ECO220Y1 Lecture Notes - Lecture 11: Unimodality, Non-Sampling Error, Normal Distribution
27
ECO220Y1 Full Course Notes
Verified Note
27 documents
Document Summary
Can be discrete or continuous, depending on population, statistic, sample size. Have distribution because are subject to sampling error. Find probability of every possible value of the sample statistic, using sampling rules: list all possible samples with n observations. Find statistic of interest for each possible sample. Draw samples and observe relative frequency of each value of sample statistic. Central limit theorem, laws of expectation and variance. Number of samples (number of simulation draws) sample size: sd( ) = sqrt(p(1-p)/n) 1: sample size assumption, randomization condition, 10% condition, success/failure condition. Np and nq must be at least 10 or greater than 3 standard deviations. Central limit theorem: as sample size approaches infinity, sample distribution of means is normal distribution. Shape of population distribution doesn"t matter: the larger the sample, the better the approximation, sample distribution of proportions is a special case of central limit theorem. Can vary from due to sampling error, non-sampling error, or parameter not being as.
