RandomizedDesignEffectSizes function

RandomizedDesignEffectSizes

This function creates the theoretical effect sizes for data from one of four different distributions for specified parameter values for the distribution specified by the type parameter. It assumes there are two samples, one corresponding to a control group and the other to the treatment group. It returns the theoretical effect sizes for a fully randomized experiment.

RandomizedDesignEffectSizes(m1, std1, m2, std2, type = "n")

Arguments

• m1: The theoretical mean for the control group
• std1: The theoretical variance for the control group
• m2: The theoretical mean for the treatment group
• std2: The theoretical variance for the treatment group
• type: String identifying the distribution, 'n' for normal, 'ln' for lognormal, 'lap' for Laplace, 'g' for Gamma

Returns

dataframe containing the expected values of the unstandardized mean difference effect size, the pooled within group variance, the standardized mean difference effect size and the point bi-serial correlation.

Examples

RandomizedDesignEffectSizes(m1=0, std1=1, m2=1, std2=3, type = 'n')
#  ES Var     StdES      rPBS
#1  1   5 0.4472136 0.2182179
RandomizedDesignEffectSizes(m1=0, std1=1, m2=1, std2=3, type = 'l')
#        ES       Var     StdES        rPBS
#1 243.0432 242552663 0.0156056 0.007802562
 RandomizedDesignEffectSizes(m1=0, std1=1, m2=0.266, std2=1, type = 'l')
#          ES      Var     StdES       rPBS
# 1 0.5024232 6.310995 0.1999957 0.09950162

Author(s)

Barbara Kitchenham and Lech Madeyski