models & measurements
Chapter 5
There are two main analyses performed in a scientific context: measurement and model selection.
Here, we explore the topic of measurement. We see that models of the data are always a part of measurement, although we show that the presence of a model is sometimes 'transparent' to our calculations. We then develop a measurement algorithm that allows measurements to be made for a wide range of models. The utility of this algorithm is demonstrated in a series of worked examples, such as measuring rates (death rates, recovery rates, etc.), rate differences (differential recovery), times and time intervals (reaction time and qrs duration), means, variances, straightline slopes, exponential decay, and stimulus detectability (dprime).
Programming Asides:

computing confidence intervals [p281]

body temperature measurement [p285]

temperature measurement with multiplicative noise [p289]

binomial rate measurement [p294]

multinomial rate measurement [p295]

measuring the qrs duration [p300]

measuring saccade latency [p303]

straightline model i [p306]

priors in slope and angle [p309]

straightline model ii [p309]

dprime [p316]

complex observation functions in threshold detection [p323]

measuring exponentially decaying sensorymotor error [p328]

dark adaptation [p336]

multiplesource transparent measurement [p342]

temperature from multiple thermometers [p346]

measuring multiplesource slope parameter [p351]