Broadly speaking, there are two ways of collecting statistics. 

1. One is to use data collected for another purpose – for example, figures for recorded crime, for prosecutions or numbers of court orders applied for and made. 
   
   
2. The other way is to ask people directly about the issue – for example, their experiences of certain kinds of crimes. 

Data that has been collected for a different purpose – often by a government agency or other official body – may be seen in some respects as more “objective”; however, they are subject to whatever classification criteria that body is applying. 

For example, the process by which an incident of domestic abuse becomes recorded as a “crime” involves several stages, starting with the decision by the victim whether or not to report an incident, followed by the responses of the police (e.g. whether or not to record the incident), the Crown Prosecution Service, the courts, and so on. 

Asking people about their experiences may result in information that is better suited to a particular purpose – but is also subject to various limitations: 

• It is more time consuming – and hence more expensive. 

• What questions are asked and how they are asked lead to different responses – some being more reliable and more meaningful than others. 

• Results will be based on a sample of the population rather than everyone and will therefore be subject to some degree of sampling error. 

• Sometimes the sample will be based in one geographical area, or include a particular kind of person only, and will not therefore be universally applicable.

When looking at any statistics, there are two issues to consider: 

1. Reliability – i.e. are these results replicable? – would another person or organisation asking exactly the same questions, or collecting the same kinds of information, come up with the same figures?
   
   
2. Validity – i.e. are they meaningful? 
   
   

Factors which can affect both reliability and validity include: 

• The definition of domestic violence and/or abuse which is used. 

• How information is collected who or what. 

• How many people or organisations are included who is collecting the information, and for what purpose. 

Most domestic abuse statistics (e.g. crime figures) are based on specific incidents and kinds of behaviour. 

However, domestic abuse has to be seen within a context of power and control, which is usually gender-based. 

Repeated and escalating abuse which takes place within a context of fear and intimidation does not easily show up in an incident-based form of statistical record; and emotional abuse – which may be perpetrated in various ways, and with various degrees of subtlety – may be completely disregarded, particularly when the focus is on crimes. 

Information about people’s experiences of domestic abuse may be collected in different ways: 

• By written questionnaire which may be posted or emailed to a sample of potential respondents. This procedure is rarely used in domestic abuse research and is likely to result in a low response rate, under-reporting, and low validity and reliability. 

• By written questionnaire which may be accessed via the internet, or a magazine, newspaper or other publication. In this case, respondents are self-chosen, and there is no way of judging whether or not their experiences are typical of the population as a whole. 

• By asking in a telephone interview. Privacy cannot be assured, and it can be dangerous for respondents (who may therefore tend not to disclose domestic abuse.) 

• By asking them face to face, as in the Crime Survey for England and Wales. Both validity and reliability are improved, as is the response rate; but under-reporting is common. 

• Self-completion, after contact has been made face to face, and privacy assured. This is the preferable method, as it maximises both the respondent’s safety, and the reliability of the data: many victims of domestic abuse are reluctant to disclose the abuse they have experienced in face-to-face interviews.