The study referred to in the Warwickshire survey is "The Relationship between Part-time Employment and A-level Results" published in the Winter 1992 edition of Educational Research by Peter Tymms and Carol Fitz-Gibbon who used the ALIS database to make the comparisons. At first glance, the story seems straightforward - you go out to work, you get lower grades; you work long hours, you get lower grades still. Or as the study puts it. "It was generally the case that those who reported working part-time gained lower grades than those who did not. Furthermore, except for chemistry, those who reported working more than nine hours gained lower grades than those who worked less than nine hours."
The real story, though, is in the detail. First, the, observed effect of working is very small - on average, less than a quarter of a grade in each subject for under nine hours' work; less than half a grade for more than nine hours.
Then, the link between work and grades is by no means necessarily one of cause and effect (the study deliberately speaks only of an "association" between the two). The authors point out that the differences "...may be explicable in terms of a self-selection mechanism whereby students who are already struggling with A-level work tend to seek part-time work." This reading is reinforced by the fact that general studies (for which there is often no formal course, and no homework) seems to be affected by part-time work neither more nor less than are other subjects.
The common sense interpretation here, as the study says, is that "...many students are able to maintain a part-time job without adversely affecting their grades, especially if the hours spent working are modest."
Nevertheless, the same common sense also says, as Peter Tymms pointed out to me: "If you are a parent or an A-level tutor then this is obviously something you worry about."