Then there is the man who drowned crossing a stream with an average depth of six inches. ~W.I.E. Gates
Almost
all published statistical data consist of fatal flaws that one can never be rid
of. For example, bias is just one. For the purpose of the political election,
we will use the current statistics as our example. Currently the Washington
Post is showing a poll that has Mitt Romney leading by 47% to President Obamas
46%. So, how does bias play a role in this data? First off how was the poll
conducted? On the bottom of the poll in small letters I read this;
Source: This Washington
Post-ABC News poll was conducted by telephone July 5 to 8, 2012, among a random
sample of 1,003 adults. The results have a margin of sampling error of plus or
minus four percentage points for the full sample and four points for the sample
of 855 registered voters. Sampling, data collection and tabulation by Abt-SRBI
of New York. Results may not add to 100 percent because ''Other/don't know''
not included. Full results available at www.washingtonpost.com/polls.
By
looking at how the poll was conducted we can see where the bias affects the
results. The poll consists of 1,003 adults whom were called on the phone in
July. According to the census bureau taken in 2011, the population of the
United States is 311,591,917. Lets assume that half are adults, that leaves
us with 155,795,958.5 people to be represented by only 1,003. Another way to
view this, 1 person surveyed represents the opinions of 155,329 people.
That’s only part of it, depending on what time the calls were made will
determine who may have answered the phone. Phone calls made in the middle of
the day are more likely to be answered by people that are either working from
home, are unemployed, or stay at home moms raising children.
Lets
assume that they did call during the day, and lets further assume that it was
stay at home moms that answered, they may be more apt to side with Romney
because of the story of his wife Ann. That is only one small example of how a
bias plays a role in statistics. Furthermore, it is impossible to eliminate all
bias in a survey. If you change the time of the call, you may be largely
eliminating the group of people that work at night, or stay at home moms, or
even those that have evening actives. A factor that is unknown to the reader in
this poll is where in the nation were these people called. The results would be
different if the call was made only to urban location versus a farming
community.
The truth is relative only to the data that is presented. The author of whatever version of the “truth” you are viewing could have manipulated the data as to gain a more favorable result.
I always keep in mind that there are 3 "sides" to truth......yours, mine, and the truth!!
ReplyDelete