 data analysis. What most people don't know, 80% of that is actually data cleaning and whipping the data into a shape so we can analyze it. There's often that image that we have huge servers here and we we write complicated code and derive formulas on the whiteboard like the movies. And there are some of that, but actually 80% is very tedious cleaning data, merging different data sources together, confirming the data quality to make sure that what later enters the model and our statistical models is sound and makes sense and it's of high quality. It always helps in that phase to have this connection with the field, to have been there, to have seen the reality. So when data speaks, when there are mistakes or inconsistencies, you can actually identify them and can tell this number is odd or I'm surprised by this, my intuition was not telling me that. And most of the time the complementarity between field work and data analysis make the story and the survey great. A fair bit of statistical modeling is to make sure that the final estimates that we present to policymakers. Is there a 10% increase or 10% decrease or you know, does this campaign work or doesn't it work? To make sure that this estimate is very reliable and to inform projects and their policies. The last phase is the writing phase and the dissemination phase. My job there is to make these numbers something that tells a story.