A growing literature studies exogenous shocks in election turnout to illuminate the relationship between turnout and election results. The normative and policy implications of these studies are unclear because an increase in turnout does not necessarily make the election result more representative of the electorate. We introduce a general method to calculate to what extent turnout shocks make election results more or less representative. We apply this to an interesting case, being the 2018 referendum in the Netherlands in which municipal mergers caused a shock in turnout. We find that, for the referendum result to be perfectly representative, only 71% of voters sensitive to the shock should have been exposed to it, instead of the actual 92%.