This is a question I pondered last season as I watched Eastleigh at their Silverlake ground. My memory told me that a couple of seasons before at the place I was standing – along the far side of the pitch opposite the main stand, the crowd had been two, even three deep. But that day there stood only a thin line of people forlornly strung out like a remote island chain. The queue at the tea hut also seemed shorter. Where had everyone gone?
I decided to take a look at the figures which confirmed that Eastleighs average attendances had indeed peaked at 723 during the 2008-2009 season. That season had in many ways been a high-water mark for the club and the average attendance figure had subsequently slipped back – to 543 for the 2009/10 campaign, and 537 for 2010/11
In contrast 2008/09 had for near-neighbours (a mere 4 miles away door-to-door) Southampton FC proven to be the darkest moment in their history; relegated from the Championship and heading into financial oblivion the club had spent much of that season in turmoil and unsurprisingly attendances slumped – an average of 17 849 turned up to St Mary’s that season – a drop of some 6001 from only two seasons before. Since the Saints began their road to recovery however, attendances recovered with 09/10s average coming in at 20 982 and 10/11s bouncing back to a more healthy 22 160.
Could it have been that attendances at the two clubs were in some way linked? did some of the thousands staying away from St Marys opt for non-league football instead – boosting Eastleigh’s attendances to record levels? (only a small number would need to do so – 2.5% of the lost 6001 would make 150 extra spectators at Eastleigh’s Silverlake ground – enough to be have a visible effect), but is this effect (if it exists) now in reverse – are crowds being drawn away from Eastleigh by the magnetic pull exerted by the bigger club just like the tides responding to the moon?
Time to get statistical….
Using Excel we can calculate the correlation coefficient between average attendances at the two clubs. This measures the strength of a relationship between two sets of variables (in this case average attendances) giving us a correlation coefficient. This will be on a scale from -1 to +1. A strong positive correlation will be around +1 whilst a strong negative correlation will be nearer -1. A figure of 0 indicates no correlation between the two variables. If the coefficient is squared we can also get a measure of the percentage of change in one variable that is explained by change in the other.
First we need to get hold of some data. The biggest issue with this type of research is ensuring it is comparable i.e that there haven’t been different methods used in compiling the raw data. To ensure comparability only league attendances will be counted.The data for Southampton are sourced from ESPN Soccernet. For Eastleigh it appears the only data available online is from Wikipedia. I have however managed to check the 2008/9 figure with a separate source and the two mach up so it would seem on the face of it to be reliable.
Test 1 Southampton vs Eastleigh:
The analysis includes the years 2005/6 to 2011/12 this time period covers the period Eastleigh have been members of the Conference South. Earlier they had been playing in the Wessex League which attracts only modest crowds so this would unbalance any comparison. Southampton meanwhile have played in the Premier League, Championship, and League 1 so this period allows us to see the impact of Saints crowds on Eastleigh’s attendance figures.
The correlation coefficient from this test is -0.46 which suggests there is a moderate correlation between the two attendance figures. If this is squared we get a figure of 0.21 which means that the variation in one attendance figure explains 21% of the variation in the other. The negative relationship also indicates that as one figure goes up, the other goes down. Whilst the test does not tell us about causality, or the direction of causality it seems safe to assume that the bigger club, in this case Southampton, would have a bigger gravitational pull and that it is attendance figures at St Marys which would determine attendances at Eastleigh rather than the reverse.
But, before drawing our conclusions there are some other questions….
Test 2: Portsmouth Vs Havant & Waterlooville
One question is can this relationship be observed with other clubs, or is it particular to Eastleigh and Southampton? Just a short trip down the coast Portsmouth and Havant & Waterlooville provide another case to investigate. Portsmouth have recently had a successful period followed by a number of well documented financial problems and have suffered relegation from the Premier league and the Championship.
The figures for Havant & Waterlooville is obtained from this source. For Portsmouth data again comes from ESPN Soccernet, though as this only goes back to the 01/02 season data from 98/99 – 00/01 has been obtained from european-football-statistics.co.uk – again both are league attendances
As Havant & Waterlooville have been competing at the same level for longer it is possible to use data from further back – right to the first season as a merged club in 1998/99.
The correlation coefficient here is more towards the weak end of the spectrum at R = 0.33 meaning that only 11% of the difference in one attendance figure is attributable to difference in the other. This is relatively low and suggests that there is little connection between attendances at the two clubs.
The relationship also differs with Southampton and Eastleigh in being positive, meaning higher attendances at one club are associated with higher attendances at the other. There are several possible explanations for this one possible explanation for this is coincidence. Another is that attendances at both clubs are equally affected by an unknown variable or variables. A third explanation is that there is a dynamic between the clubs which benefits Havant & Waterlooville; Portsmouth reserve matches being played at Westleigh Park and discounts for Portsmouth season ticket holders may mean mean that a number of Portsmouth fans may also visit Westleigh to see Havant and Waterlooville play. The higher the attendances at Fratton Park the bigger this effect.
Eastleigh V Saints revisited;
The original hypothesis is based on the observation that in terms of average attendances Southampton’s recent low point and Eastleigh’s high point coincided but is there another explanation for this?
I decided to dig out an old programme for the last game of the 2008/9 season, the play-off semi final between Eastleigh and Hayes and Yeading. Inside was a list of attendance figures for the whole of the season. One stood out; Eastleigh Vs AFC Wimbledon drew a record breaking crowd of 2283.
Many of the crowd were, as this match report indicates, supporters of AFC Wimbledon. This is a club with a unique history which explains its high level of support relative to the Conference South. AFC Wimbledon would only spend one season in this league with the effect of distorting average attendances for this season. The impact of this one game on the Eastleigh’s average is such that drops considerably from 723 to 645 if we disregard the Wimbledon game.
Using the revised figure to re-run the test we get this:
The result is slightly different. The correlation remains negative, but removing the AFC Wimbledon factor R now drops to -0.30, again in the weak part of the range. If we square R we get 0.09, or 9% of the variation in one variable being explained by variation in the other variable – this is relatively low and suggests there are other far more potent factors which explain the variation in the attendances at Eastleigh.
Taking into account the revised figure our conclusion therefore is that whilst we cannot wholly rule out some relationship between attendances at Eastleigh and Southampton the evidence appears to be that any connection between the two is fairly weak.
The data from the Havant and Waterlooville and Portsmouth average attendance figures also suggest any relationship is weak and go against the view that if attendance increased at a league club then attendances at nearby smaller clubs will necessarily fall. In fact one possibility hinted at by the data is the suggestion that higher attendances at Portsmouth may even have been beneficial to Havant & Waterlooville.
Combined we can say that there is little evidence that league clubs draw spectators away from neighbouring non-league clubs.
It would seem, based on this limited research, that there is no significant relationship between attendance at league clubs and attendance at non-league clubs. In this sense they inhabit two very different spheres. Spectators abandoning a league club do not substitute the experience of attending a league clubs match for non league football, just as non-league supporters do not appear to be among the ranks of those attracted by greater success to league clubs.