PSY 2174 Lecture Notes - Lecture 4: Observational Error, Sample Size Determination, Stratified Sampling
Document Summary
The easiest way to deceive is to change the y-axis. Using diversion tactics, such as size on plots. Show graphs when reporting statistics (disambiguates discrepancies) Sample: individuals from the population, who appear in study. Parametric statistic: refers to a statistical approach studying populations. Non-parametric: less powerful statistical approach that generally cannot make conclusions about populations. In order to make conclusions from sample to population, sample must be representative i. e. , it must reflect attributes of population. Representativeness is less of an issue if population has low variability among individuals. Sample that has different characteristics from population. Sampling bias: a recruitment process that favours the selection of certain individuals. Must know all individuals in a population. Random process of selection with equal chance of being selected. Simple random, systematic, stratified, proportionate stratified, and cluster. Requires that each individual has equal chance of being selected. Sample with or without replacement: with replacement: ensures independence of scores, without: no independence of scores.