Showing posts with label sample bias. Show all posts
Showing posts with label sample bias. Show all posts

Monday, July 2, 2012

Statistics: Bias and Sample Bias


In statistics we come across two types of errors, random errors and systematic errors. Random error is the error due to sampling variability or at times measurement precision. It occurs essentially in all quantitative studies and can be minimized to an extent but not avoided. We might wonder, What is Bias? Bias or Systematic error is a reproducible inaccuracy that produces a consistently false pattern of differences between observed (estimated) and true values. We get Definition Bias or definition of Bias as the systematic or average difference between the true value of the parameter and the experimental estimate value.

Bias Statistics or Statistical Bias is an error we cannot correct by repeating an experiment many times and averaging together the results, it is a directional error in an estimator.In a population which includes 50% males and 50% females, if we know that an important variable is the gender. Then the sample needs to include the same proportion. But if the sample includes 30% males then we can say that the results are likely biased as there are not enough responses from the male gender. This is how we know Sample Bias or in other words if the given sample is biased.

In statistics, Bias is used to refer the directional error in an estimator. Though we repeat the same experiment many times there would be some randomness existing.  By repeating an experiment we might see a slight variation in the estimation values. The bias is a systematic or average difference between the experimental estimate and the true value.

Examples of Bias
For example, if the national census could be completed online using the internet, then the sampling is said to be biased as only those people who could afford a computer and internet connection would be included and the people who cannot afford a computer or have the necessary skills to operate one will be excluded.

Jason was assigned by his editor to determine what most Americans think about the new law that will place a special tax on all electronic goods purchased and was told to mention in the survey form that he revenue returns collected from the tax would be made use of to enforce new online decency laws. Jason decided to use the email poll for convenience. In this poll 90% of those surveyed opposed the tax. Jason was quite surprised when 65% of all Americans voted for the taxes. So, online and call-in polls are particularly biased as the respondents are self-selected.

For example, a movie is released and a survey is done on the first day first show to know how people liked it. In the survey some say the movie was awesome and they all liked it and some say the movie was boring and they did not like it at all. Here ‘they’ might refer to only their group.