Abstract
This paper presents a novel combinatorial approach for voting rule analysis. Applying reversal symmetry, we introduce a new class of preference profiles and a new representation (bracelet representation) of preference profiles. By applying an impartial, anonymous, and neutral culture model for the case of three alternatives, we obtain precise theoretical values for the number of election scores for the plurality rule, the Kemeny rule, the Borda rule, and the scoring rules in the extreme case.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Aleskerov FT, Chistyakov VV, Kalyagin VA (2010) Social threshold aggregations. Social Choice Welfare 35(4):627–646
Boutilier C, Rosenschein JS (2016) Incomplete information and communication in voting. In: Moulin H, Brandt F, Conitzer V, Endriss U, Lang J, Procaccia A (eds) Handbook of computational social choice. Cambridge University Press, Cambridge, pp 223–258
Bubboloni D, Gori M (2015) Symmetric majority rules. Math Social Sci 76:73–86
Bubboloni D, Gori M (2016) On the reversal bias of the Minimax social choice correspondence. Math Social Sci 81:53–61
Conitzer V, Sandholm T (2005) Communication complexity of common voting rules. In: Proceedings of the ACM conference on electronic commerce, pp 78–87
Crisman KD (2014) The Borda count, the Kemeny rule, and the permutahedron. In: Crisman KD, Jones MA (eds) The mathematics of decisions, elections, and games, Contemporary Mathematics, vol 624. American Mathematical Society, Providence, pp 101–134
De Donder P, Le Breton M, Truchon M (2000) Choosing from a weighted tournament. Math Soc Sci 40(1):85–109
Egecioglu Ö, Giritgil AE (2013) The impartial, anonymous, and neutral culture model: a probability model for sampling public preference structures. J Math Sociol 37(4):203–222
Fishburn PC (1977) Condorcet social choice functions. SIAM J Appl Math 33(3):469–489
Fitzsimmons Z, Hemaspaandra E (2018) Election score can be harder than winner. arXiv:1806.08763
Gehrlein WV, Fishburn PC (1976) The probability of the paradox of voting: a computable solution. J Econ Theory 13:14–25
Gehrlein WV, Fishburn PC (1979) Proportions of profiles with a majority candidate. Comput Math Appl 5:117–124
Gehrlein WV, Lepelley D (2011) Voting paradoxes and group coherence: the condorcet efficiency of voting rules. Studies in choice and welfare. Springer, Berlin
Gehrlein WV, Lepelley D (2017) Elections, voting rules, and paradoxical outcomes. Studies in choice and welfare. Springer, Berlin
Goldsmith J, Lang J, Mattei N, Perny P (2014) Voting with rank dependent scoring rules. In: Proceedings of the 28th AAAI conference on artificial intelligence (AAAI’14). AAAI Press, pp 698–704
Guilbaud GT (1952) Les théories de l’intérét général et le problème logique de l’agrégation. Economie Appliquée 5:501–584
Hashemi V, Endriss U (2014) Measuring diversity of preferences in a group. Series frontiers in artificial intelligence and applications. Ebook Vol. 263: ECAI 2014, pp 423–428. https://doi.org/10.3233/978-1-61499-419-0-423
Hoskins WD, Penfold Street A (1982) Twills on a given number of harnesses. J Austr Math Soc Ser A 33(1):1–15
Karpov A (2017) Preference diversity orderings. Group Decision Negotiation 26(4):753–774
Karpov A (2019) On the number of group-separable preference profiles. Group Decision Negotiation 28(3):501–517
Lackner ML, Lackner M (2017) On the likelihood of single-peaked preferences. Social Choice Welfare 48(4):717–745
MacMahon PA (1918) Combinations derived from m identical sets of n different letters and their connexion with general magic squares. Proc Lond Math Soc s2–17(1):25–41
May K (1948) Probabilities of certain election results. Am Math Monthly 55(4):203–209
OEIS. The on-line encyclopedia of integer sequences. Published electronically at http://oeis.org
