PSY 301 Lecture Notes - Lecture 5: Coin Flipping, British Rail Class 360, Sampling Error
Document Summary
Data (observations)= our tie to reality/nature: so data need to be good. Data are a special class a type of model, all model are false. Same is true of data, but here means that all data are measured with error or the possibility of error: error does not mean blunder, means that data often differ- perhaps slightly- from what they represent. 4 types of error that can affect raw data, how data are manipulated: sampling error, human + technical, bias (unintentional, rounding, precision, accuracy (rpa, flawed analysis. Showed distribution on sc 78% of you got something besides 5/10 heads: reason- sampling error (no surprise) In a general sense, sampling error means we cannot exactly measure underlying probability (of heads) Yes, more data replication: as sample size increases, the sampling error of our average goes down- our average gets closer + closer to true value, ex. Coin flip: lots of ways to include replication in a study.
