Support ESC Insight on Patreon

A Deep Analysis of Melodifestivalen’s Rehearsal Poll Written by on February 17, 2023

On each Friday night of Melfest season audience members at the rehearsal are asked “Who had your favourite performance of the night?”

Their answers make up the publikundersökningen, the poll of audience members that is often seen as a prediction of the results in the final. But is it a good predictor? Ben Robertson reports from his experience in Linköping and Lidköping.

On the last two Friday nights I have been helping out the Swedish branch of OGAE Sweden, Melodifestivalklubben, with one of their traditions each and every Melfest week. The Swedish fan club is usually provided press passes each week to cover the show and rehearsals like others do, but they also fulfil an important role to the storyline of each week of Melfest. They organise the publikundersökningen, known colloquially in English as the poll, that takes place at this Friday rehearsal.

This Friday evening rehearsal is hotly anticipated by those of us following the contest closely, especially during the heats. That is because these songs have been a trade secret up until that point, with only the bubble of journalists, circa 40 or 50 people, able to talk freely about what the songs sound like and their favourites.

The audience members get one vote and one vote only for the OGAE Sweden poll (Photo: Alma Bengtsson, SVT)

However the Friday evening rehearsal may have fake voting but generally succeeds from Lidköping to Örnsköldsvik in attracting of a few thousand members of the public happy to pay a hefty sum to watch a practice (the adult prices for Friday night’s rehearsal in Lidköping started at 540 kr, around £43.

The poll works in a quite simple way. A few faces from OGAE stand around the arena as members of the public leave the arena and are asked a simple question. “Who had your favourite performance in the show?” Respondents are able to give one name and one name only that is added to a tally chart, before the OGAE members reconvene about thirty minutes after the show to share their results with each other and share them to the world.

And so there I was last Friday night, clipboard ready, surveying each and every person I had a chance to speak to who their favourite was of the night.

Our results forecast a landslide, but our results were very wrong.

More Of You

In total between the three of us we spoke to 671 people that night and 324 of them, almost half, had Theoz and ‘Mer Av Dig’ as their favourite performance of the night. This level of support, and more than double the 21% of the vote that Panetoz received in second place, meant we felt we were in for a landslide. We were witnessing entire families striding out of the arena and in unison declaring that Theoz was their numero uno.

If all you witnessed was the mood outside the SAAB Arena that night then you have assumed Saturday was to be a coronation for the 17-year-old star. But so it didn’t turn out to be.

Theoz ended up in 3rd place during the heat, which was won by Maria Sur (3rd place in the poll) with Panetoz taking the second automatic qualifying spot. This leads us to ask what went wrong this week, when the week before the poll correctly identified that Jon Henrik Fjällgren and Tone Sekelius would stroll into the Friends Arena showcase with ease.

One huge problem that we had that night was our location in Linköping. If your Swedish comprehension was decent on Saturday you would have noticed that Theoz’s postcard on Saturday night was filmed from the main square of Linköping, as it so happens Theoz was from the same town as this second heat was held. When I glanced into the stands on Friday night every placard visible was for the young star, and while the audience may have only heard one minute of his song before the show started they very clearly came with a favourite set in stone.

Another factor at play is the demographics of the audience that attend such rehearsals. Firstly the types of people who attend a Melodifestivalen show live are going to be those who tune in religiously and are fans of the entire show’s glitz and glamour. Theoz’s song played into their hands, with a spectacular dance routine, obligatory key change in the right place and a humongous pyro budget that blew the other competitors away. Of all the acts that performed on Saturday this was the one that screamed modern day Melfest and appealed to this target audience.

Those demographics though also had one further problem that they don’t represent the full spectra of ages that vote in Melodifestivalen equally. As viewers will note, votes through the show’s app are divided into seven different age categories, starting off with 3 to 9 year olds and ending with the over 75s. But the rehearsals I have been to have been dominated by families with small children, and devoid of many fans from 16 to 29 for example. This skews the data collected towards the younger fan taste, and it’s notable here that Wiktoria was a clear 4th place in Friday night’s poll, but it was Tennessee Tears, with 12 points from the two older categories and the televote, that came through to comfortably claim 4th place in the final ranking.

So with all of these biases that exist in producing the poll, I decided to investigate how reliable this prediction tool was for the Melodifestivalen heats.

The Data Analysis

I decided to look at the data from the poll in the years going from 2020 back to 2016, to account for the years where the app was the main voting method (due to the way the app was used in 2015, where many people were unable to vote during the show and running order was very important as one voted during each performance, I am not using 2015 for this analysis). I am also looking in this investigation at the polls taken during the heats, and not the one’s during the show’s Semi Final/Andra Chansen rounds or the final, as it is that first impression of the songs that is most interesting to get an opinion for.