Pérez-Fernández R, De Baets B (2017) Recursive monotonicity of the Scorix: Borda meets Condorcet. Group Decision Negotiation 26(4):793–813
Saari DG, Barney S (2003) Consequences of reversing preferences. Math Intell 25(4):17–31
Sato S (2009) Informational requirements of social choice rules. Math Social Sci 57:188–198
Sato S (2016) Informational requirements of social choice rules to avoid the Condorcet loser: a characterization of the plurality with a runoff. Math Social Sci 79:11–19
Stanley RP (2012) Enumerative combinatorics. Vol. 1. Cambridge University Press, Cambridge
Veselova Y (2016) The difference between manipulability indices in the IC and IANC models. Social Choice Welfare 46(3):609–638
Acknowledgements
The author would like to thank Vincent Merlin, Dominik Peters, Stefan Napel, Fuad Aleskerov, Aleksei Kondratev, Yuliya Veselova, Daniel Karabekyan, and two anonymous reviewers for their valuable comments, and Zaruhi Hakobyan for Latex text processing. The author is also thankful to the seminar audience of the 8th Murat Sertel Workshop (held at the Université de Caen Normandie, during May 22–23, 2018), the 14th Meeting of the Society for Social Choice and Welfare (held at Seoul, Korea during June 14-17, 2018). An earlier version of this paper circulated under the title “An Informational Basis for Voting Rules”. The paper was prepared within the framework of the Basic Research Program at the National Research University Higher School of Economics (HSE) and supported within the framework of a subsidy by the Russian Academic Excellence Project “5–100”. The work was conducted by the International Laboratory of Decision Choice and Analysis (DeCAn Lab) of the National Research University Higher School of Economics.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
5 Appendix
5 Appendix
Proof of proposition 5
For \(m=3\), a homogenous preference profile (n preference orders \(P_0\)) has vector of rank sums \(\begin{pmatrix}2n\\ n\\ 0\end{pmatrix}\).Any preference profile has vector of ranks sums either \(\begin{pmatrix}2n\\ n\\ 0\end{pmatrix}\) or vectors \(\alpha \),\(\beta \),\(\gamma \),\(\delta \),\(\epsilon \), defined here:
For all vectors \(\alpha \),\(\beta \),\(\gamma \),\(\delta \),\(\epsilon \), we have that the first component is not less than the second; the second component is not less than the third. Substituting x preference orders from \(P_0\) to \(P_5\), we obtain type \(\alpha \) preference profile. There are \(\lfloor \frac{n}{2}\rfloor \) preference profiles of type \(\alpha \).Similarly, we have \(\lfloor \frac{n}{2}\rfloor \) voting situations of type \(\beta \).
Substituting \(\frac{y}{2}\) preference orders from \(P_0\) to \(P_3\) and \(x-y\) preference orders from \(P_0\) to \(P_1\), we obtain type \(\gamma \) voting situations with even y.Substituting \(\frac{(y-1)}{2}\) preference orders from \(P_0\) to \(P_3\), \(x-y-1\) preference orders from \(P_0\) to \(P_1\), and one preference order from \(P_0\) to \(P_2\), we obtain type \(\gamma \) voting situations with odd y.
We can construct all possible type \(\gamma \) voting situations using these two design methods. There are two natural restrictions on x,y for saving the order of alternatives:
From this, we find the number of type \(\gamma \) voting situations (if \(n \ge 4\)):
for \(y \le \frac{x}{2}\), it is \(\sum _{i=2}^{\lfloor \frac{n}{2}\rfloor }p_2(i)+\sum _{i=\lfloor \frac{n}{2}\rfloor +1}^{\lfloor \frac{2n}{3}\rfloor }p_{2,n-i}(i)\);
for \(y \ge \frac{x}{2}\), it is \(\sum _{i=2}^{\lfloor \frac{2n}{3}\rfloor }p_2(i) + \sum _{i=\lfloor \frac{2n}{3}\rfloor +1}^{n-1}\left[ p_2(i)-p_{2,2i-n-1}(i)\right] \);
for \(y = \frac{x}{2}\), it is \(\lfloor \frac{\lfloor \frac{2n}{3}\rfloor }{2}\rfloor \).
