Monthly Archives: April 2014

Charitable Giving

This post won’t be nearly as entertaining as some of the others, but it’s analysis I’ve been working on so I thought I’d include it. This research is part of a term research paper in my Labor Economics course that I will be turning in a final draft of on Friday.

All data comes from the Bureau of Labor Statistics’ Consumer Expenditures Survey which is one helluva data source that I have plans for in the future of this blog. It tracks how people spend their money and is very easy to access and use. Unfortunately for me, around the 1970s an organization called Giving USA started tracking charitable giving data and in the last decade put it behind a pay-wall. So the best thing the BLS has is a statistics called ‘cash contributions’, which combines all money given to 501(c)3s with all money that is part of continuing legal claims (alimony/child support), not sure why. But it’s fine because I looked at variation in charitable giving as a portion of income with respect to years and there’s no reason to expect significant variance in court settlements. Specifically I used 2007, 2008, 2009 to capture the most recent US economic downturn and because I am building on research by Russell James at the University of Georgia.

1. Found average cash contribution by income bracket
2. Found average income within each income bracket
3. step 1/step 2= % of income contributed

and now we’re just a few simple t-test away from seeing the AMAZING RESULTSt test

Oh. Nevermind.

The variance between the sets of contributions as a % of income is not statistically significant.

Why not? A couple reasons
– Income Effect: several economists have studied charitable giving as a fixed portion of income, meaning that any household with x income will give y in charitable contributions
– Substitution Effect: other economists argue that in economic downturns there is no income effect, only a substitution effect that makes the marginal value of each dollar given to charity more valuable
– Survey limitations: the CEX attempts to reflect a set number of households from each income bracket in each year’s survey, so the survey does a poor job reflecting the population’s shifting between brackets as it does in an economic downturn
– Income brackets: this is the main one for me. The CEX’s income brackets are not equal, they skew way to the left and they cut off at $70,000+, meaning the upward sloping portion of the theoretical U Curve at charitable giving normally forms is out the window

other graph



And this time I wasn’t even trying to use shitty data.


Ode to BAZOOKA Font


Pictured above is BAZOOKA font. In 2002 I was in Mrs. Compton’s third grade class and I wrote a report about the sun entirely in this font. Mrs. Compton retired at the end of the year.

Recently I wanted to use this font again but found out it’s not included in my version of Microsoft Office which got me wondering what the font selection has looked like over the years in Office. Crazy this dataset wasn’t already made so I made my own starting with Office 1997 and including each version up to the most recent, Office 2013.

First here’s the count for each set. Who the hell makes the decisions about fonts.


1. Who the hell makes these decisions? How do you go from 67 to 26 to 217?
2. It sucks to be MS Mincho, Lucida Sans Demibold Roman, and Lucida Sans Demibold Italic, who got left out of 2010

For our protagonists of this analysis we have the original 20 fonts from 1997

Arial Narrow Arial Narrow Bold Arial Narrow Bold Italic Arial Narrow Italic Arial Black Bookman Old Style Bookman Old Style Bold Bookman Old Style Bold Italic Bookman Old Style Italic Garamond Garamond Bold Garamond Italic Haettenschweiler Impact MapSymbols MT Extra Monotype Sorts MS Outlook Tahoma Tahoma Bold

It would have been cooler if they were all presented in the font they’re named for (file under: ideas for future intern)

Here they as they exist in the other versions
the numbers in the right hand column are probabilities, not percentages, if you didn’t have that figured out then I don’t have much to offer you. Shouts out to Trevor “Hatin’ Ass Krugman” McCormick for this edit

And here are their individual heroic journeys: ThE fElLoWsHiP oF tHe FoNtS

fonts fonts2  fonts4 fonts5 fonts3fonts6


There’s some pretty incredible story lines in there.
– Just when I lost faith that any of the fonts would make the immolating fires of 2013 Arial Black and Impact rise from the ashes after staying dead 1337% longer than Jesus Christ (see, there’s precedent Nelson!).
– It looked like we were going to lose Haettenschweiler for sure after 2003, but its graphs looks exactly like my EKG as I was recording its data
– You would think that after the fonts offered increased 8.34x from 2003 to 2007 you would think that some of our compatriots we lost in that fight would come back, but because Microsoft’s typographers give fuck all about logic there’s only a 16% chance the font was brought back in 2007. Shouts out Arial Narrow Italic.


