STAT 301 Lecture Notes - Lecture 14: Confidence Interval, Errors And Residuals, Statistical Parameter
Document Summary
Sampling error vs "plain old" error: the 5% error rate for 95% confidence intervals refers only to sampling error (sometimes called. "statistical error": this is the error that is just due to the randomness of the data. If we collect a new data set, we"ll get different values for our statistics: this error rate does not account for error that could arise from violating the assumption of normality. Other potential sources of error: confidence intervals attempt to account for sampling error. Ci does not account for: biased samples, faulty measuring devices, poor experimental design, the violation of underlying assumptions, using parameters that don"t reflect reality, failure to account for relevant differences between sub-groups. Module 5 summary: we make confidence intervals because population parameters are unknown. Chapter 6: we have seen that we reject h0 when a 95% ci does not contain the null value, to summarize.