Who is the strictest referee in the Premier League?
The question has a real answer. It is not a name.
The referees currently on the list
Yellow cards per match across each referee's whole career, with the range the figure could plausibly take. Ordered by rate because overlapping ranges cannot be ordered any other way — not because the order means anything.
Dot: the career rate. Bar: the 95% range. Dashed line: the league average (3.28 yellow cards a match). Where two bars overlap, the two referees cannot be told apart.
Why the top of the list is not the answer
| Pairs of referees | Separate | Of |
|---|---|---|
| All pairs | 245 | 595 |
| Adjacent pairs — each referee against the man immediately behind him | 0 | 34 |
| Fouls, all pairs | 307 | 595 |
| Fouls, adjacent pairs | 0 | 34 |
Plenty of referees genuinely differ; none differs from his own neighbour. D. England leads at 3.95 a match, but his range (3.58–4.33) contains every referee for some distance below him. "Stricter than average?" the data can settle. "Strictest?" it cannot.
It is a real trait — over a career
| Split-half reliability | Yellow cards | Fouls |
|---|---|---|
| Full career | 0.86 | 0.95 |
| Last three seasons only | 0.48 | — |
Split a referee's matches into two random halves and compare them. Over a career they agree; over three seasons, barely better than a coin — which is why every referee figure here runs across all 26 seasons. Pinning one rate to within a quarter of a card takes 219 matches, about 6 seasons.
Every referee with 100 or more Premier League matches
| Referee | Matches | Yellows / match | 95% range | Against the league average |
|---|---|---|---|---|
| D. England | 104 | 3.95 | 3.58 – 4.33 | above the norm |
| P. Bankes | 118 | 3.93 | 3.58 – 4.29 | above the norm |
| D. Cooteformer | 112 | 3.88 | 3.48 – 4.27 | above the norm |
| R. Jones | 113 | 3.85 | 3.50 – 4.20 | above the norm |
| S. Attwell | 247 | 3.74 | 3.49 – 3.99 | above the norm |
| Mike Rileyformer | 205 | 3.74 | 3.46 – 4.02 | above the norm |
| Mike Deanformer | 561 | 3.65 | 3.48 – 3.81 | above the norm |
| S. Hooper | 141 | 3.65 | 3.31 – 3.98 | above the norm |
| C. Kavanagh | 197 | 3.61 | 3.31 – 3.91 | above the norm |
| C. Pawson | 291 | 3.58 | 3.35 – 3.82 | above the norm |
| Phil Dowdformer | 306 | 3.55 | 3.33 – 3.77 | above the norm |
| A. Taylorformer | 433 | 3.51 | 3.31 – 3.71 | above the norm |
| P. Tierney | 197 | 3.40 | 3.13 – 3.66 | — |
| Steve Bennettformer | 243 | 3.40 | 3.19 – 3.60 | — |
| Kevin Friendformer | 272 | 3.38 | 3.15 – 3.62 | — |
| Neil Swarbrickformer | 132 | 3.33 | 3.04 – 3.62 | — |
| Rob Stylesformer | 211 | 3.33 | 3.05 – 3.60 | — |
| Howard Webbformer | 297 | 3.31 | 3.09 – 3.52 | — |
| Mark Clattenburgformer | 293 | 3.25 | 3.02 – 3.48 | — |
| Martin Atkinsonformer | 462 | 3.24 | 3.08 – 3.40 | — |
| A. Madley | 136 | 3.23 | 2.86 – 3.61 | — |
| Jonathan Mossformer | 274 | 3.23 | 3.01 – 3.45 | — |
| M. Oliver | 434 | 3.22 | 3.05 – 3.39 | — |
| Mike Jonesformer | 202 | 3.17 | 2.92 – 3.42 | — |
| Andre Marrinerformer | 392 | 3.13 | 2.95 – 3.30 | — |
| Lee Masonformer | 288 | 3.08 | 2.87 – 3.29 | — |
| Graham Pollformer | 186 | 3.06 | 2.78 – 3.35 | — |
| Alan Wileyformer | 253 | 3.04 | 2.82 – 3.25 | below the norm |
| Peter Waltonformer | 169 | 2.85 | 2.60 – 3.09 | below the norm |
| Chris Foyformer | 262 | 2.73 | 2.52 – 2.95 | below the norm |
| G. Scottformer | 105 | 2.68 | 2.36 – 2.99 | below the norm |
| Uriah Rennieformer | 116 | 2.65 | 2.32 – 2.97 | below the norm |
| Dermot Gallagherformer | 125 | 2.55 | 2.25 – 2.85 | below the norm |
| Lee Probertformer | 182 | 2.52 | 2.25 – 2.78 | below the norm |
| Mark Halseyformer | 265 | 2.17 | 2.00 – 2.35 | below the norm |
9,707 of 10,260 matches since 2000-01; the rest lack a complete card record or an identified referee. Public convention, Premier League only — a Championship rate belongs to a differently refereed competition. Under 100 matches the range is too wide to print.
What we will say instead
- Counts. How many yellow cards a referee showed, how many penalties he gave, how many errors the panel confirmed. Nobody estimated those, so there is nothing to separate.
- Distance from the norm, where it clears the margin of error — 12 referees above, 8 below, 15 indistinguishable from average.
- Never a ranking of one referee against the next. How we decide what is publishable.