Abstract
I use this case quite frequently when I am teaching research ethics, as it is a really useful way of drawing out tensions between consequentialist and deontological approaches to ethics. Although the initial reaction to the research is typically horrified dismay, I think it benefits from some careful thinking about whether this research is actually as bad as it appears. In summary, the researchers intended to study the safety of fridge door design by trying various types of doors − luring small children into boxes with the door design and then closing it, and observing their behaviour on closed circuit television.
Firstly it is worth noting that the researchers are certainly cognisant of the potential for harm to the children participating and do take precautions both to minimize this (opening the door immediately if they show signs of distress) and to detect lasting damage (following up after six months and a year). So they are not callous or disregarding of the potential costs to participants.
Secondly it is worth noting the real world impacts of this research − it was part of the evidence which underwrote the adoption of the modern magnetic seal fridge door as opposed to the modification suggested by manufacturers at the time, namely the much cheaper introduction of a catch on the inside that children could use to release the door internally. This, along with other activities, arguably contributed to a decrease in the death rate of children stuck in fridges and freezers from about 2 per million in 1960 to less than 0.5 per million in 1981 in the US 1 (Kraus, 1985). That is a significant decrease in deaths, particularly given how influential the US design was internationally; many, many lives have been saved by this research.
Nonetheless there are a number of significant ethical challenges that need to be considered:
Is it necessary? One question that Research Ethics Committees tend to ask when examining this and other risky research proposals is whether the risk is needed to achieve the end results. In this case, for example, is the deception necessary − couldn’t we just ask the children to try to escape the box? In this case the researchers argued that it was necessary to observe the children in a natural environment. This claim seems to have been supported by the experimental results, in particular the surprising outcome that 24 per cent of children made no effort whatsoever towards escaping and 37 per cent of children reacted with violent but non-purposive behaviour (Bain et al., 1958). This underwrote the design of fridges such that even very minimal movement such as sitting down would break the seal. And if children had been asked to try and escape this would not have been observed.
Alternatively we might query whether the risks and harms involved in the research are justified. There are four distinct questions we may ask here: (i) Are the risks/harms in general justifiable? (ii) Are the risks/harms outweighed by the benefits? (iii) Are the risks/harms outweighed by the benefits to the participants? (iv) Are the risks/harms outweighed by the benefits to those like the participants?
With regard to question (i), the risks here primarily are those of distress; although there is a small chance of lasting psychological damage (for example, claustrophobia) this is minimized by the research design. This is the sort of scenario that children sometimes get themselves into (my son has, for example, opened the under-stairs cupboard from the outside, gone in and been unable to open it from the inside) and typically does not do lasting damage. As such they seem in general justifiable. In regard to question (ii), it likewise seems that the potential benefits of the research (reducing the death rate from 2 per million) outweigh the possible psychological distress and harm to 200 participants. Question (iii) is where our consequentialists and deontologists will diverge, with consequentialists disagreeing that this is important. It seems unlikely that this research will benefit the participants – with only two deaths per million being potentially avoided, it is very unlikely that any of our participants will be among those benefited by the research. Under some frameworks this makes the research unacceptable, because it poses more than minimal risks to vulnerable participants without a corresponding benefit to those participants. I am sceptical about the importance of this as an absolute rule myself, however, given that the harms are mostly small and transient in this case, with a very low risk of more serious damage, but the harms being mitigated by the research are significant and permanent. One way to resolve this is to appeal to question (iv), which focuses on the type of participant rather than the participants themselves, and asks whether benefits accruing to the type of participants outweigh the harms to the actual participants. The answer to this question is clearly Yes − the small risk of psychological harm and a certain amount of distress for 200 children is outweighed by the potential lives saved among children. I am likewise sceptical about this solution, mostly because I am not sure that I always identify in a morally relevant way with groups that I belong to. Nonetheless, in this case I think the benefits so outweigh the harms that I am tempted to think that it does not matter that they do not accrue to those participants.