The statistics look strong for the poll. In those 20 shows the poll has predicted the qualifiers with a 72.5% success rate, and has always managed to correctly predict at least one of the two artists who have qualified directly to the Friends Arena finale. If we look even lower down the leaderboard at the top 4, those artists who remain in the competition after the heat one way or another, then that prediction success rate rises to 85%. On 10 of those twenty occasions did the poll correctly predict which songs would be in the top four, and 5 times out of twenty the poll accurated called the top 2 and who would end up in the 3rd/4th places.

Peter Baston, the Melodifestivalklubben member responsible for co-ordinating the poll this year, believes that the poll is as good a predictor as anybody else is, claiming that the poll is “no better or worse than reporters from Aftonbladet or from the [OGAE] club.” This data is backed up by results from the Swedish branch of OGAE, which plays a prediction game each week to try and guess the correct outcome. Even with hundreds of members playing the game the poll has managed to place inside the top 10 within the fan club some years, with the poll and the reporters’ tips being practically equally as successful long-term.

Some of these correct calls stand out for surprising the expectations of the media before the show. It was the poll that correctly predicted that ‘Statements’ would miss out to Wiktoria and Jon Henrik Fjällgren in 2017, and the previous year it was a huge landslide for Frans’ ‘If I Were Sorry’ after the rehearsal show in 2016 where he took more than half of the votes that flipped expectations overnight to made him a contender for victory.

However those expectations are coupled with big misses, and arguably bigger misses than Theoz last week. The example in modern times where the poll did worst was heat 4 in 2018, where the poll predicted only two of the eventual top 4, with one of the finalists correctly predicted and none of the Semi Final/Andra Chansen entries

Results of the Melodifestivalklubben poll from heat 4 in 2018 (Image: Melodifestivalklubben)

This poll shows the weakness that the poll has most clearly. While Mariette wins, and she went through to win this heat, she is followed by Elias Abbas in second place who eventually finished fifth and knocked out on the night. Elias’ song ‘Mitt Paradis’ is the type of Melfest entry made for the live setting and for families to clap and jump around to and leave the arena telling people they enjoyed it.

On the flip side we can see that Felix Sandman, the artist who eventually finished in second place in the entire Melodifestivalen circus that year, ended up placing 6th out of 7 in the poll with just 8.6 % of the vote. Felix’s poll failure is a great example that the quieter more gentle Melodifestivalen numbers are likely to fly by the viewers coming to the show live, who are thrust upon entry with sweets, crisps and chocolate from sponsors and geared up for a Friday night family party. Gentle ballads will without much exception pass the poll by, while the uptempo crowd pleasers tend to overperform.

And we see this bias for the child friendly catchy hit to overscore in the poll above permeate other poll failures. Theoz is the obvious example from Linköping, but in 2019 there was shock around the community when Dolly Style popped a win on the Friday night result, eventually coming 5th with ‘Habibi’, swapping placings with Lina Hedlund’s ‘Victorious’ that scored just 7% of the votes that night compared to Dolly Style’s 26.5%. Other similar twists include Anis Don Demina (29%) losing out to Mariette (14%) in 2020 and SaRaHa (39%) losing out to Lisa Ajax (12%), where the crowd-pleasing family favourite was pipped at the line to something more mature.

Peter argues that this is a demographic problem. “Parents are just taking their kids out”, he explains, “and parents often let the children decide and follow their taste.” With the advent of the age range voting we have seen how wide the differences can be between younger and older demographics, but the poll skews younger. Interesting in 2019 and 2020, years where the age based app vote was used at Melodifestivalen, the poll is in all practical senses no worse a predictor tool than previously (69% predictor rate for the final, 84% for the top 4).

And To Tonight’s Poll

Tonight’s poll in Lidköping (no, not Linköping, and yes, there is a joke about that in tonight’s show) had three of us from Melodifestivalklubben ask over 600 people from the audience who their favourite was. In little surprise Marcus and Martinus’s song ‘Air‘ was ranked top by nearly half of the audience members.

Such is the gap to those in contention I have full faith that this poll will correctly call them as the winner of this heat. Yes the pair may have the same child-friendly appeal to Theoz, but I’d also argue that Paul Rey in second place also appeals to that family friendly audience as the two Norwegians. There is no Maria Sur in this lineup with the narrative nor stand out vocal that I believe will come up the rails.

However the battle for those other places appears tight. I’d argue that Melanie’s song perhaps is the dark horse here, quieter and gentler than Nordman’s that did get the crowd roused by the final chorus into that oh-so-Swedish clap-a-long. Casanova’s may also be a name to watch out for, their demographc was split with Nordman tonight and with app voting at home and an older audience I would anticipate a higher result.

How accurate will this week’s poll be? Let’s find out!

About The Author: Ben Robertson

Ben Robertson has attended 23 National Finals in the world of Eurovision. With that experience behind him he writes for ESC Insight with his analysis and opinions about anything and everything Eurovision Song Contest that is worth telling.

Read more from this author...

You Can Support ESC Insight on Patreon

ESC Insight's Patreon page is now live; click here to see what it's all about, and how you can get involved and directly support our coverage of your Eurovision Song Contest.

If You Like This...

Have Your Say

Leave a Reply