1. It has been a superhectic week. Had to present the research at the Second Annual ACE Graduate Student Research Celebration (my god, why do they use such a long, terrible name?). The presentation was fine, but not as good as I had expected. (And hey, no evaluation form this year?). It was just a report of previous results plus some motivation for the next stages. I was so hoping to include some welfare calculation based on a simulated estimation, to compare with the classical, non-simulated previous result. But, what could I do, the program did not converge until 10 hours before the presentation! In fact it has not converged till now, for crying outloud.
2. DSB raised an interesting point in my presentation. I mentioned a study by Lichtkoppler and Blaine for Ashtabula County as a comparison to my estimation. Their welfare figure, after being scaled up to community level-- is well below ours. I was alluding that it might be due to the form of their survey: they used a tax-referendum contingent valuation, asking directly how much tax increase respondents are willing to bear in order to enjoy a better particular condition. I am afraid that kind of survey invites protest responses, for I have found such sentiment all over the comment/suggestion box in our returned surveys. But DBS implicitly contended that WTP as espressed through referendum might be the unbiased one. Well, I think I would disagree. As JBB said afterward, the central issue here is who should pay for the cleanup of legacy pollution. Confronting respondents with referenda would definitely induce protest bids -- and thus downward biased welfare measures. However, I agree that our numbers might also be overestimated since embedding effect might have played some role. Need to check on this further.
3. MK and AA questioned my stepwise elimination of insignificant parameters in order to get the most parsimonious specification for welfare measurement. Their main concern was the possibility of bias from variables omission. Admittedly, I had overlooked this issue. The reason I eliminated "unimpotant" variables one-at-a-time was to avoid obscuring the log sum formula of probabilistic welfare calculation. But, yes, I should have addressed this issue, too.
4. Urbana-Champaign gets colder. Wildly colder, though with no snow yet. One day the weather changed so extreme. In the morning it was 60 but it dropped to 23 in the evening.
5. Out of tiredness and boredom came desperation for new environment. To Borders I went (where else in this town, given limited time -- as tight as it can be?). No particularly interesting new books found. But got so tempted to buy Hilary Hahn's new album . And failed the temptation: In fact not only bought Hahn's, but added two more: Royal Philharmonic Orchestra interpretation of Andrew Lloyd Webber (have been loooongg time of waiting) and Best of von Karajan. Hope these new babies will help me speed up the dissertation, to say the least :-)