Interaction variables yet again

This is the first time I've put any thoughts on this but I had thought that everyone had read the post on New Economist and the links in the post. The catalyst for this post was a paper by Matthew Kotchen and Laura Grant, "Does Daylight Saving Time Save Energy? Evidence from a Natural Experiment in Indiana". What bothered me a little was that they labeled their paper "Very Preliminary" yet allowed themselves to present their findings in the Wall Street Journal. See Marginal Revolution's post for the link. What bothered me even more was that this paper was presented at the NBER and no one seemed to have caught the error. This indicates to me that economists make this error more often than not -- just like political scientists. See this link for the same error made by political scientists.

However, I have to admit that even after getting a PhD I was not pointed to this type of error until I started working - by statisticians. One analyst (another PhD economist) proposed estimating an equation with interaction terms but without including all of the variables as main (or level) variables. The statistician on the project had to point out that this is not correct. I can now see why this is the case. For instance, let's say we want to estimate:

Y = a + bX1 + cX2 + dX1X2

1. From an ANOVA standpoint there is no reason to exclude X1 and X2 separately (one or both) and just include X1X2.
2. Leaving out one of the main effects (or level variables), for instance, X2 is tantamount to assuming/imposing the restriction c = 0. There is no a priori reason to do this. Econometrics lets us test this restriction and there really is no harm to keeping it in.
3. Leaving out one variable is similar to doing model selection by dropping insignificant variables but in this case the authors do not test that this is the case. In any case, even if a variable is not statistically significant there is still no good reason to drop the variable in these types of analyses.
4. At most analysts should consider including the variable as a main effect as part of sensitivity analysis (even if they do not believe that the variable should be included as a main effect).

In their paper, Kotchen and Grant focus on the coefficient of the interaction, d, in this case which they use to support their claim that DST increases energy usage. My guess is that if they were to estimate the model correctly, the size of the coefficient, d, would fall. Right now their estimates of d are partially capturing the effects of the omitted variable. I suppose the other possibility is that including all the relevant variables as main effects could have resulted in some perfect collinearity although they don't indicate this is the case.

Some thoughts on DST

A post at MR and even earlier at Half Changed World got me thinking about DST and its effects on power usage. Here are 2 possible changes:
1. Usage does not change - it stays the same but just slides along by the hour change.
2. It changes at the margins at the beginning and at the end of the day.
Both of these indicate that at best the effects of DST on are small. #2 indicates that as the weeks pass the changes at the margin are not constant across the entire DST period.

A paper by Ryan Kellogg and Hendrik Wolff estimate the effects of DST on energy use based on an "experiment" in Australia when because of the Olympics, 2 states in Australia extended their DST by 2 months. This paper used half hourly data on energy use in the two states in Australia (1 that extended DST and another that did not -- another state which hosted the Olympics was not included. ) This was an impressive attempt at trying to obtain the causal effect of DST. They find that extending DST did not save energy which with some introspection is not unlikely. There were some weaknesses in the paper none of which can be attributed to the authors:
1. The aggregated nature of the data did not allow them to separately identify the effects of DST on residential versus industry energy demand.
2. The two extra months of DST did not allow them to measure the effects of DST over the entire DST period which runs from March through October.
Overall, Figures 3 and 6 in the paper pretty much summarized their findings although they had to estimate whether the increase in consumption early in the day outweighed the savings at the end of the day (Figure 6). The technical details of their estimator was a little overwhelming for me however. The title of the paper is: Does Extending Daylight Savings Time Save Energy? Evidence from An Australian Experiment. Their paper can be found here: http://repositories.cdlib.org/ucei/csem/CSEMWP-163/

Some of the deficiencies in the Kellogg and Wolff data were overcome by Matthew Kotchen and Laura Grant "Does Daylight Savings Time Save Energy? Evidence from a Natural Experiment in Indiana". They take advantage of the fact that some counties in Indiana never observed DST while some did and in 2006 a law forced all counties to comply. This allowed for indentification of the effects of daylight savings. They focus on residential energy use -- their data are the monthly bills of all residential customers in Indiana for three years (2004, 2005, 2006) which allows them to also estimate the effects of DST over the entired period covered by DST. They find that DST actually increases energy usage! This finding made it to the general public on various blogs and news releases. Unfortunately, their findings are marred by an error in their specification. They use interaction variables to identify the effect of DST but fail to include all of the interactions as main effects as well. More on this here. I liked their attempt to confirm their findings using a DOE simulation of home energy use, though. Their findings (?) however are not really representative of the country.
Their paper can be found here (at the time of writing): http://www2.bren.ucsb.edu/~kotchen/links/DSTpaper.pdf

Final thoughts: If all we're interested in are the differences in energy use due to the one hour difference, it seems like we could look at energy use for towns that are sufficiently "close" to the border of different time zones and just look at the differences in differences of those as well. Unfortunately, the dividing line for time zones usually go through rural areas. Perhaps I'm missing something because this seems like another obvious place to look. Another obvious comparison is Arizona with another state since AZ does not observe DST as well (?) The drawback here is that AZ is not comparable to other states since differences in differences will not work in this case.

