It occurred to me that teams may play differently under floodlights. Maybe shoot more on the off chance that the ‘keepers visibility wasn’t as good as in natural light?
Now my data doesn’t have kick off times so I’ve split it two ways. Firstly I’ve split the sample into midweek and weekend games, on the assumption that midweek games are played under floodlights whereas weekend ones are not. This is shown in the table below, where ST = shots on target, and TS = total shots, and Sh% = shooting percentage:
However there’s also a lot of weekend games played under the lights around the equinox, so below I’ve switched up the samples such that the midweek sample also incorporates games played on a weekend in November, December, and January:
So what can we surmise? Well the two tables mirror each other pretty well; weekend games have more goals as a result of proportionately more shots, however over a 38 matches the difference is about 1 goal, 9 shots on target, and 12 total shots between the two sides. There’s a tiny effect here, and in the opposite direction to that I’d expected, but it’s barely distinguishable from the noise.