Thus, the number of type \(\gamma \) voting situations is equal to
It is also the number of type \(\delta \) voting situations.
Substituting \(\frac{x}{2}\) preference orders from \(P_0\) to \(P_3\), we obtain type \(\epsilon \) voting situations with even x. Substituting \(\frac{(x-1)}{2}\) preference orders from \(P_0\) to \(P_3\), one preference order from \(P_0\) to \(P_1\), and one preference order from \(P_0\) to \(P_5\), we obtain type \(\gamma \) voting situations with odd x.
We can construct all possible type \(\epsilon \) voting situations using these two design methods. There is one restriction on x for saving the order of alternatives:
If \(n \ge 2\) then the number of type \(\epsilon \) voting situations is equal to n.
Summing over all types, we have (if \(n\ge 4\))
Modifying this, we obtain
Calculating the sums, we obtain the result, which is also correct for \(n=2\), and \(n=3\).
Proof of proposition 7
Let \(f_0(k){:}\;\mathbb {N}_0 \rightarrow \mathbb {N}_0\) be the sum of distances between preference order \(P_0\) and preference order \(P_i\), where \(i \equiv k\) (mod 6). Let \(g_0 (k)=f_0 (k)+f_0 (k+2)\) and \(h_0 (k)=(2g_0 (k)-g_0 (k+3))/6=(2f_0 (k)+2f_0 (k+2)-f_0 (k+3)-f_0 (k+5))/6\). This transformation is presented in Fig. 3a. The first circle is the circle of distances from order \(P_0\) to all six preference orders. After transformation, each preference order is presented by three subsequent ones. This transformation has the inversion presented in Fig. 3b. We design one-to-one correspondence between \(f_0 (k)\) and \(h_0 (k)\).
In the same fashion, we define functions \(f_j(k)\),\(g_j (k)\),\(h_j (k)\), \(j=\bar{0,5}\). Instead of summing distances from preference orders \(\sum _{j=0}^5f_j(k)n_j\), we sum transformed values \(\sum _{j=0}^5h_j(k)n_j\). We have bisection between these sums. Any sum of transformed values \(\sum _{j=0}^5h_j(k)n_j\) can be represented by the circle presented in Fig. 4.
Permutating one pair of alternatives (one or three swaps in a preference order) leads to Fig. 4 circle turnover. Two possibilities of permutating three alternatives (two swaps in a preference order) lead to the circle rotating. Reversing the preference profile leads to a rotating on 3 preference orders in a clockwise manner. By these operations we can always construct a circle, such that \(x+y+z\le 3n-x-y-z\) and \(x\ge y \ge z\). From this definition, we have the following additional restrictions on x,y,z
From these inequalities, we have \(n \le x+y+z \le \lfloor \frac{3n}{2}\rfloor \). We will calculate the number of partitions of \(i = \bar{n,\lfloor \frac{3n}{2}\rfloor }\) with 1,2, or 3 parts, such that each part does not exceed n (and the special case of \(i = \frac{3n}{2}\)).A complementary partition has sum \(3n-i\).
If \(x+y+z = \frac{3n}{2}\), then the partitions of \(i = \frac{3n}{2}\) and \(3n-i=\frac{3n}{2}\) can be the same. Partitions with \(y= \frac{n}{2}\) are are symmetric (\(n-x=z\), \(n-z=x\), \(n-y=y\)). The number of such partitions is equal to
For all other i, we have
For \(n \ge 3\), we have
Summing (*) for even n and (**) for all n, we obtain the result, which is also correct for \(n=1\) and \(n=2\).
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Karpov, A. (2021). Combinatorics of Election Scores. In: Diss, M., Merlin, V. (eds) Evaluating Voting Systems with Probability Models. Studies in Choice and Welfare. Springer, Cham. https://doi.org/10.1007/978-3-030-48598-6_15
Download citation
DOI: https://doi.org/10.1007/978-3-030-48598-6_15
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-48597-9
Online ISBN: 978-3-030-48598-6
eBook Packages: Economics and FinanceEconomics and Finance (R0)