Shrill bumper stickers

buck fush

To ski or not to ski?

ELB at Half Changed World struggled with this. We're having the same kinds of issues with horseback riding. Right now, skiing tops horseback riding. As much as my impression that skiing is for rich snotty kids, horseback riding is even for richer snottier kids. While I'm ranting about this, what not throw in tennis and golf lessons as well?

In any case, like ELB here's a list of Pros and Cons (I'm having trouble coming up with any Pros actually so I'll start with the Cons.)
Cons:
1. It's expensive.
2. It's not a life skill that is useful like swimming.
3. It's associated with rich snotty kids who drink champagne while discussing horses asses.
Pros:
1. It gets them closer to nature.
2. It might teach some responsibility and how to care for animals.

All in all, maybe it's not too bad since we are breaking down and considering the Rock Creek Horse Center which is still pricey.

1 WEEK SESSION $500 (non-refundable)
Monday thru Friday 9:00 am - 3:00 pm
For boys and girls ages 8-14
From the beginner to the more experienced, our campers learn to ride and care for horses at all levels of accomplishment. Each session includes instruction, lectures on horsemanship theory and safety, as well as trail rides, picnics and games.

We're hoping that horse care and stable cleaning are part of the lessons -- they should learn that it's not all fun and games and there responsibilities as well.

Do sales encourage hoarding?

Well, not hoarding as when there are short supplies but this title sounded catchier. Whenever there are sales I tend to buy a lot more than I need at the time rationalizing that I will eventually get around to using whatever it is that is on sale: toilet paper, paper towels, soap, shampoo, etc. A consequence of all this is that I need to find room for all this stuff. So the question now is: Do sales (and Cotsco) cause us to want more room in our houses, or even catchier (and probably more inaccurate), do sales and Cotsco cause McMansions?

Is suicide rational?

I used to think it was but this article "The Golden Gate: A Bridge Too Deadly?" convinced me otherwise:
Gazing at the Pacific Ocean over a railing only four feet high, he found the sole remaining impediment to his death was his own willpower, which turned out to be fleeting. "I counted to 10, and I couldn't do it," Baldwin said. "And I counted to 10 again, and I vaulted over. "And my hands were the last thing to leave, and once they left, I thought: 'This is the worst decision I've ever made in my life.' " CalTrans officials point to a University of California survey's finding that nine out of 10 people prevented from jumping off the Golden Gate were still alive years later or had died of natural causes, despite the rationale that a barrier would prompt them only to "go somewhere else to end it." The study is part of a growing body of scientific literature that explodes persistent myths about suicide while reinforcing a simple principle: When it is harder to kill oneself, fewer people do so.

Bad hotels

I was looking into Tripadvisor customer reviews at two places: Timberline Lodge in Oregon and Canaan Valley Resort in West Viriginia. Both places are different but they seem to offer great views (Oregon) or activities (Canaan Valley). Both had very mixed (tending toward the negative) reviews on the quality of accomodations (given the price).

I've noticed that the signal to noise ratio from:
1. web site pictures
2. brand names (e.g. Hilton, Marriott, etc.)
are very low. I don't know when the pictures are taken. I don't know if the hotel will no longer be a brand name in the next few months because it is not up to standard.

So I've relied on Tripadvisor and Yahoo ratings which when the results are mixed leave me with a very quesy feeling as on our trips to Penang (2005) where we stayed at the Gurney Hotel and Kuala Lumpur (2007) where we stayed at the Lanson. The Gurney billed it self as a 5 star hotel but by Best Western standards it was below Best Western. Although service was good and the people were friendly, the carpet was threadbare, the furniture felt cheap, the couch was stained, and the sheets had holes in them. The toilet kept overflowing even though they came every other day to fix it. The Lanson turned out to be a pleasant experience although it was a little far from KLCC.

In general, hotels seem to lack any desire to upgrade their facilities until they "have to". When is this "have to" point - when marginal costs of not renovating exceed marginal benefits. They have to close e.g. Boca Raton Resort and Club ("The Boca Raton Resort and Club is excited to announce a complete renovation of our Boca Beach Club, including the arrival, lobby, guestrooms, restaurants & bars, fitness center, children’s activity center, and pools. A stunning new Pool Oasis will feature three redesigned swimming pools, an oceanfront bar, additional beach access, upgraded cabanas plus sunning terraces, lounging platforms & lush landscape. The new Pool Oasis is scheduled to reopen April 1, 2008; hotel rooms, lobby, and the interiors are scheduled to reopen in 2009." accessed 3/10/2008) and what do they do with their employees at this time?

The marginal benefits for some hotels are exceptionally high for a lot of these places mainly because of their location and hence, their monopoly power i.e. they are located on an island, or the view of the Grand Canyon for instance, , the view of Red Rocks in Sedona, the Cascades from Timberline Lodge and in Canaan Valley the draw are the facilities (such as skiing) make it more likely that people will tolerate bad accomodations (for a high price).

Unfortunately, for people like me who prefer both nice accomodations and views/facilities these places turn me off.