Figure - uploaded by Patrick Onghena
Content may be subject to copyright.
Source publication
Randomization in scientific research refers to the deliberate introduction of a stochastic element in the design of a scientific study. Broadly defined, randomization may refer to any aspect of the study that contains a stochastic element. For example, items in a questionnaire may have a randomized order or the decision rule of a nonparametric test...
Contexts in source publication
Context 1
... number K increases rapidly as the number of participants increases. For 10 participants , 5 in the experimental group and 5 in the control group, there are 252 distinct assignments. For 20 participants, 10 in the experimental group and 10 in the control group, there are 184,756 distinct assignments. And for a more realistic scenario of 30 participants, 15 in the experimental group and 15 in the control group, there are already 155,117,520 distinct assignments. Notice that in these examples the number of participants in the experimental group and the number of participants in the control group are equal (so-called balanced designs), but that Equations 1 and 2 also apply for unequal numbers (unbalanced ...Context 2
... the numerical values of the confounding variables are unknown, but suppose in this case, for the purpose of demonstration, we have measured weight and length of the patients before the experiment and computed the BMI for the four patients to be respectively 13.7, kg/m², 14.1 kg/m², 15.2 kg/m², and 17.0 kg/m². The matrix of all possible assignments and the marginal averages are given in Table 1. Table 2 shows the averages for each treatment and the differences between the averages for each assignment. ...Context 3
... number K increases rapidly as the number of participants increases. For 10 participants , 5 in the experimental group and 5 in the control group, there are 252 distinct assignments. For 20 participants, 10 in the experimental group and 10 in the control group, there are 184,756 distinct assignments. And for a more realistic scenario of 30 participants, 15 in the experimental group and 15 in the control group, there are already 155,117,520 distinct assignments. Notice that in these examples the number of participants in the experimental group and the number of participants in the control group are equal (so-called balanced designs), but that Equations 1 and 2 also apply for unequal numbers (unbalanced ...Context 4
... the numerical values of the confounding variables are unknown, but suppose in this case, for the purpose of demonstration, we have measured weight and length of the patients before the experiment and computed the BMI for the four patients to be respectively 13.7, kg/m², 14.1 kg/m², 15.2 kg/m², and 17.0 kg/m². The matrix of all possible assignments and the marginal averages are given in Table 1. Table 2 shows the averages for each treatment and the differences between the averages for each assignment. ...