Fig. 2

Reported impact of information on consultations/antibiotic requests disaggregated by whether or not antibiotics would ‘help’ ILI. (i) Prior Survey results are for May–June 2015 data reported in Roope et al. [24]; wave 1 conducted in October–November 2016; wave 2 conducted in March 2017. (ii) V1 is ‘fear-only’ message; V2 is ‘mild-fear-plus-empowerment’; V3 is ‘strong-fear-plus-empowerment’. (iii) Column charts exclude respondents who answered ‘do not know’ (< 7%). (iv) p values are from Wilcoxon’s single-sample sign-rank tests of whether respondents were more/less likely to consult/request antibiotics for ILI in response to the information they were given, under the null hypothesis that they were neither more likely nor less likely to consult/request antibiotics