Random Assignment

  • Gideon J. MellenberghEmail author


A substantial part of behavioral research is aimed at the testing of substantive hypotheses. In general, a hypothesis testing study investigates the causal influence of an independent variable (IV) on a dependent variable (DV). The discussion is restricted to IVs that can be manipulated by the researcher, such as, experimental (E-) and control (C-) conditions. Association between IV and DV does not imply that the IV has a causal influence on the DV. The association can be spurious because it is caused by an other variable (OV). OVs that cause spurious associations come from the (1) participant, (2) research situation, and (3) reactions of the participants to the research situation. If participants select their own (E- or C-) condition or others select a condition for them, the assignment to conditions is usually biased (e.g., males prefer the E-condition and females the C-condition), and participant variables (e.g., participants’ sex) may cause a spurious association between the IV and DV. This selection bias is a systematic error of a design. It is counteracted by random assignment of participants to conditions. Random assignment guarantees that all participant variables are related to the IV by chance, and turns systematic error into random error. Random errors decrease the precision of parameter estimates. Random error variance is reduced by including auxiliary variables into the randomized design. A randomized block design includes an auxiliary variable to divide the participants into relatively homogeneous blocks, and randomly assigns participants to the conditions per block. A covariate is an auxiliary variable that is used in the statistical analysis of the data to reduce the error variance. Cluster randomization randomly assigns clusters (e.g., classes of students) to conditions, which yields specific problems. Random assignment should not be confused with random selection. Random assignment controls for selection bias, whereas random selection makes possible to generalize study results of a sample to the population.


Cluster randomization Covariate Cross-over design Independent and dependent variables Random assignment and random selection Randomized block design 


  1. Cox, D. R. (2006). Principles of statistical inference. Cambridge, UK: Cambridge University Press.Google Scholar
  2. Hox, J. (2002). Multilevel analysis: Techniques and applications. Mahwah, NJ: Erlbaum.Google Scholar
  3. Lai, K., & Kelley, K. (2012). Accuracy in parameter estimation for ANCOVA and ANOVA contrasts: Sample size planning via narrow confidence intervals. British Journal of Mathematical and Statistical Psychology, 65, 350–370.PubMedGoogle Scholar
  4. McNeish, D., Stapleton, L. M., & Silverman, R. D. (2017). On the unnecessary ubiquity of hierarchical linear modelling. Psychological Methods, 22, 114–140.PubMedGoogle Scholar
  5. Murray, D. M., Varnell, S. P., & Blitstein, J. L. (2004). Design and analysis of group-randomized trials: A review of recent methodological developments. American Journal of Public Health, 94, 423–432.PubMedPubMedCentralGoogle Scholar
  6. Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis. London, UK: Sage.Google Scholar
  7. van Belle, G. (2002). Statistical rules of thumb. New York, NY: Wiley.Google Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  1. 1.Emeritus Professor Psychological Methods, Department of PsychologyUniversity of AmsterdamAmsterdamThe Netherlands

Personalised recommendations