Abstract
Committee protocols typically involve a deliberation stage in which members try to influence and convince other regarding the “right” decision. Beyond information exchange, such deliberations also aim to affect the preferences and the votes of other members. Using a model of social influence, we demonstrate how deliberation procedures affect the voting outcome and how different protocols of consultation by committees’ chairs may affect their decisions. We then analyze the ability of a “designer” to control the deliberation protocol and to manipulate the deliberation procedure to increase the probability that the outcome he favors will be selected.
Similar content being viewed by others
Notes
In the first round the referees cannot influence each other. In many journals, however, when there is a second round of “revise and resubmit,” referees see each other’s first round reports, and may therefore be influenced by them.
This is how papers are accepted for presentation in some computer science conferences, e.g., the ACM Conferences on Economics and Computation.
Condorcet (1785) showed that the quality of decisions made by committees with diverse information is increasing with the group size.
See also the survey by List (2018).
Note that there is some analogy between these two properties since by proper representation of our setting we can transform \(\alpha\)—the weight of the second attribute—to \({\hat{\alpha }}\) which will be the weight of the first attribute. In this case, whenever \(g(\alpha ,\alpha )>\alpha\), we would have \(g({{\hat{\alpha }}},{{\hat{\alpha }}})<{{\hat{\alpha }}}\) so SR with respect to the second attribute implies UR with respect to the first one.
Our setup assumes that g depends on one’s core preferences and the average behavioral preferences of the rest without specifying how many other individuals there are in the influence group. One can modify this assumption by indexing the g function according to the number of individuals in the influence group. Claims 4 and 6 below only require that the vector \(\beta _{-i}\) can be replaced with a parameter \(\beta \in \left[ \min _{j\ne i}\beta _j,\,\max _{j\ne i}\beta _j\right]\).
As an alternative, we can let \(\Gamma\) be an \(n\times n\) adjacency matrix, with entry \(\Gamma _{ij}\in \{0,1\}\) denoting whether i interacts with player j. In this formulation, one can permit \(\Gamma _{ij}\in [0,1]\), indicating intensities of influence, so it is possible to consider committees in which some individuals have a greater social influence than others.
We can extend the existence claim to the case of a general (bounded and equi-Lipschitz) preferences over [0, 1] following the same step of the proof in FS.
This proof is stronger than the claim itself, as it shows that it is possible that prior to the deliberation all members favor one candidate but as a result of the deliberation all of them favor the other candidate.
For example, all members are expected utility maximizers with the vNM utility \(\alpha u+(1-\alpha ){\tilde{u}}\), where u is a concave transformation of \({\tilde{u}}\).
The case where person 2 does not participate is more involved and the analysis depends on whether \(\beta _2\) is above or below the average of \(\beta _1\) and \(\beta _3\).
We can put a higher weight on \(\beta _0\) if the opinion of the chair is more influential.
References
Alger, I., & Weibull, J. W. (2013). Homo moralis—preference evolution under incomplete information and assortative matching. Econometrica, 81, 2269–2302.
Ariely, D., & Levav, J. (2000). Sequential choice in group settings: Taking the road less traveled and less enjoyed. Journal of Consumer Research, 27, 279–90.
Aronson, E., Wilson, T. D., & Akert, R. M. (2010). Social Psychology (7th ed.). Upper Saddle River: Prentice Hall.
Austen-Smith, D., & Feddersen, T. J. (2006). Deliberation, preference uncertainty, and voting rules. American Political Science Review, 100, 209–217.
Becker, G. S. (1970). Altruism, egoism and genetic fitness: Economics and sociobiology. Journal of Economic Literature, 14, 817–826.
Billot, A., & Qu, X. (2022). Deliberative democracy and utilitarianism. Social Choice and Welfare. https://doi.org/10.1007/s00355-022-01404-8
Chan, J., Lizzeri, A., Suen, W., & Yariv, L. (2018). Deliberating collective decisions. The Review of Economic Studies, 85, 929–963.
Charness, G. & Kuhn, P. (2011). “Lab labor: What can labor economists learn in the lab?,” Handbook of Labor Economics, Volume 4a, 229–330.
Condorcet, M.J.A.N. de Caritat, 1785. “An essay on the application of analysis to the probability of decisions rendered by a plurality of voters,” abridged and translated in I. McLean and A.B. Urken, eds. Classics of Social Choice, 1995. Ann Arbor: University of Michigan Press.
