Abstract: . . .  psycholinguistics Page 1 Pia Knoeferle & M. W. Crocker Department of Computational Linguistics Saarland University SS 2006 Introduction to Psycholinguistics Lecture 4 Experimental Methods I P. Knoeferle 2 Overview What’s an experiment and why do we run experiments ? Empirical research cycle Building statistical models for observed data Standard deviation Frequency distributions Samples: representative of the population? Standard error Hypothesis testing Choosing a statistical test Field, 2005; . . .
. . .  psycholinguistics Page 1 Pia Knoeferle & M. W. Crocker Department of Computational Linguistics Saarland University SS 2006 Introduction to Psycholinguistics Lecture 4 Experimental Methods I P. Knoeferle 2 Overview What’s an experiment and why do we run experiments ? Empirical research cycle Building statistical models for observed data Standard deviation Frequency distributions Samples: representative of the population? Standard error Hypothesis testing Choosing a statistical test Field, 2005; Howell, 2004 P. . . .
. . .  psycholinguistics Page 1 Pia Knoeferle & M. W. Crocker Department of Computational Linguistics Saarland University SS 2006 Introduction to Psycholinguistics Lecture 4 Experimental Methods I P. Knoeferle 2 Overview What’s an experiment and why do we run experiments ? Empirical research cycle Building statistical models for observed data Standard deviation Frequency distributions Samples: representative of the population? Standard error Hypothesis testing Choosing a statistical test Field, 2005; Howell, 2004 P. Knoeferle 3 Why do we run experiments? To answer a research question . . .
. . .  N-1 Variance : average error between mean and observations Standard deviation : square root of the variance Small s relative to mean Data points are close to mean The mean is an accurate representation of the data s equal to zero would mean ? var iance ( s 2 ) = SS N 1 = ( x i x ) 2 N 1 = 6.8 4 =1.7 Why N-1?: Degrees of freedom s = ( x i x ) 2 N 1 = 1.7 =1.3 P. Knoeferle 19 Why N-1?: Degrees of freedom Building statistical models Degrees of freedom Number of observations that are free to vary Example: Number of languages a CL knows Sample of four observations from a population: can vary freely If we use this sample to calculate standard deviation: we have to use mean of the sample as an estimate of the population mean (i.e., we hold one parameter . . .
--3000,4,375,2638,17987