Flexible design is a form of design which allows for interim feedback that may change the course of a trial or experiment. It’s sometimes used synonymously with adaptive design. However, there is a subtle difference: The term “adaptive design” refers to specific design types (e.g. adaptive allocation or sample size re-assessment), while “flexible design” is more of an informal term to refer to any design that isn’t rigid or fixed in place.
In flexible design either the sampling methodology or the statistical methods used for analysis may be subject to change. One example of a flexible design would be a drug trial in which the sample size is reassessed after the initial data comes in.
Benefits of Flexible Design
Flexible types of design are useful because they allow us to target our experiments more closely, and reduce the waste of time and resources that a blind trial involves.
Flexible design principles have allowed medical testers, for instance, to condense a planned series of two separate trial experiments into just one; An interim analysis halfway through targets and fine-hones the data. In a field where each trial can take years of planning, involve reams of official approvals, and can be very expensive, this way of cutting corners can have huge benefits. It means we may get answers much sooner to big questions involving as yet untreatable illness or new cutting edge drugs.
Negative Points and Cautions for Flexible Design
Flexible designs have the downside, though, that they introduce added possibilities for selection bias and other experimenter-based inaccuracies. They should always be used carefully, and the full rationale behind all changes in design over the course of the trial or experiment should be fully recorded and explained in research reports.
It’s also important that if changes are made halfway through an experiment a plan is made—before implementation of those changes— on how to combine the data from different stages.
If these issues are not addressed, an experiment run with flexible design will not be able to satisfactorily confirm or reject any hypothesis.
Chow SC (2012) Flexible, Adaptive or Attractive Clinical Trial Design. Drug Des 1:e104. doi:10.4172/2169-0138.1000e104
Retrieved from https://www.omicsonline.org/open-access/flexible-adaptive-or-attractive-clinical-trial-design-ddo.1000e104.php?aid=7347 on January 12, 2018
EU Committee for Medicinal Products for Human Use, Reflection Paper on Methodological Issues in Confirmatory Clinical Trials with Flexible Design and Analysis Plan. Retrieved from http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/09/WC500003617.pdf on Jan 13, 2018
Hung, Want, O’Neill, Flexible design clinical trial methodology in regulatory applications
Retrieved from http://onlinelibrary.wiley.com/doi/10.1002/sim.4021/full
Roser, Christoph, and David O Kazmer. “Flexible Design Methodology.” In Proceedings of the 5th Design for Manufacturing Conference Design Engineering Technical Conferences, 3:139–48. Baltimore, Maryland, USA: ASME, 2000.
Preprint retrieved from http://www.allaboutlean.com/wp-content/uploads/2014/05/2000_Roser_DETC-Flexible-Design-Methodology_preprint.pdf on January 12, 2018
If you prefer an online interactive environment to learn R and statistics, this free R Tutorial by Datacamp is a great way to get started. If you're are somewhat comfortable with R and are interested in going deeper into Statistics, try this Statistics with R track.Comments? Need to post a correction? Please post on our Facebook page.