The pollen of the maize (corn) plant is a source of food to larval mosquitoes of the species Anopheles arabiensis, the main vector of malaria in Ethiopia. The production of maize has increased substantially in certain areas of Ethiopia recently, and over the same time period, malaria has entered in to new areas where it was previously rare. This raises the question, is the increase of maize cultivation partly responsible for the increase in malaria?

One line of evidence is to look for an association between maize production and malaria incidence at different geographically dispersed sites (Kebede et al.2005). The data set "malaria vs maize.csv" contains information on several high-altitude sites in Ethiopia, with information about the level of cultivation of maize (low, medium or high in the variable maize_yield) and the rate of malaria per 10,000 people (incidence_rate_per_ten_thousand).

Plot a multiple histogram to show the relationship between level of maize production and the incidence of malaria.

ANOVA is a logical choice of method to test differences in the mean rate of malaria between sites differing in level of maize production. Calculate the standard deviation of the incidence rate for each level of maize yield. Do these data seem to conform to the assumptions of ANOVA? Describe any violations of assumptions you identify.

Compute the log of the incidence rate and redraw the multiple histograms for different levels of maize yield. Calculate the standard deviation of the log incidence rate for each level of maize yield. Does the log-transformed data better meet the assumptions of ANOVA than did the untransformed data?

Test for an association between maize yield and malaria incidence. Should I use Tukey-Kramer test or the Kruskal-Wallis tes to test for the association between maize yield and malaria incidence?