Tuesday, April 14, 2015

Randomised Controlled Trial: The Gold Standard?

In designing applied social research by experiment, researcher can choose to use either randomised experiments or quasi-experiments (Hall 2008: 82). Randomised experiment – often called randomised control trials or RCT (Wade 1999 and Bryman 2008: 36) – was initially used in natural science research but know it becomes common research method use by social scientists (Bryman 2008: 35-37 and Hall 2008: 82). Stolberg, Norman and Trop (2004: 1539) state that the history of the RCT began long time ago in 600 B.C. by Daniel of Judah when he conducted clinical trial to assess the effect of vegetarian diet.

The RCT is a study in which people or participants are allocated at random to receive one of several interventions or treatments (Stolberg et al 2004: 1539). Hall (2008: 83) distinguishes the RCT and observational studies in a very important respect because the observational study examines phenomenon in the normal course of events but in the RCT, the researcher controls the conditions of the observation. The primary goal of RCT is to assess the intervention whether it works by comparing it to a control condition (West and Spring 2010). 

Many experts and researchers argued that the RCT is the “golden standard” for research method in social sciences. In my opinion, it might be true if the RCT will become the best standard to assess impact for most of the social researches, but it also depends on the several factors within and outside the research which enable or restrict the RCT to be implemented.

Wade (1999) and Bryman (2008) support the contention that the RCT is the “golden standard” for research methods, some of the researchers did not used the same term but they also implied that the RCT is the most powerful tool for research. The main reason for this is that – in principle – the method has the alternative answer or explanations for cause-effect relationships questions for treatment effects using randomisation process (Hall 2008: 82-89). Another reason is unbiasedness as an important aspect in statistics is the special concern of the RCT (Middleton and Aronow 2012: 2).

However, in the “real world” experiments, the RCT sometimes cannot be used due to threats to internal validity and also its limitations such as: ethical and moral concerns, attrition or experimental mortality, researcher expectancy, instrumentation and diffusion of treatment (Hall 2008: 89-90).

In contrast, Grossman and Mackenzie (2005: 516) argue that the RCT is not the “golden standard”, it is a good design for experimental research in some situations, but that's all. Moreover, Christ (2014) contends that the RCT is difficult and expensive to implement and may not suitable to the aim of the research. This argument also supported by Stolberg, Norman and Trop (2004: 1540) that the RCT may not feasible because of financial constrains. In addition, Lindsey (as cited by Donaldson and Christie 2004) strongly argues that the RCT is not a gold standard and the researchers are poorly served by the gold standard terminology. Lindsey said: ‘I think that for impact assessment randomised experiments are the worst methodology except for some of the others that have been tried from time to time. That is pretty much my theme here..’ (Donaldson and Christie 2004).

To conclude, many researchers and experts agreed that the RCT is the best method currently available for assessing program impact or casual-effect relationship. However, according to some experts, there are very view situations where the RCT can be successfully implemented. The claim that the randomised experiment is the gold standard for research methods in social sciences is only valid for selected researches or experiments in some circumstances, but it is not the gold standard for all experimental research in social sciences due to its limitations. So, if the meaning of the “gold standard” was the best for all researches in all circumstances, citing argument given by Berk (2005: 431), if the truth be told, there is no gold standard.***

Written by: Agung Wasono (April 2015)


Berk, R, A. (2005) Randomized Experiments as the Bronze Standard, Journal of Experimental Criminology, Vol (1): 417-433

Bryman, A. (2008) Social Research Methods (Third Edition), New York: Oxford University Press

Christ, T, W. (2014) Scientific-Based Research and Randomized Controlled Trials, the “Gold” Standard? Alternative Paradigms and Mixed Methodologies, SAGE Journals / Qualitative Inquiry, Vol 20(1) 72–80 available at: http://qix.sagepub.com/content/20/1/72.full.pdf+html (assessed 31 March 2015)

Donaldson, S, I, and Christie, C, A. (2004) The 2004 Claremont Debate: Lipsey vs. Scriven Determining Causality in Program Evaluation and Applied Research: Should Experimental Evidence Be the Gold Standard?, Journal of Multi Disciplinary Evaluation, Volume (JMDE: 3): 60 – 77, available at: http://journals.sfu.ca/jmde/index.php/jmde_1/article/viewFile/101/116 (assessed 29 March 2015)

Grossman, J. and Mackenzie, F, J. (2005) The Randomized Controlled Trial: Gold Standard, or Merely Standard?, Volume 48(4):516-34, available at: http://www.ncbi.nlm.nih.gov/pubmed/16227664 (assessed 31 March 2015)

Hall, R. (2008) Applied Social Research: Planning, designing and conducting real-world research, Melbourne: Palgrave Macmillan.

Middleton, J, A and Aronow, P, M. (2012) Unbiased Estimation of the Average Treatment Effect in Cluster-Randomized Experiments. Available at: http://www.joelmiddleton.com/uploads/4/3/1/4/4314860/clusterrand-2011-12-24.pdf (assessed 2 April 2015)