Now there's some evidence to the contrary. Well, actually the first edition of the paper was published in 2001, so maybe 'now' is a bad word to use, but anyway I've never seen it before so it's news to me. From Peter Kurrild-Klitgaards paper presented at The Annual Meeting of the European Public Choice Society: Voting paradoxes under proportional representation: Evidence from eight Danish elections:

The present analysis [...] suggests that some of the social choice paradoxes potentially occurring in proportional election systems actually do occur in reality. Specifically, one social choice paradox is present in virtually all elections using list-systems with proportional representation, albeit to different extents, namely the More-Preferred-Less-Seats Paradox, while two other paradoxes, the Condorcet-Winner-Turns-Loser Paradox and the Condorcet-Loser-Turns-Winner Paradox, seem to occur occasionally.

What does the existence of these paradoxes mean, I hear you ask. Well:

More generally the presence of this particular paradox would seem to contradict a premise underlying much of contemporary democratic theory, namely that if more people prefer one party than another, then it would be wrong for the latter party to receive more seats. It would in particular seem contrary to the views and justifications usually offered in favour of proportional representation, namely that this somehow is more in line with "what the people want". Proportional representation obviously tends to add pluralism to a party system compared to, e.g., the first-past-the-post system, but it also seem not only to do so by benefiting some parties at the cost of others (which is inevitable) but also to do so in a way that seems intuitively in direct opposition to the majority principle underlying the democratic idea.

[...]

...apparently, it is quite often the case that the preference of a majority of the voters is quite different from what the seat allocations are under proportional representation. This is an insight which should make democrats careful about how they emphasize the values of the democratic process and its outcomes - and be more interested in the consequenses of alternative electoral arrangements.

...

I have only looked at a few of the papers from the conference, but it looks like there's a lot of interesting stuff there. Do take a look if you have the time.

...

På det lidt mere seriøse plan, hvis vi ønsker at forklare journalisternes fremgangsmåde og problemerne dermed, så er det værd at bemærke at, som Arnold Kling så rigtigt udtrykker det:

Journalists write stories.

Sværere er det ikke at forklare meget af det vrøvl som bliver lukket ud, når en ny meningsmåling har set dagens lys. Det er ikke en historie, at ingen partier har flyttet sig en milimeter siden valget, eller at for eksempel en hed politisk debat om dette eller hint ingen konsekvenser har haft for partiernes tilslutning. Eller måske er det, men ikke en historie som mange journalister vil finde det interessant at skrive om - det er sjældent at resultater à la "we fail to reject the 0-hypothesis" fører til avisoverskrifter. Så journalisterne, med deres begrænsede viden om statistik som naturlig følgesvend, kigger med lys og lygte efter overskrifter i tallene, og der er altid et par stykker eller tre. Politiske eksperter vil herefter gøre sig selv til grin og underminere enhver form for "ekspertstatus" de måtte have, ved at fodre journalisten med mere eller mindre plausible politiske forklaringer på de teoretiske vælgerbevægelser, som ikke findes ret mange andre steder end i journalisternes hoveder. Eller de "politiske eksperter" vil måske på egen hånd fylde journalisterne med forvrøvlede ideer om, "hvad tallene viser".

Hvordan løser "man" så "problemet"? Rigtig gættet: Der er ikke noget "problem", som kræver en "løsning".

Folk får den journalistik de betaler for.