Tit-for-Tat voting by contestants in the TV quiz-show ‘The Weakest Link’

Goddard, Paul, Hylton, Patrick, Parke, Adrian and Noh, Zamira (2013) Tit-for-Tat voting by contestants in the TV quiz-show ‘The Weakest Link’. In: SABE/IAREP/ICABEEP 2013 Conference in Atlanta, GA,, 25-29 July 2013, Atlanta, GA.

Tit-for-Tat voting by contestants in the TV quiz-show ‘The Weakest Link’
__ddat03_staffhome_pgoddard_RESEARCH_Ongoing projects_weakest link_talks_Atlanta 2013_TFT amended atlanta abstract.pdf

Item Type:Conference or Workshop contribution (Paper)
Item Status:Live Archive


Background It has not escaped the notice of researchers that TV quiz-shows like ‘The Weakest Link’ (WL) make
ideal observational field experiments because they comprise the key ingredients of game theory: a finite group of
players must select from a fixed set of actions to play for well defined payoffs. For example, WL has been used
to assess the optimal banking strategy in economic decision making (Haan, Los & Riyanto, 2011), the trade-off
between risk and return strategies in game playing (Février & Linnemer, 2006; Barmish & Boston, 2009), as a
test of gender and race discrimination in voting (Levitt, 2004; Antonovics, Arcidiacono & Walsh, 2005;
Goddard, 2012) and to demonstrate ‘neighbour’ effects in voting practice (Goddard, Ashley & Hunter, 2011).
Research Questions:- We tested for three kinds of voting bias by players of WL. i.) spatial, ii.) gender and iii.)
‘Tit-for-Tat’ (TFT).
Methodology-i.) Rules of WL:- A group of players (n=9) accumulated a pot of money by fielding a first round of
questions. Next, each player identified one of their fellows as the ‘weakest’ in that round. The player accruing
the majority of votes was summarily eliminated from the show. A second accumulation round of questions
preceded another elimination vote, and so on, until the group was whittled down to the final pair, who then
played out a tie-breaker to determine an outright winner. Methodology- ii.) Analysis:- The observed frequencies
of votes cast in the first and second rounds of 72 episodes of WL were recorded. Simple probability theory was
then used to calculate the corresponding expected frequencies due to chance. Significant departures from these
expected patterns, identified by χ2 tests, indicated voting bias.
Findings:- TFT voting occurred when recipients of round 1 votes responded in kind by voting for the perpetrator
in round 2. TFT votes occurred significantly more often than expected, and, significantly more often than those
made by the equivalent controls who had not received a vote in round 1. Spatial and gender biases were found:
players avoided voting for direct neighbours and females received significantly more votes than males.
Interpretation:-We suggest that TFT was played as a deliberate, explicit strategy, but, spatial/gender voting
anomalies emerged implicitly. To elaborate, we suggest that a player’s voting decision was informed by two
sources of information: situational, the game-specific, public performance of the other players, and,
dispositional, their individual, internal, subjective-dependent attributions. In rounds where situational
information was unequivocal, so the weakest player was easily identified by the other players (hi-consensus),
there was no voting bias. However, significant biases emerged as uncertainty increased (consensus decreased)
about the identity of the weakest player. In the absence of clear-cut situational information, because all players
performed equally well (or badly!), players resorted to their private, bias-prone dispositional information source.
Conclusion:- The format of WL quiz-shows provided an ideal context to analyse forced-choice decision making
and the implicit biases and explicit strategies therein.

Keywords:neighbour effect, compulsory voting, weakest link
Subjects:C Biological Sciences > C880 Social Psychology
L Social studies > L110 Applied Economics
Divisions:College of Social Science > School of Psychology
Related URLs:
ID Code:16869
Deposited On:04 Mar 2015 14:49

Repository Staff Only: item control page