What Works in Parenting Support: Glossary
Policy Research Bureau

* What Works in Parenting Support? Main report.
* What Works in Parenting Support? Summary report.

* Search the database

* Return to PRB Home Page

Abbreviations and Acronyms
Mean (Average)
Not Applicable
No Information reported
Method of Administration
Randomised Controlled (or Control) Trial
Special populations. Denotes particular populations of policy interest for the Review. Includes: fathers, foster parents, grandparents, step-parents, low income parents, ethnic minorities, refugees, asylum seekers, travellers, gifted children, children in transitions, children and parents with disabilities, teenage parents, parenting teenagers, children with SEN, parents in prison, drug using parents
Time 1. A time point for data collection - the beginning of an intervention
Time 2. Time point for data collection - the end of the intervention
Time 3. Time point for data collection - follow up after end of intervention

Glossary of Key Terms
Early intervention
Services that are aimed at families where early signs of problems may be visible at a low level, provided with the intention of nipping difficulties in the bud and preventing their evolution into bigger problems
Effect/Effect size
The effect size indicates the underlying strength of the relationship between the independent variable i.e. the factor that researcher is systematically varying (e.g. parenting skills training), and the dependent variable i.e. the factor that the researcher is expecting changes in (e.g. child behaviour). While significance focuses on the probability of the relationship between the variables occurring by chance, effect size focuses on the magnitude of the effect of one variable on another. It therefore provides an estimate of the extent of effect of an independent variable on a dependent variable
Group work
Services that are delivered to groups of parents at the same time, either in a classroom style setting or more often in a workshop or informal discussion-based format
Individual work (also called one-to-one work)
Services that involve a worker (professional or voluntary) and a single individual recipient
Intensity and duration of interventions (corresponds to the medical concept of 'dosage')
Low/short six weeks or less, in sessional format (one off or regular)
Medium six to twelve weeks, in regular sessional format
High/long twelve to twenty six weeks, in regular sessional format
Intensive over twenty six weeks; or continuous for some sustained period (e.g. 'wraparound' and residential services)
Manualised intervention
The intervention has a written manual or protocol describing the objectives of the intervention, the 'theory of change' that underlies the intervention, what precisely should be delivered in terms of content (ie, a curriculum), and giving guidance to the practitioner about how to structure and lead sessions.
Unlike experimental designs, non-experimental designs do not provide evidence of causality, although some non-experimental designs may be strongly suggestive of causal effects. Non-experimental approaches such as correlational designs, for example, can show associations between two variables that may imply a causal relationship. However, whereas experimental designs are able to demonstrate that systematically changes one variable (e.g. parenting skills) leads to change in another variable (e.g. child behaviour), the same cannot be concluded from non-experimental studies.
Often used in the US literature to denote workers who have a basic level of training in a given area, but who are not highly qualified
Peer worker (or peer supporter)
Used to describe parents who are involved in delivering an intervention (paid or unpaid), often offering informal befriending and support on the basis that they share the same social circumstances and have had similar experiences
Primary prevention
Services that are aimed at addressing a need that may not yet have become apparent, or at preventing a problem from developing in the first place
A worker with specialised training and qualifications in a particular area.
Quasi-experiments do not use random assignment of participants to treatment and control groups, but do involve an experimental approach in which the effect of one variable on another is assessed. However, lack of randomisation of participants means that inferences about the causal relationship between factors are more ambiguous. Quasi-experimental designs vary in terms of the number of experimental groups involved and the number of time points at which data are collected to assess particaptns' progress. They include: a one group post test only design, a one group pre test & post test design, a non-randomised post-test only design, and non-randomised groups pre test & post test design.
Randomised studies (also known as random allocation; randomised controlled trials; RCTs)
Randomised studies are experimental studies characterised by the random assignment of participants to experimental conditions, such as an intervention (or 'treatment') group and a control group. They enable the researcher to assess casual effects by examining the effects of one factor (the 'independent' variable) on another factor or outcome (the 'dependent' variable). The randomised allocation of participants to treatment groups reduces the probability of the results being due to prior differences in characteristics of the groups, and increases the likelihood of the results being due to the intervention.
Secondary prevention (services for 'at risk' groups)
Services that are aimed at families who are deemed especially at risk for a particular problem - for example, services for parents who are experiencing problems with parenting that could if left unattended lead to child abuse.
A test of significance helps to determine whether results are 'genuine' or not. Typically the cut off point for deciding on the significance of a result is set at the 5%level, or one in twenty probability level, expressed as p<.05. This means that there is a 5% probability of the observed results having occurred by chance, and a 95% probability that the results indicate the presence of a 'genuine' or 'real' effect. If the significance level of a test shows that the probability is greater than 1 in 20 (i.e. p>.05), then the results are said to be non-significant. In practice this means that any difference between, for example, a group receiving an intervention and a control group, could have arisen by chance rather than resulting from the effects of the intervention.
Single mode and multi-mode interventions
Single mode interventions involve a single activity (e.g. prescribing medication for ADHD); multi-modal interventions offer combinations of activities usually aimed at complementing each other (e.g., medication and a programme of education about ADHD).
Targeted services
A service that is provided to only some groups or individuals on the basis of an assessment of their particular need (e.g. social work; speech therapy; services for parents of children with conduct disorders).
Tertiary prevention / Treatment
Services provided for families who have identified problems in an area of functioning and who need help to regain adequate or 'normal' levels of functioning.
Universal services
A service that is provided to everyone, irrespective of levels of need (in the UK, examples would include primary health care services (GPs, the Health Visiting service, and ante-natal classes)
Validated (and unvalidated)
Generally only used in the context of quantitative methods, and with similar meaning to the term 'standardised'. For the purposes of the Review, we defined a validated measure (or 'instrument') as a scale or questionnaire that had been pre-tested for statistical reliability (i.e., stability over time) and for face and construct validity (i.e. that the instrument actually does measure what it purports to measure), and found to be reasonably robust. Often this would involve testing and re-testing the instrument on a 'normal' sample of the population of interest, so that the values to be expected in a representative group of the population of interest are known and 'norms' can be established against which study results involving samples can be assessed. Unvalidated measures on the other hand are usually tailor-made for particular research studies and are of unknown (or unreported) reliability and validity.

For enquiries about The POLICY RESEARCH BUREAU's past work,
email dghate@prb.org.uk.
(c) 2004 Policy Research Bureau and Arches Consulting Ltd
$Id: glossary.html,v 1.3 2007/03/30 12:05:06 tom Exp $