Altruistic Voting: Warm Glow or Cool Thinking? Information Acquisition and Information Avoidance in a Swiss Popular Vote on Animal Welfare
Last registered on November 13, 2018

Pre-Trial

Trial Information
General Information
Title
Altruistic Voting: Warm Glow or Cool Thinking? Information Acquisition and Information Avoidance in a Swiss Popular Vote on Animal Welfare
RCT ID
AEARCTR-0003551
Initial registration date
November 10, 2018
Last updated
November 13, 2018 1:11 AM EST
Location(s)
Primary Investigator
Affiliation
University of Hamburg
Other Primary Investigator(s)
PI Affiliation
University of Pittsburgh
PI Affiliation
University of Hamburg
PI Affiliation
Toulouse School of Economics
PI Affiliation
University of Vienna
Additional Trial Information
Status
In development
Start date
2018-11-12
End date
2019-09-30
Secondary IDs
Abstract
This research project investigates pure versus impure altruism in voting, with a focus on pre-vote information acquisition. Using the Swiss popular vote on the so-called “Horncow-Initiative” on November 25, 2018, we study one exemplary vote in which the voters themselves are not directly affected in material terms by the voting outcome, while the voting outcome will have externalities on others (in this case, animals). Such votes are liable to so-called “expressive voting” or “moral bias”, i.e., voters express altruism. Altruism can be pure – i.e., consequentialist and based on information about the true effects on those concerned – or impure, i.e., motivated by self-signalling, or, equivalently, by the so-called warm glow. In the latter case, voters are not necessarily motivated to acquire information about true consequences on those affected but use the vote to express, and signal to themselves, that they have morally good intentions. Impure altruism can even cause deliberate information avoidance. A welfare-maximizing approach that puts higher weights on externalities than on ego utility would imply minimizing information avoidance and hence reducing impure altruism. We conduct a survey experiment with a sample of approx. 2000 individuals to investigate the conditions under which voters acquire objective and balanced information, rather than to avoid information.
A second focus of this project is on animal welfare in its own right. We investigate whether consumption behavior that improves animal welfare (such as refraining from meat consumption) and voting in favor of animal-welfare regulation are positively correlated, or whether “altruistic voting” is a compensation for egoistic consumption behavior. Both possibilities are predicted by economic theories on (self-)signalling and moral behavior.

Literature:

Andreoni, J., Rao, J.M., and H. Trachtman (2017): Avoiding the Ask: A Field-Experiment on Altruism, Empathy, and Charitable Giving. Journal of Political Economy 125(3), 625-653.

Arluke, A. and E. Madfis (2014): Animal abuse as a warning sign of school massacres: A critique and refinement. Homicide studies 18(1), 7-22.

Brocas, I., Carillo, J. D., Wang, S. W., and C.F. Camerer (2014): Imperfect choice or imperfect attention? Understanding strategic thinking in private information games. Review of Economic Studies, 81(3), 944-970.

Bruns, H. and G. Perino (2018): Point at, nudge, or push private provisions to a public good? Field experimental evidence for experts, politicians, and nobodies. SSRN Online: https://ssrn.com/abstract=3178597.

Crumpler, H., and P. J. Grossman (2008): An Experimental Test of Warm Glow Giving. Journal of Public Economics 92 (5), 1011–21.

Dittmann, I., Kübler, D., Maug, E., and L. Mechtenberg (2014): Why Votes Have Value: Instrumental Voting with Overconfidence and Overestimation of Others' Errors. Games and Economic Behaviour 84, 17-38.

Erlanger, A. C. E. and S.V. Tsytsarev (2012): The relationship between empathy and personality in undergraduate students’ attitudes toward nonhuman animals. Society and Animals 20(1), 21-38.

Feddersen, T., Gailmard, S., and A. Sandroni (2009): Moral bias in large elections: theory and experimental evidence. American Political Science Review, 103(2), 175-192.

Grossman, Z. and J.J. van der Weele (2017): Self-Image and Willful Ignorance in Social Decisions. Journal of the European Economic Association 15(1), 173-217.

Hesterman, N., Treich, N. S., and Y. L. Yaouanq (2018): An economic model of the meat paradox. Mimeo.