Anyway, back to that kick ass BAZOOKA font. It has never existed in any Office suite ever offered. Maybe the rubber cement was stronger in 2002, check back next time for that analysis.

Want my data? God help your soul

Don Jon’s Market Utility Ratio



I recently watched the movie Don Jon and by watched I mean that my roommate watched it and described it to me. On two separate occasions visiting the confessional booth Don, Jon, or Don Jon, I never caught his name, confesses to masturbating 17 times and having sex twice in one week and masturbating 22 in the next week. He seemed pretty satisfied with himself so let’s assume he gets the same utility out of both combinations. What he basically have here is a lesson in production possibility frontiers but nix guns & butter for vaseline & Trojan condoms.


Using Don Jon’s numbers:
Sex w/Human = 2.5(Masturbation) meaning he would need to have sex 8.8 times before he wouldn’t masturbate. Jesus.

DJ is clearly an extraordinary man, but how does he compare to the market?
– At their cheapest masturbation and sex are free, so I’ll be looking (very poorly) for what their cost options look like and comparing that ratio to DJ’s
– Took the mean from some 2006 data on street prostitutes and then calculated inflation, because that seemed reasonable
– Here at we pride ourselves on excellent data collection, so the cost of masturbation is calculated using the average of porn rental estimates from Yahoo Answers

Sex = $87.91
Masturbation = $18.333333

87.91/18.3333333 = 4.795

Good Ol Don Jon is overvaluing masturbation by nearly 192%. Damn.

Pokemon Thoughts pt. 1


This will likely be part of series about Pokemon because it’s very data rich.

For my first post I decided I wanted to try to settle a disagreement between two friends and avid Pokemon Trainers who represent two different battling ideologies. The first advocates an all out aggressive approach to battling, where your highest powered, most accurate move is always your optimum strategy, because after all the goal of a battle is to kill the other pokemon. The second takes a more finessed approach which favors status and stat altering moves.

The difference between these two strategies got  boiled down to a debate over the usefulness of the move Sand-Attack, a move which does no damage but lowers your opponents accuracy. The first trainer would never use it and the move would be replaced by the first option that does damage. The second would keep the move and argues that it is more beneficial in the long run than a move that does damage. These are of course two extreme positions taken for the sake of making the analysis easier.

So is Sand-Attack ever more beneficial than the next best alternative?

First, some controls and assumptions

– This analysis uses a Lv. 5 Pidgey with stats that are averaged from all possible Lv. 5 Pidgeys and the opponent is simply a leveled up version of the same Pidgey
– The first move is granted to the user, ceteris paribus
– The optimum strategy for weaker and equally strong pokemon is always to spam your strongest move, killing the opponent in the fewest turns possible
– Critical hits do not exist in this analysis
– This is theoretical rather than practical because you probably won’t be battling an infinite number of Pidgeys an infinite number of times

Pidgey  LV.  5               Gust                    Sand-Attack
HP 21.5                 Power      40             Power      0
Atk 12.5                Accuracy 100            Accuracy 100


%change in accuracy= 3/(3+x) where (x) is the number of times Sand-Attack is used

stats increases= original stat*(1/50*base stat)



Stats labeled ‘Turns’ is the number of turns that our level 5 Pidgey would survive in any given battle

Numbers are not rounded as they would be in real gameplay to reflect the average


Unsurprisingly, we see that using Sand-Attack as your first move (SA1) increases the number of turns you will have in a fight on average because it increases the probability that your opponent will miss. Also note the diminishing returns caused by your opponent needing fewer turns to kill you as it gets stronger. (X-axis=level of opponent, Y-axis=# of turns before our valiant hero is dead).

Next take a look at a comparison between the percentage of health of the opponent depleted before our noble lad is struck down (SA1 again denotes when Sand-Attack is used as the first move and %HPdiff is the difference between the two lines)


Interestingly, a score above 1 means using Sand-Attack on average allows a level 5 Pidgey to say “domo arigato, Mr. Pidgeotto” to a level 6 opponent. As the opponent grows stronger the benefit of Sand-Attack decreases simply because there are fewer turns for the opponent to miss.

So there you go, first post, Sand-Attack isn’t worthless. However, if you’re battling a level 5 Pidgey against stronger Pidgeys and you have no other options you’re probably as shitty of a Pokemon Trainer as I am a data analyst.

Want my data? God help your soul and