“Stories about anti-semitism on college campuses are both interesting and important, for example, precisely because they are an expression of a broader societal trend.”
Broader trend, yes and we should shudder.
It is interesting you seem (if I am not wrong)casually intellectualise hatred and falsehoods spewing from newspaper articles and from those students’mouths. What these students do is not a reflection of a trend.Trends come and go, but anti semitism has never not been a trend. It has been unfortunately established thousands of years ago, it seems it is here to stay.
When you are portrayed the way Jews are (as the killers) and the actual terrorists as martyrs and brave, there is something seriously wrong with these young people. Journalists are working for a business and most would write about any dirt to sell more papers ( investigative reporting excluded) But, how about the impact these misguided student have on the younger generations? What a 12 year old finds cool is his brother’s friends in college celebrating on 7 October the massacre, gloating celebrating how many Jewish babies were killed and hoping for more.
I am not certain we can take anti semitism lightly.
"Even one hundred college students who say outlandish things won’t allow you to draw any conclusions about the 20 million. 100 college students is still a mere .0005 percent of them!"
First of all, a sample of 100, if representative of the target population, definitely can provide a good deal of useful information. Secondly, when one is sampling from a target population the total proportion of the target population that is sampled does not matter. All that matters is the size (and representativeness) of the sample.
Right, and thanks for pointing this out. Still, "you need a large, representative sample" seems like a good approximation to the truth in a blog post for a general audience. It is true, as you point out, that you can make do with a small sample *in relative terms* provided that it is representative and sufficiently large on some absolute scale: 2 randomly selected individuals still won't tell you a lot about the population as a whole. It is also true that you can make do with an unrepresentative sample if it's large enough (in relative terms): once you've sampled 98% of the population, for example, you can confidently make dependable inferences about the population as a whole. In the kind of case that I have in mind, though, the relevant population is neither large nor representative. I'll try to clarify the text to avoid misunderstandings.
You may know this cool trick. If you take 2 randomly selected individuals, A and B, there's a 50% chance that A is above the median for the variable you're interested in, and the same goes for B. Of course, there's also a 50% chance that A is below the median, and the same goes for B. So the chance that the median value is higher than both or lower than both is 2*.5^2 = 50%. Generally speaking, if you sample N individuals, the chance the median is higher than all of them or lower than all of them is 2*.5^N. If N = 5, that comes to 6.25%. This is where we get Hubbard's "Rule of Five": If you take 5 random samples, there's a very good chance, 93.75%, that the median is between the highest and lowest values.
" Serious journalists who write those stories should reconsider their career and life choices."
love. :)
“Stories about anti-semitism on college campuses are both interesting and important, for example, precisely because they are an expression of a broader societal trend.”
Broader trend, yes and we should shudder.
It is interesting you seem (if I am not wrong)casually intellectualise hatred and falsehoods spewing from newspaper articles and from those students’mouths. What these students do is not a reflection of a trend.Trends come and go, but anti semitism has never not been a trend. It has been unfortunately established thousands of years ago, it seems it is here to stay.
When you are portrayed the way Jews are (as the killers) and the actual terrorists as martyrs and brave, there is something seriously wrong with these young people. Journalists are working for a business and most would write about any dirt to sell more papers ( investigative reporting excluded) But, how about the impact these misguided student have on the younger generations? What a 12 year old finds cool is his brother’s friends in college celebrating on 7 October the massacre, gloating celebrating how many Jewish babies were killed and hoping for more.
I am not certain we can take anti semitism lightly.
This is a good article. But...
This part is a bit misleading -
"Even one hundred college students who say outlandish things won’t allow you to draw any conclusions about the 20 million. 100 college students is still a mere .0005 percent of them!"
First of all, a sample of 100, if representative of the target population, definitely can provide a good deal of useful information. Secondly, when one is sampling from a target population the total proportion of the target population that is sampled does not matter. All that matters is the size (and representativeness) of the sample.
Right, and thanks for pointing this out. Still, "you need a large, representative sample" seems like a good approximation to the truth in a blog post for a general audience. It is true, as you point out, that you can make do with a small sample *in relative terms* provided that it is representative and sufficiently large on some absolute scale: 2 randomly selected individuals still won't tell you a lot about the population as a whole. It is also true that you can make do with an unrepresentative sample if it's large enough (in relative terms): once you've sampled 98% of the population, for example, you can confidently make dependable inferences about the population as a whole. In the kind of case that I have in mind, though, the relevant population is neither large nor representative. I'll try to clarify the text to avoid misunderstandings.
Thank you for this response, which shows how my own comment needed to be qualified.
Finding sentences that approximate the truth in just the right way is always a challenge when writing for a popular audience (or indeed any audience).
You may know this cool trick. If you take 2 randomly selected individuals, A and B, there's a 50% chance that A is above the median for the variable you're interested in, and the same goes for B. Of course, there's also a 50% chance that A is below the median, and the same goes for B. So the chance that the median value is higher than both or lower than both is 2*.5^2 = 50%. Generally speaking, if you sample N individuals, the chance the median is higher than all of them or lower than all of them is 2*.5^N. If N = 5, that comes to 6.25%. This is where we get Hubbard's "Rule of Five": If you take 5 random samples, there's a very good chance, 93.75%, that the median is between the highest and lowest values.