Levin, J., Arluke, A., and L. Irvine (2017): Are People More Disturbed by Dog or Human Suffering?. Society and Animals 25(1), 1-16.

Morton, R., Piovesan, M., and J.R. Tyran (2018): The dark side of the vote: Biased voters, social information, and information aggregation through majority voting. Games and Economic Behaviour, forthcoming.

Tyran, J. R. and A.K. Wagner (2016): Experimental Evidence on Expressive Voting. Forthcoming in Congleton, R., Grofman, B. and S. Voigt (eds.): Oxford Handbook of Public Choice.

External Link(s)
Registration Citation
Citation
Mechtenberg, Lydia et al. 2018. "Altruistic Voting: Warm Glow or Cool Thinking? Information Acquisition and Information Avoidance in a Swiss Popular Vote on Animal Welfare." AEA RCT Registry. November 13. https://doi.org/10.1257/rct.3551-1.0.
Former Citation
Mechtenberg, Lydia et al. 2018. "Altruistic Voting: Warm Glow or Cool Thinking? Information Acquisition and Information Avoidance in a Swiss Popular Vote on Animal Welfare." AEA RCT Registry. November 13. http://www.socialscienceregistry.org/trials/3551/history/37120.
Experimental Details
Interventions
Intervention(s)


Intervention Start Date
2018-11-13
Intervention End Date
2018-12-31
Primary Outcomes
Primary Outcomes (end points)
pre-vote acquisition of information supporting the Horncow Initiative, pre-vote acquisition of information supporting opponents of the Horncow Initiative, pre-vote informedness, reported voting behavior (YES, NO, abstention), consequentialism