Coughlan, P. J. (2000). In defense of unanimous jury verdicts: Mistrials, communication, and strategic voting. American Political Science Review, 94, 375–393.
Cuhadaroglu, T. (2017). Choosing on influence. Theoretical Economic, 12, 477–492.
Damiano, E., Li, H., & Suen, W. (2009). “Delay is strategic information aggregation,” discussion paper, University of Toronto.
Dawkins, R. (1976). The Selfish Gene. Oxford University Press.
DeGroot, Morris H. (1974). Reaching a Consensus. Journal of the American Statistical Association, 69, 118–121.
Dekel, E., & Piccione, M. (2000). Sequential voting procedures in symmetric binary elections. The Journal of Political Economy, 108, 34–55.
Dekel, E., & Piccione, M. (2014). The strategic disadvantage of voting early. American Economic Journal: Microeconomics, 6, 162–179.
Dietrich, F. & Kai, S. (2022). “Deliberation and the Wisdom of Crowds,” https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4145408.
Dietrich, F. & Kai, S. (2022). “Deliberation and the Wisdom of Crowds.”
Ding, H., & Pivato, M. (2021). Deliberation and epistemic democracy. Journal of Economic Behavior and Organization, 185, 138–167.
Dryzek, J. S., & List, C. (2003). Social choice theory and deliberative democracy: A reconciliation. British Journal of Political Science, 33, 1–28.
Farrar, C., Green, D. P., Green, J. E., Nickerson, D. W., & Shewfelt, S. (2009). Does group composition affect policy preferences? Results from three randomized experiments. Political Psychology, 30, 615–647.
Feddersen, T., & Pesendorfer, W. (1998). Convicting the innocent: The inferiority of unanimous jury verdicts under strategic voting. American Political Science Review, 92, 23–35.
Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition and Cc-operation. Quarterly Journal of Economics, 114, 817–868.
Fershtman, C., & Segal, U. (2018). Preferences and social influence. American Economic Journal: Microeconomics, 10, 124–142.
Gersbach, H., & Hahn, V. (2012). Information acquisition and transparency in committees. International Journal of Game Theory, 41, 427–453.
Golub, B., & Jackson, M. O. (2010). Naïve Learning in Social Networks and the Wisdom of Crowds. American Economic Journal: Microeconomics, 2, 112–49.
Hoff, K., & Stiglitz, J. E. (2016). Striving for balance in economics: Towards a theory of the social determination of behavior. Journal of Economic Behavior and Organization, 126, 25–57.
Isenberg, D. J. (1986). Group polarization: A critical review and meta-analysis. Journal of Personality and Social Psychology, 50, 1469–1476.
Levy, G. (2007). Decision making in committees: Transparency, reputation and voting rules. American Economic Review, 97, 150–168.
List, C. (2018). “Democratic deliberation and social choice: A review,” in The Oxford Handbook of Deliberative Democracy. Bächtiger, A., J. Dryzek, J. Mansbridge, and M. Warren, eds. Ch. 29 pp. 463–489. Oxford: Oxford University Press.
Li, H., & Suen, W. (2009). Decision making in committee. Canadian Journal of Economics, 42, 359–392.
Mendelberg, T. (2006). Small group deliberation. Philadelphia, PA: Presented at the Annual meeting of the American Political Science Association.
Moldovanu, B., & Shi, X. (2013). Specialization and partisanship in committee search. Theoretical Economics, 8, 751–774.
Myers, D. (1982). “Polarizing effects of social interaction,” In H. Brandstatter, J. Davis, and G. Stocker-Kreichbauer (Eds.), Group Decision Making pp. 125–161. New York: Academic Press.
Myers, D. G. (1975). Discussion-induced attitude polarization. Human Relations, 28, 699–714.
Myers, D. G., & Lamm, H. (1976). The group polarization phenomenon. Psychological Bulletin, 83, 602–627.
Perote-Pena, J., & Piggins, A. (2015). A model of deliberative and aggregative democracy. Philosophy and Economics, 31, 93–121.
Schade, D., Sunstein, C. R., & Kahneman, D. (2000). Deliberation about dollars: The severity shift. California Law Review, 100, 1139–1175.
Sobel, J. (2005). Interdependent preferences and reciprocity. Journal of Economic Literature, 43, 396–440.
Visser, B., & Swank, O. H. (2007). On committees of experts. Quarterly Journal of Economics, 115, 967–994.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Declarations
The authors did not receive support from any organization for the submitted work and no funding was received to assist with the preparation of this manuscript or with conducting this study.
