The effects of noise on discerning differences between two sets of data



Minimal Noise 
Some Noise 
More Noise 
 Each ball represents a data point in the experiment.
 The two colors of balls represent two conditions of a hypothetical experiment (e.g., a yellow ball is a data point from condition A and a pink ball is a data point from condition B.
 The location along the top of the graph from which they drop is analogous to the true value of the dependent variable.
 The location along the bottom, where the balls fall, is analogous to the actual measured value for that trial.
 Each ball dropping simulation represets scenarios with different levels of noise in the experiment, represented by the rows of pegs that result in an actual measurement being diferent from the true value.
 Testing whether the two conditions are really different corresponds to looking only at the distribution of where the balls land and determining whether yellow balls and pink balls fell from different locations.
 The less noise, the fewer data points that are necessary to determine that the measured values in the two conditions are different.