
[0: i
neffi
cien
t - 1
0: e
ffici
ent]
Daily Spending Yearly Spending
Daily Costs Yearly Costs
N=132, +/- 1 standard error shown
à Supple-‐menIng cost info with a spending context in the same frame can increase efficiency intenIons
Study 3: Provide context for unfamiliar frames
When targeCng an energy-‐inefficient audience without an explicit spending context – consider framing costs as per month. Providing a context for total energy spending in the same frame as costs of energy inefficiency may help reduce resistance to adopCng energy-‐efficient behaviors. Yearly framing of costs/savings coupled with annual total energy spending may encourage the most energy efficient behavior intenCons. Future research will invesCgate when aggregaCng costs broadly is most effecCve. Future research will apply findings to improve Energy Guide labels and translate intenCons of energy efficiency to electricity and monetary savings.
Conclusions
Thanks to Ethan Rix, URI’s Undergraduate Research IniCaCve Grant, Qualtrics Behavioral Research Grant, URI’s Mental AccounCng and Pricing Lab, ENRE Experimental Economics Group, and to a^endees of the 2015 Annual MeeCng of the Society for Judgment and Decision Making.
Acknowledgements
Alter, A. L., & Oppenheimer, D. M. (2009). UniCng the tribes of fluency to form a metacogniCve naCon. Personality and social psychology review.
Camilleri, A. R., & Larrick, R. P. (2014). Metric and Scale Design as Choice Architecture Tools. Journal of Public Policy & Marke>ng, DOI: 10.1509/jppm.12.151
Dinner, I. et al. (2011). ParCConing Default Effects: why people choose not to choose. Journal of Experimental Psychology: Applied, 17(4), 332. EIA (2009). ResidenCal Energy ConsumpCon Survey. Gourville, J. (1998). The Effect of Temporal Reframing on TransacCon EvaluaCon. Journal of Consumer Research, 24(4), 395-‐408. Kahneman, D., Knetsch, J.L., & Thaler, R.H. (1990) Experimental Tests of the Endowment Effect and The Coase Theorem. Journal of Poli>cal
Economy, 98(6), 1325-‐1348. Kahneman, D. & Tversky, A. (1979) Prospect Theory: An Analysis of Decision under Risk. Econometrica, 47(2), 263-‐91. Lembregts, C., & Pandelaere, M. (2013). Are All Units Created Equal? The Effect of Default Units on Product EvaluaCon. Journal of Consumer
Research. 39, 1275-‐1289. Novemsky, N., et. al. (2007). Preference fluency in choice. Journal of Marke>ng Research, 44(3), 347-‐356. Samuelson, W., & Zeckhauser, R. (1988). Status quo bias in decision making. Journal of risk and uncertainty, 1(1), 7-‐59. Song, H., & Schwarz, N. (2008). If It’s Hard to Read, It’s Hard to Do: Processing Fluency Affects Effort PredicCon and MoCvaCon. Psychological
Science, 19(10), 986-‐988. Thaler, R. H. (1985) Mental AccounCng and Consumer Choice. MarkeCng Science, 4(3), 199-‐214.
Select References