Conflict of interest
The authors have no competing interests to declare that are relevant to the content of this article. Moreover, all authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or non-financial interest in the subject matter or materials discussed in this manuscript.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix: Proofs
Appendix: Proofs
Proof of Claim 1:
If g is monotonic in both arguments and \(g(\alpha ,\alpha )\equiv \alpha\) then for \(\alpha <\beta\), \(\alpha =g(\alpha ,\alpha )\leqslant g(\alpha ,\beta )\leqslant g(\beta ,\beta )=\beta\). If, on the other hand, for some \(\alpha\), \(g(\alpha ,\alpha )<\alpha\), then by continuity for sufficiently small \(\varepsilon >0\), \(g(\alpha ,\alpha +\varepsilon )<\alpha\). Likewise, if \(g(\alpha ,\alpha )>\alpha\) then \(\alpha <g(\alpha -\varepsilon ,\alpha )\), a contradiction. \(\square\)
Proof of Claim4:
If committee members do not deliberate among themselves, then there is no change in their preferences and therefore \(\beta ^{cb}=\alpha\) and \(\beta _0^{cb}=g(\alpha _0,\alpha )\). When the n agents deliberate among themselves their equilibrium behavioral preference is given by \(\beta ^{cm}=g(\alpha ,\beta ^{cm})\). By Claim 6 in FS, \(\beta ^{cm}=g(\alpha ,\beta ^{cm})>\alpha\) if and only if \(g(\alpha ,\alpha )>\alpha\). And since \(g_2>0\), \(\beta _0^{cm}=g(\alpha _0,\beta ^{cm})>g(\alpha _0,\alpha )=\beta _0^{cb}\). \(\square\)
Proof of Claim 5:
We start with the case db in which the chair deliberates with each adviser separately. Consider the behavioral preferences of each adviser which is a function of his core preferences and the behavioral preferences of the chair, \(\beta (\alpha ,\beta _0)\equiv g(\alpha ,\beta _0)\). Given that \(g_2>0\), this function is increasing in \(\beta _0\). Also, \(\beta _0(\alpha _0,\beta )\equiv g(\alpha _0,\beta )\) are the behavioral preferences of the chair given his core preferences and the behavioral preferences of the advisers. Figure 1 depicts both functions in a \((\beta \times \beta _0)\) space where \(\alpha ,\alpha _0\) are fixed parameters. The intersection \((\beta ^{db}\),\(\beta _0^{db})\) (point s in Fig. 1) is an equilibrium of the bilateral deliberation process as \(\beta ^{db}=g(\alpha ,\beta _0^{db})\) and \(\beta _0^{db}=g(\alpha _0,\beta ^{db})\). Note also that since \(\alpha <\alpha _0\), it follows that the equilibrium point is above the \(45^{\circ }\) line because in equilibrium \(\beta ^{db}<\beta _0^{db}\) (see point (ii) at the end of Sect. 2.2).
Examine now the equilibrium of the multilateral deliberation process. The function \(\beta _0(\alpha _0,\beta )\) is the same as before. The function that determines the behavioral preferences of the advisers given the behavioral preferences of the chair is given now by \(\beta =g(\alpha ,\frac{\beta _0+(n-1)\beta }{n})\).Footnote 13 This curve is a left (anticlockwise) rotation of the curve \(\beta (\alpha ,\beta _0)\) around the symmetric point r (which is the intersection of the \(45^{\circ }\) line and the \(\beta (\alpha ,\beta _0)\) curve). The reason is that when \(\beta =\beta _0\), the two functions imply the same behavioral preferences. It is a left rotation because for all the points above the \(45^{\circ }\) line \(\beta <\beta _0\) and therefore the function for the multilateral procedure yields a lower value. The intersection q of the two curves, \(\beta _0(\alpha _0,\beta )\) and \(\beta =g(\alpha ,\frac{\beta _0+(n-1)\beta }{n})\) yields the equilibrium for the multilateral deliberation case. As depicted in Figure 1, comparing the two equilibrium points yields that \(\beta _0^{db}>\beta _0^{dm}\). \(\square\)
Proof of Claim 6:
Under the consultation procedure, \(\beta ^{cb}=\alpha\) and \(\beta _0^{cb}=g(\alpha _0,\alpha )\), while under the deliberation procedure \(\beta ^{db}=g(\alpha ,\beta _0^{db})\) and \(\beta _0^{db}=g(\alpha _0,\beta ^{db})\). We show first that \(\beta ^{db}>\alpha\). If the core preference of the chair is \(\alpha\), then the advisers and the chair have the same core preferences and consequently the same behavioral preferences \(\beta\). Given our SR assumption, \(\beta\) must be larger than \(\alpha\). They can obviously not equal \(\alpha\), because \(g(\alpha ,\alpha )>\alpha\). If \(\beta <\alpha\), then since \(g(\alpha ,\alpha )>\alpha\) and \(g(\alpha ,\beta )=\beta\), we get
In contradiction to the assumption that \(g_2<1\). As \(\alpha _0>\alpha\), the \(\beta\) value of the chair is higher than when his core preferences are \(\alpha\), and consequently, so is the \(\beta\) values of the advisers.
