The main features of a randomised controlled trial are the randomisation process and the presence of a control group.

Randomisation "“ The use of this technique should aim to produce a representative sample, typical of the population under investigation (e.g. consumers of a specific service). Therefore, care must be taken in the method of randomisation employed, with 'pseudo' methods of randomisation avoided, such as clients attending a service on alternative days, because bias can arise if such an approach to deciding which group (arm) of a trial participants are allocated is used (e.g. it is possible to modify the allocation process to ensure that a particular individual receives a specific intervention). Biased sampling occurs when the sample chosen fails to represent the true make up of the overall population of interest. For example, if service users are asked to volunteer in a study they may be more enthusiastic and motivated compared to those who declined and, consequently, unrepresentative. A biased sample relates to those in which some individuals had a greater or lesser chance of being included. Random sampling should overcome such bias, by ensuring that every individual within the target population has the same chance of being chosen to take part.

It is assumed that through randomisation all extraneous variables will be evenly distributed across different groups. Therefore, the rationale behind randomisation is to attempt to produce equal groups in terms of participants. The use of randomisation differentiates experiments and quasi-experiments (see below).

Use of a control group"“ Individuals in control groups will either not receive the intervention under investigation, or will receive something else, often 'standard care' applied to the specific problem or situation. The control group used should be as closely matched as possible to the experimental group to avoid any confounding factors from interfering with results produced, such as age, gender, social class.

People within the control group will not have received the intervention under investigation (the 'independent variable'). This allows for a comparison between control and experimental group. Since the former did not receive the independent variable any differences exhibited between the two should relate to the effect of the independent variable, as long as the groups are of a similar makeup at the outset. For example, in a Children and Families Service, those in the control group may receive the standard service, while those in the experimental group receive a specialist, more intensive intervention (the 'independent variable').

Allocation to Treatment Group - There are three principal means by which an individual may be assigned to an experimental or control group during a trial: independent sample design, matched pair design, or repeated measures design.

Independent sample design

This design involves a group of participants being obtained for the experiment as a whole, and then individuals are allocated randomly to one or other of the experimental conditions

Matched pairs design

When using this design, participants are usually matched into pairs and then allocated randomly to either a control or experimental condition. A researcher may decide to carry out this form of design if they feel there is a third variable that they suspect could affect the dependent variable. For example, if gender was thought to affect the outcomes measured, the research would match people according to their gender and then randomise participants to an experimental or control group in relation to these paired groups (one of the pair going to the experimental group, the other to the control). Such an approach helps to ensure that this potentially moderating variable is controlled, by being equally distributed within each group. It is therefore less likely to interfere with the relationship between the independent and dependent variables.

Repeated measures (or within subjects) design

Participants in this approach under go both sets of conditions under investigation, i.e. all participants are involved in all conditions. The problem with this design is that of 'order effects', i.e., the results are influenced by the sequence in which conditions are administered. A learning effect may ensue when a participant is involved with two conditions, so that whatever is carried out second tends to receive a higher score, which cannot be directly attributed to the independent variable. Alternatively, a 'fatigue effect' may occur, giving a negative overall result, so that whatever is carried out second will tend to receive lower marks. To overcome this difficulty, 'counterbalancing' may be adopted, so that half of the participants undergo condition A followed by condition B, and the other half are involved with condition B followed by condition A.