The Best Design is Not Enough
The value of randomized controlled trials (RCTs) resides in comparisons of an “experimental” treatment with a controlled condition and in the effectiveness of a randomization strategy that essentially nullifies or minimizes unmeasured factors among participants that could adversely affect results.
But in many areas where much further study is needed, it is very difficult to create a meaningful “control” condition. For example, after a loved one becomes severely physically disabled, there is no routine “standard of care” for family caregivers in the community. Consequently, any attention or service provided to these caregivers will be above and beyond what is typically associated with a “controlled” condition.
Caregivers in many communities have everyday routines which put considerable demands on their time. As a consequence, they may be prone to dropping out of a control condition that appears to offer little benefit or have little potential value.
Even under the best of circumstances, randomization may not work well in small samples. This may be true in understudied caregiver scenarios that involve high-cost yet low-incidence conditions, such as spinal cord injury (Ramkumar & Elliott, 2010). Some unmeasured characteristics may be over-represented in one group or may exert an effect over time in other, unanticipated ways. For example, caregivers with a lower income may be more likely to drop out of a study than those with a higher income.
There are many other “unmeasured” characteristics that influence the degree to which a caregiver is willing to participate in an RCT. These characteristics may be more likely to exert an influence on the results with smaller sample sizes.
Ramkumar, N., & Elliot, T. (2010). Family caregiving of persons following neurotrauma: Issues in research, service and policy. NeuroRehabilitation, 27, 105-112.