Since \(\beta _0^{b0}=g(\alpha _0,\beta ^{db})\) and \(\beta ^{db}>\alpha\), it follows that \(\beta _0^{db}>g(\alpha _0,\alpha )=\beta _0^{cb}\), hence the claim. \(\square\)
Proof of Claim 7:
If person 1 does not participate in the deliberation, then \(\beta ^1_1=\alpha _1\). We show first that \(\beta ^1_2>\beta _2\) and \(\beta ^1_3>\beta _3\). Observe that by Claim 8 in FS, \(\beta ^1_2>\alpha _2\). If \(\beta _2\leqslant \alpha _2\), then clearly \(\beta ^1_2>\beta _2\), and since \(\beta _2 >\beta _1\),
Suppose that \(\beta _2>\alpha _2\) but \(\beta ^1_2\leqslant \beta _2\). Since by FS (2018) \(\beta _1<\beta _3\) (see end of Sect. 2.2),
Also, since \(g_2<1\),
Similarly, using inequality (2)
Combining inequalities (3) and (4) together and recalling that \(\beta _1<\beta _2\), we get
A contradiction, hence \(\beta ^1_2>\beta _2\). And since \(\beta _1<\beta _2\), it follows that \(\beta ^1_2>\frac{1}{2}[\beta _1+\beta _2]\), hence \(\beta ^1_3>\beta _3\). It thus follows that both unanimity rule (determined by person 3) and majority (determined by person 2) rule accept less projects than the case in which all members participate in the deliberation.
Suppose now that person 3 does not participate. Then by Claim 8 in FS, \(\beta ^3_3=\alpha _3>\beta _3\) and the unanimity rule will accept less projects. Since \(\beta ^3_2>\beta ^3_1\) (Claim 7 in FS), in order to show that the majority rule will accept more project it is enough to show that \(\beta _2>\beta ^3_2\). Since by the aforementioned claim, \(\alpha _2>\beta ^3_2\), this is clearly the case when \(\beta _2\geqslant \alpha _2\). We therefore prove the impossibility of \(\alpha _2>\beta ^3_2>\beta _2\). Otherwise,
Since \(g_2<1\), we get
A contradiction to the fact that \(\beta _3>\beta _2>\beta _1\). \(\square\)
Proof of claim 8:
We use the following table
Person B is selected by a unanimity rule if and only if \(M:=\min \{\beta _1,\beta _2,\beta _3\}\geqslant \gamma\).
- (i):
-
1-2-3: For \(i=1,2,3\), \(\beta ^{(v)}_i\geqslant \beta ^{(i)}_1\), hence \(M^{(v)}\geqslant M^{(i)}\).
- (ii):
-
1-3-2: For \(i=1,2,3\), \(\beta ^{(v)}_i\geqslant \beta ^{(ii)}_1\), hence \(M^{(v)}\geqslant M^{(ii)}\).
- (iii):
-
2-1-3: For \(i=1,2,3\), \(\beta ^{(iv)}_i\geqslant \beta ^{(iii)}_1\), hence \(M^{(iv)}\geqslant M^{(iii)}\).
- (iv):
-
2-3-1: For \(i=1,3\), \(\beta ^{(v)}_i>\beta ^{(iv)}_i\). Consider two cases—
- 1.:
-
\(g(\alpha _1,\alpha _3)<\alpha _2\). Then \(\beta ^{(v)}_1=g(\alpha _1,\alpha _3)=g(g(\alpha _1,\alpha _3),g(\alpha _1,\alpha _3))<g(\alpha _2,g(\alpha _1,\alpha _3))\leqslant g(\alpha _2,\theta \alpha _3+(1-\theta )g(\alpha _1,\alpha _3))=\beta ^{(v)}_2<\beta ^{(v)}_3\), hence \(M^{(v)}=\beta ^{(v)}_1>\beta ^{(iv)}_1\geqslant M^{(iv)}\).
