Midnight Voices poll of Pete Atkin Favourites

Midnight Voices poll of Pete Atkin "Top Ten" songs, spring 1998

Results analysed by Richard Corfield

The way that the voting was organised reflected three measures of the popularity of a song:

The score that it was given (between 1 and 10) which can be expressed as an average i.e. the total score divided by the number of people who voted for it.
The spread of the individual votes around that mean (technically the standard error).
And of course (as pointed out by Michael Cross) the number of people who voted for it in the first place.

So, the measure that I've used reflects all of these: (x/se)*n

This index not only reflects how highly a song scored, it also reflects how consistent were the scores for it. In other words it penalises averages calculated from a large spread of original values and emphasises averages where the voters agree very well on the score. An example of this would be 'Tonight your love is over'. This song ranks 5 out of 58 because the people who voted for it all agreed that it should score highly (10+9+8+9+9+8). On the other hand The Hypertension kid is penalised (and performs less well in the final ranking) because of the very large spread of scores it was given (5+9+7+10+8+2+10+9+3).

Finally weight is given to the most important measure of all: How many people voted for it.


Ranking  Song          Sum  mean      se         n   mean/se  mean/se*n

      1  GIRL ON       154  7.000000  0.538557   22  12.9977  285.950
      2  THIRTY        137  6.523810  0.604030   21  10.8005  226.810
      3  FADED         146  7.684211  0.671268   19  11.4473  217.499
      4  THIEF         128  6.736842  0.615787   19  10.9402  207.864
      5  TONIGHT        53  8.833333  0.307318    6  28.7433  172.460
      6  BOTBS         121  6.368421  0.705909   19   9.0216  171.410
      7  PERFECT       110  6.111111  0.666122   18   9.1742  165.135
      8  WRIST          56  7.000000  0.377964    8  18.5203  148.162
      9  PAYDAY         87  6.692308  0.663949   13  10.0796  131.034
     10  FLOWERS        76  4.470588  0.582323   17   7.6772  130.512
     11  HIST AND       88  7.333333  0.828775   12   8.8484  106.181
     12  CANOE          82  5.466667  0.844403   15   6.4740   97.110
     13  CARNATIO       53  3.785714  0.566254   14   6.6855   93.598
     14  SUN GATE       63  4.846154  0.749753   13   6.4637   84.028
     15  A KING AT      55  6.111111  0.654990    9   9.3301   83.971
     16  SENIOR C       63  5.727273  0.751857   11   7.6175   83.793
     17  SESSION        54  4.909091  0.802887   11   6.1143   67.257
     18  THE HYPER      63  7.000000  1.000000    9   7.0000   63.000
     19  BETWEEN        60  5.000000  0.953463   12   5.2440   62.929
     20  PRINCE         37  5.285714  0.865043    7   6.1103   42.772
     21  ARRAY          30  6.000000  0.707107    5   8.4853   42.426
     22  TOUCH          35  5.000000  0.845154    7   5.9161   41.413
     23  ISTJOKER       32  4.571429  0.895947    7   5.1023   35.716
     24  SECRET D       36  5.142857  1.100402    7   4.6736   32.715
     25  LAST HILL      34  5.666667  1.054093    6   5.3759   32.255
     26  NAT STEL       26  5.200000  0.860233    5   6.0449   30.224
     27  TONGUE         35  5.000000  1.175139    7   4.2548   29.784
     28  MASTER         28  5.600000  0.979796    5   5.7155   28.577
     29  LAUGH BO       18  3.000000  0.632456    6   4.7434   28.461
     30  SUNRISE        30  5.000000  1.154701    6   4.3301   25.981
     31  NO DICE        28  4.666667  1.085255    6   4.3001   25.800
     32  SHADOW         30  5.000000  1.183216    6   4.2258   25.355
     33  BALLAD         36  7.200000  1.529706    5   4.7068   23.534
     34  HOLLOW         27  4.500000  1.258306    6   3.5762   21.457
     35  SCREEN         26  4.333333  1.256096    6   3.4498   20.699
     36  PEARLDRIL      28  7.000000  1.354006    4   5.1698   20.679
     37  S AND D        26  4.333333  1.282359    6   3.3792   20.275
     38  LADY DAY       28  5.600000  1.400000    5   4.0000   20.000
     39  ERRANT         23  3.833333  1.222475    6   3.1357   18.814
     40  DOUBLE         31  7.750000  1.652019    4   4.6912   18.765
     41  CAN'T EX       26  5.200000  1.462874    5   3.5546   17.773
     42  RAIN           19  3.166667  1.077549    6   2.9388   17.633
     43  ALL I EVE       9  2.250000  0.629153    4   3.5762   14.305
     44  RIDER          13  2.600000  0.927362    5   2.8037   14.018
     45  PRAC MAN       27  6.750000  2.136001    4   3.1601   12.640
     46  APPARIT        12  6.000000  1.000000    2   6.0000   12.000
     47  TENDER         19  6.333333  1.666667    3   3.8000   11.400
     48  ORIGINAL       18  4.500000  1.755942    4   2.5627   10.251
     49  CARE CHA       18  6.000000  2.081666    3   2.8823    8.647
     50  THE BEAU       14  4.666667  1.855921    3   2.5145    7.543
     51  HAVE BIRO      10  3.333333  1.855921    3   1.7961    5.388
     52  WHERE?         10  5.000000  2.000000    2   2.5000    5.000
     53  ROAD OF        11  5.500000  2.500000    2   2.2000    4.400
     54  MAN WHO        10  2.500000  0.645497    1   3.8730    3.873
     55  BLACK F         5  2.500000  1.500000    2   1.6667    3.333
     56  EYE             5  2.500000  1.500000    2   1.6667    3.333
     57  WALL OF         6  3.000000  2.000000    2   1.5000    3.000
     58  DTMA           10  5.000000  4.000000    2   1.2500    2.500

Several low scoring songs couldn't be ranked. These are the songs with 1 or 0 voters.

	LUCK OF THE DRAW
	MY EGOIST
	FRANGIPANNI WAS HER FLOWER
	WHAT AM I SUPPOSED TO DO?
	A MAN WHO'S BEEN AROUND
	ALL THE DEAD WERE STRANGERS
	BE CAREFUL WHEN THEY OFFER YOU THE MOON
	FRIENDLY ISLAND SONG
	TIME AND TIME AGAIN
	NOTHING LEFT TO SAY
	STRANGER IN TOWN
	OUR LADY LOWNESS
	LITTLE SAMMY SPEEDBALL

A word of warning: You've heard Mark Twain's assessment of statistics, 'Lies, damned lies, etc'? Well that needs to be borne in mind. All this does is rank songs by their average popularity. I'm sure that most of us would produce a completely different top 58 (or top 71 - I can see at least three songs in the unranked section that I like very much personally).

However, when it comes to thinking about what Pete and Clive should use when invited to contribute on the radio or TV etc, I think that this list - together with Ian's and Michael's - could be helpful. The top ten are statistically very defensible - probably the top twenty actually.

So, that's it. Steve is planning to post the results and the associated graph on the web. Please bear in mind when you look at the results that where n (the number of voters) is less than three, statistically we're on thin ice.

I'd like to thank Dr Cortina Borja (University of Oxford, Department of Statistics) for his advice on the statistical approach.

back to Pete Atkin Home Page