Lacking as they do any statistical analysis of variance, most of your tabular surveys are pretty meaningless, but in the absence of figures for age distribution the survey on pay is especially so.
Salary is closely correlated to age, so most of the differences between institutions (and genders) will simply reflect that. When the range of means for all but one institution is approximately ?8,000, and when one salary at Bath University changes the gender gap based on "averages" by about Pounds 14,000, it's time to consider statistical significances in a grown-up way. Also, the median might be a more useful metric for averages than the mean.
Douglas B. Kell Director, Centre for Integrative Systems Biology Manchester University