- 2.:
-
\(g(\alpha _1,\alpha _3)\geqslant \alpha _2\). Then \(\beta ^{(v)}_2\geqslant \beta ^{(iv)}_2\) and since for \(i=1,2,3\), \(\beta ^{(v)}_i\geqslant \beta ^{(iv)}_i\), it follows that \(M^{(v)}\geqslant M^{(iv)}\).
- (vi):
-
3-2-1: \(\beta ^{(vi)}_3>\beta ^{(vi)}_2>\beta ^{(vi)}_1\), hence \(M^{(vi)}=\beta ^{(vi)}_1\). Obviously \(M^{(vi)}\leqslant \beta ^{(v)}_1<\beta ^{(v)}_3\) and by definition, \(M^{(vi)}\leqslant \beta ^{(vi)}_1\). We show next that for all \(\alpha _2\in [\alpha _1,\alpha _3]\), \(\beta ^{(vi)}_1\leqslant \beta ^{(v)}_2\). For \(\alpha _2=\alpha _1\) we get \(\beta ^{(vi)}_1=\beta ^{(v)}_2\). Differentiate both with respect to \(\alpha _2\):
$$\begin{aligned} \frac{\partial \beta ^{(vi)}_1}{\partial \alpha _2}= & {} g_2(\alpha _1,\theta \alpha _3+(1-\theta )g(\alpha _2,\alpha _3))\times (1-\theta )g_1(\alpha _2,\alpha _3)\\ \frac{\partial \beta ^{(v)}_2}{\partial \alpha _2}= & {} g_1(\alpha _2,\theta \alpha _3+(1-\theta )g(\alpha _1,\alpha _3)) \end{aligned}$$We assumed that \(g_{12}<0\) (see end of Sect. 2.2), hence \(g_1(\alpha _2,\alpha _3)\leqslant g_1(\alpha _2,\theta \alpha _3+(1-\theta )g(\alpha _1,\alpha _3))\) and since \(g_2(\cdot ,\cdot )\leqslant 1\), \(\frac{\partial \beta ^{(vi)}_1}{\partial \alpha _2}\leqslant \frac{\partial \beta ^{(v)}_2}{\partial \alpha _2}\), implying \(\beta ^{(vi)}_1\leqslant \beta ^{(v)}_2\) for all \(\alpha _2\in [\alpha _1,\alpha _3]\). As \(M^{(vi)}\leqslant \beta ^{(v)}_i\) for \(i=1,2,3\), it follows that \(M^{(vi)}\leqslant M^{(v)}\).
Person B is selected by a majority rule if and only if L, the mid-value of \(\beta _1,\beta _2,\beta _3\) satisfies \(L\geqslant \gamma\). Using the above table, we get
- (i):
-
1-2-3: For \(i=1,2,3\), \(\beta ^{(vi)}_i>\beta ^{(i)}_i\), hence \(L^{(vi)}>L^{(i)}\).
- (ii):
-
1-3-2: For \(i=1,2,3\), \(\beta ^{(vi)}_i>\beta ^{(ii)}_i\), hence \(L^{(vi)}>L^{(ii)}\).
- (iii):
-
2-1-3: \(L^{(iii)}=\min \{\beta ^{(iii)}_2,\beta ^{(iii)}_3\}\). As \(\beta ^{(vi)}_2>\beta ^{(iii)}_2\) and \(\beta ^{(vi)}_3>\beta ^{(iii)}_3\), it follows that \(L^{(vi)}>L^{(iii)}\).
- (iv):
-
2-3-1: \(L^{(vi)}=\beta ^{(vi)}_2>\max \{\beta ^{(iv)}_1,\beta ^{(iv)}_2\}\geqslant L^{(iv)}\).
- (v):
-
3-1-2: \(L^{(vi)}=\beta ^{(vi)}_2>\max \{\beta ^{(v)}_1,\beta ^{(v)}_2\}\geqslant L^{(v)}\).
\(\square\)
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Fershtman, C., Segal, U. Social influence in committee deliberation. Theory Decis 96, 185–207 (2024). https://doi.org/10.1007/s11238-023-09949-5
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11238-023-09949-5