Challenges and solutions for N-of-1 design studies in health psychology

ABSTRACT

Theories of behaviour change and health behaviour change interventions are most often evaluated in between-person designs. However, behaviour change theories apply to individuals not groups and behavioural interventions ultimately aim to achieve within-person rather than between-group change. Within-person methodology, such as N-of-1 (also known as single case design), can circumvent this issue, though has multiple design-specific challenges. This paper provides a conceptual review of the challenges and potential solutions for undertaking N-of-1 studies in health psychology. Key challenges identified include participant adherence to within-person protocols, carry-over and slow onset effects, suitability of behaviour change techniques for evaluation in N-of-1 experimental studies, optimal allocation sequencing and blinding, calculating power/sample size, and choosing the most suitable analysis approach. Key solutions include involving users in study design, employing recent technologies for unobtrusive data collection and problem solving by design. Within-person designs share common methodological requirements with conventional between-person designs but require specific methodological considerations. N-of-1 evaluation designs are appropriate for many though not all types of interventions. A greater understanding of patterns of behaviours and factors influencing behaviour change at the within-person level is required to progress health psychology into a precision science.

Video abstract: Supplementary Material 1.

KEYWORDS:

• N-of-1
• single case study
• within-person design
• idiographic design

Acknowledgement

This conceptual review was written as a result of an N-of-1 design workshop prior to 31st European Health Psychology Society (EHPS) Conference in Padua, Italy, 2017. We would like to acknowledge the EHPS for the opportunity to meet and discuss N-of-1 design. We would like to thank all meeting participants who contributed to the discussions that instigated this paper. The authors would also like to acknowledge the helpful statistical advice from Dr George Savva and Dr Kathryn Richardson.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Dominika Kwasnicka http://orcid.org/0000-0002-5961-837X

Jennifer Inauen http://orcid.org/0000-0002-7884-3222

Wim Nieuwenboom http://orcid.org/0000-0002-4159-6897

Johanna Nurmi http://orcid.org/0000-0001-8414-3444

Annegret Schneider http://orcid.org/0000-0003-4583-7771

Camille E. Short http://orcid.org/0000-0002-4177-4251

Tessa Dekkers http://orcid.org/0000-0003-2787-3293

A. Jess Williams http://orcid.org/0000-0002-3987-3824

Walter Bierbauer http://orcid.org/0000-0003-2745-4865

Ari Haukkala http://orcid.org/0000-0001-8567-1548

Federica Picariello http://orcid.org/0000-0002-2532-3290

Felix Naughton http://orcid.org/0000-0001-9790-2796