I read the article and have a few questions to the authors: How was the megaupload estimate calculated? Various parts of the article use per week data to meas

I read the article and have a few questions to the authors: How was the megaupload estimate calculated? Various parts of the article use per week data to measure sales / rental trends, however the megaupload estimate for each country was held constant, given there's a likely correlation (with a potential time lag - i.e people buy stuff on the advice of people that have already seen it and recommended), how would this variation effect the modelling? What sort of measurement error does the google adwords estimate have? How does this effect the model? Are there any factors that might impact the adwords estimate on a per country basis (such as prevelance of use of particular browsers, adblocking plugins, etc.)? How does this effect the model? Do you have year on year data to compare this data with? Do you have any data to show correlation between sales and MPR over time? Does your data differentiate on movies that were available on megaupload before it closed vs those that were released afterwards? How does this effect your model?