Tag Archives: statistics

Aging Population a Red Herring

Does The Aging Of The Population Really Drive The Demand For Health Care?

(from Concluding Comments)

The objective of this essay has been to deconstruct the popular myth that the aging of the population by itself is a major contributor to the annual increase in the demand for health care and, thus, to total national health spending…the bulk of the rapid annual growth in national spending in the past has been driven by other factors that increase per capita spending for all age groups. Key factors include rising per capita incomes, the availability of promising but costly new medical technology, workforce shortages that can drive up the unit cost of health care, and the asymmetric distribution of market power in health care that gives the supply side of the sector considerable sway over the demand side. These other factors will be the dominant drivers of health spending in the future as well. Blaming Medicare’s future economic pressures mainly on demographic factors beyond policymakers’ control is an evasion of more important challenges. [emphasis added]

Reinhardt, Uwe E. Does The Aging Of The Population Really Drive The Demand For Health Care? Health Affairs, 22, no.6 (2003):27-39.

In 1986, Congress enacted the Emergency Medical Treatment and Active Labor Act, which forced emergency care providers (hospitals) to provide care without reimbursement. Hospitals have only survived by shifting the cost of this uncompensated (and unfunded) care to those with the ability to pay. Oddly enough, Uwe does not mention this “key factor”: an implicit subsidy driving demand. Fortunately, some are starting to take notice, ex. Ezra Klein (see below).

Washington Post: Repeal EMTALA! Jan 27, 2011

Leave a Comment

Filed under Economics, Politics

Export Your Netflix Ratings

As I approach 500 rated films on Netflix, I started wondering what the distribution of my ratings looked like. I try to watch movies that I can reasonably expect to like (based on reviews, recommendations, etc.), but without missing out on films that didn’t draw a crowd or have mass appeal. Therefore, I would not expect my distribution to be centered on 3.

  • Rated titles: 484
  • Mean rating: 3.65
  • Percent titles over 3 stars: 59.3%

Average Rating and Count by Genre

To export your Netflix ratings, follow these steps

  1. Open Mozilla Firefox.
  2. Go to Firefox > Add-ons.
  3. Enter “greasemonkey” in the search field.
  4. Click the “Install” button.
  5. Click “Restart Now” on successful installation of Add-on.
  6. Go to Netflix Movie Ratings Extractor on userscripts.org, click “Install”.
  7. Go to Movies You’ve Seen tab on Netflix.com, scroll down to “Netflix Movie Ratings Extractor (Includes IMDB Movie Data Lookup)”.
  8. (I deselected the checkbox for “Not Interested”) Click “Start”.
  9. Copy and paste the output in to something useful, like Excel.
  10. Use PivotTables to manipulate stuff.

Update (9/24/11): The script’s author has been kind enough to make updates and I can confirm it is now working again!

Leave a Comment

Filed under Consumer, Film

Marriage: An Institution on the Decline

Today, I started to wonder whether there are freely accessible statistics on cohabitation. I’d like to be able to confirm a trend I see anecdotally, but I wasn’t able to find much on unmarried couples living together without children/dependents. However, I did come across a wealth of data on “living arrangements” over at the Census (Families and Living Arrangements).

Comparing detail tables (Table A1) from 1999-2010, I created the following chart. It shows, as a percentage of people age 20+ (‘adjusted adults’*); how many people are married and with their spouse or have never been married. These categories exclude those who are married to an absent spouse, widowed, divorced, or separated.

In just 12 years, there was a 3.5% decline in people that are married with their spouse present, and a 3.4% increase in people that have never been married. In that same time, the widowed decreased by 0.55%.

Arguments over who may access state-granted ‘marriage’ benefits (including its titular reference) may be obscuring more foundational changes in how people relate to the origins of the state and its institutions.

*Note: Redefining ‘adjusted adults’ as 18+, the changes are -3.3%, +3.1%, and -0.51%, respectively.


The table below shoes the increase, by age group, in people who have never been married (from 1999 to 2010). The biggest changes are in those aged 45 – 74 years old, who now represent late boomers and early Gen Xers.

20-24 years 27%
25-29 years 40%
30-34 years 20%
35-39 years 0%
40-44 years 18%
45-49 years 72%
50-54 years 147%
55-64 years 140%
65-74 years 54%
75-84 years 13%

1 Comment

Filed under Politics, Religion, Zeitgeist

On Statistical Inference and the Internet

People often use statistical inference to make decisions, regardless of how explicit their line of reasoning is. Without testing products before buying them, information that helps consumers make better inferences is one of the biggest differentiators between Internet retailers. For example, I patronize Newegg.com because their presentation of customer feedback and ability to sort search results by “Best Rating” has enabled me to quickly and confidently purchase products.

Currently, if you know you want the best blender, or the best in-ear headphones that Amazon.com sells, the best way to go about it is to narrow your search terms so that all or mostly all of the results fit in to the relevant product category and then sorting by Avg. Customer Review. Unfortunately, a shopper has to browse to the second page of search results to find a pair of in-ear headphones (852 total results, BTW) with a sample size greater than 12 (and with the majority being 1 or 2). The 36th result has a single 5-star rating, and the 37th product has a slightly lower mean rating but has 160 ratings. Sorting through hundreds of results and loading new pages is tedious and inefficient. Sorting by Avg. Customer Review should result in products with an adequate sample size to pass some minimum confidence that the mean rating actually conveys meaningful information.*

Are there any risks with this strategy? There could be. If weightings are based on internal sales (population) figures, a third party could work backwards based on the mean review and sample size (if presented). Outside of this risk, communicating better information helps both producers and consumers.

This applies to more than Internet retail. Google Maps allows you to limit search results by user rating. It is also simple to leave feedback. In conjunction with Yelp and Citysearch, this can provide me with the information I need on short notice to choose a restaurant in a new city, for example.

Leave a Comment

Filed under Consumer