Listening to Charles Franklin in Journalism 4953 really made me have a greater appreciation for pollsters and statisticians. During out discussion we spoke about the statistics behind a small sampling size and why a seemingly small sample can be representative of the population.
This concept has always been particularly baffling to me. How can a group of 700 people be representative of the entire state or how can 1,000 people be representative of the entire nation? The funny thing that Franklin pointed out in his discussion was that when a doctor takes a sample of blood from our body he/she will only take a small vial. Do we consider the vial of blood to be unrepresentative and instead suggest that the doctor take a gallon? Of course not. Similarly, a small piece of the population can represent the whole but the key is that the population has to be representative and the poll has to be unbiased and it takes a proportional sample from each demographic group in the population.
An example Franklin used was taking a spoon full of soup. If the seasonings in the soup are evenly mixed and you take a spoonfull to taste then it will give you an accurate sampling of the whole soup. But if you take a spoonfull of soup where the seasoning is not well mixed the spoon will not be a good reflection of how the whole soup is going to taste. In polling, if the sample is not evenly mixed throughout the population then it will be like the unmixed seasoning and the sample will not represent the population.
Although Charles Franklin's lecture did not make me jump out of my seat and want to become a statistician, he did help simplify the worlds of polling and statistics for me. His simple analagies really made me better understand the thinking behind polling and I definitely appreciate polls and statistics more than I did before.
This concept has always been particularly baffling to me. How can a group of 700 people be representative of the entire state or how can 1,000 people be representative of the entire nation? The funny thing that Franklin pointed out in his discussion was that when a doctor takes a sample of blood from our body he/she will only take a small vial. Do we consider the vial of blood to be unrepresentative and instead suggest that the doctor take a gallon? Of course not. Similarly, a small piece of the population can represent the whole but the key is that the population has to be representative and the poll has to be unbiased and it takes a proportional sample from each demographic group in the population.
An example Franklin used was taking a spoon full of soup. If the seasonings in the soup are evenly mixed and you take a spoonfull to taste then it will give you an accurate sampling of the whole soup. But if you take a spoonfull of soup where the seasoning is not well mixed the spoon will not be a good reflection of how the whole soup is going to taste. In polling, if the sample is not evenly mixed throughout the population then it will be like the unmixed seasoning and the sample will not represent the population.
Although Charles Franklin's lecture did not make me jump out of my seat and want to become a statistician, he did help simplify the worlds of polling and statistics for me. His simple analagies really made me better understand the thinking behind polling and I definitely appreciate polls and statistics more than I did before.
RSS Feed