Primary Outcomes (explanation)
Information acquisition is measured as the decision to read arguments (of proponents and / or opponents of the Initiative) that are offered to be read in the 1st survey and that remain hidden if the participant chooses not to read them. We also ask a number of test questions in the 1st surveys to estimate the pre-vote degree to which participants are informed about the Horncow Initiative and its possible consequences. Voting behavior will be elicited in the 2nd-wave surveys after the popular vote. In the 2nd surveys, we ask general questions about moral attitudes, in particular whether good intentions and / or good consequences should be rewarded. In addition to the causal hypotheses described in the (hidden) experimental-design section below, we hypothesize that (1) information acquisition and (2) informedness positively correlate with consequentialist attitudes, i.e., the preference for rewarding good consequences rather than (only) good intentions.
Secondary Outcomes
Secondary Outcomes (end points)
consumption of meat and other animal products, prior (belief about) degree of informedness, overconfidence, emotional involvedness, text data from the chat: emotion-based versus information-based arguments
Secondary Outcomes (explanation)
We directly elicit consumption of meat and other animal products in the 1st surveys. Beliefs about own informedness and informedness of the average person are also directly elicited (belief about own degree of informedness, overconfidence). We directly elicit how much the participant is emotionally involved in the issue addressed by the Horncow Initiative. We hypothesize that (1) information acquisition negatively correlates with overconfidence and emotional involvedness. Since economic theories of moral behavior differ with regard to their implication on how past good / bad acts correlate with future good / bad acts, we test two hypotheses on the correlation between the consumption of animal products and voting in favor of the Horncow Initiative: (2) Subjects who refrain more from consumption of animal products are more likely to vote in favor of the Horncow Initiative (moral consistency), or (3) subjects who refrain less from consumption of animal products are more likely to vote in favor of the Horncow Initiative (moral balancing behavior). If we get sufficient text data from the chat, we will also investigate whether those who are classified as consequentialists based on their answers in the 2nd survey also use more information-based arguments than those who are not classified as consequentialists.
Experimental Design
Experimental Design
We implement six versions of one survey prior to the popular vote and two versions of another survey after the popular vote. Together, we therefore implement 12 survey treatments. We vary two different treatment variables: First, we vary (true) information that we give the subjects prior to the vote. Second, we vary the (anticipation of) the opinion of a chat-partner. The online chat of 3-5 minutes is part of the 2nd survey. Survey and chat are incentivized. Treatments are assigned randomly through randomization quota. Subjects are recruited by a professional panel provider.
Experimental Design Details
Informational variation: In all treatments, the information that we give is true. In two versions of the pre-vote survey, we provide information on the positive correlation between empathy towards other humans and empathy toward animals; whereas in two other versions, we give information on empathy toward animals that goes hand in hand with a lack of empathy toward humans. (Both pieces of information are taken from scientific studies.) In the remaining two versions of the pre-vote survey, we do not give any information on the relation between empathy toward humans and empathy toward animals (benchmark treatment). Importantly, information on the relation between the two types of empathy is irrelevant for the objective question how to vote on the Horncow-Initiative if voters are pure altruists. However, we hypothesize that it affects the extent to which voters are willing to acquire information about the true consequences of the voting outcome on those concerned, i.e., on the Horncows. To be more precise, we hypothesize that the first two versions, compared to the other versions, will motivate voters to vote without acquiring information provided by the opponents of the Initiative, since acquiring such information would reduce the self-signalling value of the popular vote on the Initiative (impure altruism). This treatment effect is to be expected since the Initiative declares to be in favor of animal welfare, which, however, is contested by its opponents. We also hypothesize that the second two versions will motivate voters to first acquire more, and balanced, information, in particular including information provided by the opponents of the Initiative, too (pure altruism). In all six pre-vote surveys, we offer subjects the option to read all arguments (both pro and contra the Initiative), or only arguments in favor of the Initiative, or only arguments against, or no argument at all. (All arguments are consequentialist, i.e., concerned with consequences for the animals affected by the vote.) However, since we want to make sure that we do not manipulate the actual information that voters can use to decide how to vote, almost all the arguments that they can decide to read are taken from the so-called “Abstimmungsbüchli” which all Swiss voters receive by post anyways, prior to the vote. (Only one argument is taken from the public debate but not from the Abstimmungsbüchli. This is necessary to balance the arguments in favor and the arguments against the Initiative.) We thus make sure that our informational variation in the pre-vote survey does NOT affect true availability of relevant information. Chat-partner variation: After the vote, we re-contact our subjects and ask them how they voted, and why. At the end of this 2nd survey, subjects are directed to an online chat-platform where they are matched with some other subject who also completed the 1st survey. Part of our subjects will be matched with someone who thought alike on the Horncow-Initiative, and the remaining subjects will be matched with someone who thought differently. All subjects are informed in the 1st survey already that they will be re-contacted. All subjects are informed in the 2nd survey at the latest that they will get the opportunity of chatting with another subject at the end of the 2nd survey. (Thus, someone who does not want to chat can simply decline to do so by completing only the survey, not the chat.) Moreover, we tell two-thirds of the subjects in the 1st survey that their chat partner will be of a similar / different opinion (as implemented). We hypothesize that those who anticipate chatting with someone of the opposite opinion will be more motivated to acquire information than those who anticipate chatting with someone of similar opinion, or those who do not anticipate the chat opportunity during the 1st survey. Prior to the chat, we instruct subjects to use friendly language and not to reveal their identity, nor to ask the chat-partner to reveal his/ hers. In addition, we do not re-contact subjects of extreme opinions and high emotional involvement for the chat. The incentivized chat lasts for 3-5 minutes.
Randomization Method
Randomization is implemented by programming randomization quota within Limesurvey.
Randomization Unit
Individual
Was the treatment clustered?
No
Experiment Characteristics
Sample size: planned number of clusters
2100 individuals
Sample size: planned number of observations
2100
Sample size (or number of clusters) by treatment arms
Since we conduct two waves (prior to the vote and after), we expect to lose some of the subjects from the 1st wave. To be more precise, we expect to have 2100 completes in the 1st wave and 1400 in the 2nd wave. Thus, we expect to have approx. 116 completes per treatment in the end.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
Dean's Office, University of Hamburg
IRB Approval Date
2018-11-07
IRB Approval Number
N/A
Post-Trial
Post Trial Information
Study Withdrawal
Intervention
Is the intervention completed?
No
Is data collection complete?
Data Publication
Data Publication
Is public data available?
No
Program Files
Program Files
Reports and Papers
Preliminary Reports
Relevant Papers