RSS
Twitter
Survey Says, US Net Surfers Bored Of Taking Online Surveys
from the please-refer-to-the-bubble-graph dept
According to a study released by Nielsen/Netratings, Americans spent 2% less time online in March than they did a year ago. Is there reason to be alarmed? Not really. In order for this data to be truly useful, we need see the amounts of time spent online broken down by tasks and have comparative metrics for their offline counterparts. Something like: taking online surveys (down 2%), reading email (up 9%), deleting spam (down 11%), surfing the web (up 13%), surfing the ocean (down 3%), working at the office (down 10%), spending time with kids (up 17%), and making sure you're up on the latest pointless statistics (up 13%). It's not that statistics aren't important, it's just that 50% of all statistics are made up anyway.
Reader Comments
(Flattened / Threaded)
How about statistics to really scare the **** out
The Social Welfare ministry has announced that, of pregnant women who test positive for the HIV virus, only 4-10% are confirmed to have the HIV virus upon further testing.
This represents the classical dilemma of the trade-off between sensitivity (the ability to detect the virus) and specificity (the ability to detect there is no virus).
For example, if 1 out of 10,000 people have the HIV virus, the test can detect 100% of infected people, and the test is able to give normal results 99.9% of the time for uninfected people, then there will still be 10 people falsely diagnosed with HIV infection, for every infected person who is correctly diagnosed -- thus, only about 10% of people who test positive for the virus will actually have the virus.
One could devise a test to improve the specificity to 99.99%. In that case, only 50% of all people who test positive will actually have the virus. But what if that causes the sensitivity to fall from 100% to 90%? Then we will have a far more sinister problem, where 1 out of 10 HIV patients will test negative for the virus, thus they will think they are "safe" and go around infecting more people with the virus.
The report acknowledges the heavy psychological burden that a "positive" result places upon pregnant women, thus physicians should provide better explanations of the test in advance. (Yeah right, they will.) How many women miscarry because of a false positive diagnosis, the report doesn't say.
http://headlines.yahoo.co.jp/hl?a=20050403-00000054-kyodo-soci
(reply to this comment) (link to this comment)
Re: How about statistics to really scare the ****
In practice, since HIV tests are expensive, blood samples from many people are mixed in a pool and tested at once. If the pool tests positive, then successively smaller pools are tested until the individual is identified. There are oceans of blood being mixed up in laboratories all over the world as you are reading this sentence.
The USA does not perform Tuberculosis vaccinations, because the (currently available) tests cannot distinguish between a vaccinated person and a diseased person. Tuberculosis is too rare here to be worth having a vaccination program. If someone invents a better test that can make the distinction, then that will be one more big syringe for baby.
In truth, vaccinations are not entirely effective either -- maybe 50-90% effective. But we might as well give them for some diseases, if it can stop the spread of the disease.
In other news, the Angolan outbreak of Marburg virus (ebola-like disease) has killed 132 out of 140 infected people. Marburg virus makes you literally melt, with blood spurting out of your hair pores.
http://www.medicalnewstoday.com/medicalnews.php?newsid=22171
(reply to this comment) (link to this comment)
Add Your Comment