My response for this week will not be a consistent look at chapter 7, but a selective address of a few terms in a random format.
"The Sample with Built-in Bias"
When reading about this poor use of statistics I am reminded of the way the electoral college works in the United States. The "sample with built-in bias" may appear at first to be an acceptable statistic, but actually contains a specific lean. In a similar way, people would like to believe that the electoral college reflects the popular vote, when in reality, it can be far from it. The example in the book demonstrates bias in a statistic because many people chose not to answer a survey question. The result was a statistic that conveniently ignores a problem possibly with the question. Similarly, in our elections, people might draw the conclusion that Donald Trump won the popular vote, or even came significantly close. The electoral college, however functions much like the "built-in bias." Donald Trump won by a wide margin in the electoral college, but Hillary had more of the popular vote. People might draw the conclusion that more voters supported Donald Trump, but that is not true.
"The Gee-Whiz Graph"
Graphs have the potential to look like strong confirmation without necessarily being very strong. An example of this that I have encountered is when I studied the connection between lead poisoning and crime. On one website I found, run by a fanatic about the issue, there were graphs that showed nearly identical curves indicating that crime waves have followed the use of lead in the past. While these graphs almost look conclusive, they involved a great amount of manipulation and adjustment. The use of lead in gasoline over the 20th century was compared to the incidence of crime. These graphs were both manipulated to create a sense that they are closely linked. Margins, time, and units of measure were all adjusted in order to match the two graphs. While I take this example with a grain of salt, I still think that it worsened the argument because there was so much manipulation involved.
Reluctant Evidence
Reluctant evidence is described as evidence that a position held by an opposing party is or has been problematic. Immediately I think of arguments against the Clintons that were brought up during the 2016 presidential race. Bill Clinton, for example, supported and signed into law strong measures against drug-related crime. The legislation he supported resulted in rapidly increased mass incarceration. While Clinton was a liberal then and is one now, this position has been used to demonstrate failure and a somewhat conservative lean. In the social media firestorm that happened prior to the election, this and many other pieces of reluctant evidence were used to undermine people's support of Hillary.
"The Sample with Built-in Bias"
When reading about this poor use of statistics I am reminded of the way the electoral college works in the United States. The "sample with built-in bias" may appear at first to be an acceptable statistic, but actually contains a specific lean. In a similar way, people would like to believe that the electoral college reflects the popular vote, when in reality, it can be far from it. The example in the book demonstrates bias in a statistic because many people chose not to answer a survey question. The result was a statistic that conveniently ignores a problem possibly with the question. Similarly, in our elections, people might draw the conclusion that Donald Trump won the popular vote, or even came significantly close. The electoral college, however functions much like the "built-in bias." Donald Trump won by a wide margin in the electoral college, but Hillary had more of the popular vote. People might draw the conclusion that more voters supported Donald Trump, but that is not true.
"The Gee-Whiz Graph"
Graphs have the potential to look like strong confirmation without necessarily being very strong. An example of this that I have encountered is when I studied the connection between lead poisoning and crime. On one website I found, run by a fanatic about the issue, there were graphs that showed nearly identical curves indicating that crime waves have followed the use of lead in the past. While these graphs almost look conclusive, they involved a great amount of manipulation and adjustment. The use of lead in gasoline over the 20th century was compared to the incidence of crime. These graphs were both manipulated to create a sense that they are closely linked. Margins, time, and units of measure were all adjusted in order to match the two graphs. While I take this example with a grain of salt, I still think that it worsened the argument because there was so much manipulation involved.
Reluctant Evidence
Reluctant evidence is described as evidence that a position held by an opposing party is or has been problematic. Immediately I think of arguments against the Clintons that were brought up during the 2016 presidential race. Bill Clinton, for example, supported and signed into law strong measures against drug-related crime. The legislation he supported resulted in rapidly increased mass incarceration. While Clinton was a liberal then and is one now, this position has been used to demonstrate failure and a somewhat conservative lean. In the social media firestorm that happened prior to the election, this and many other pieces of reluctant evidence were used to undermine people's support of Hillary.
Comments
Post a Comment