Experiences of victims of fraud and cyber crime: technical report

Question 1

1.  Introduction

Accepted Answer

1.1 Background

In September 2018, the Home Office commissioned Ipsos to conduct research to develop the evidence base regarding harms from fraud and cyber crime. The Home Office had 3 objectives for the research:

To understand the extent and nature of harms experienced from fraud and cyber crime.
To understand victim perceptions of the support provided by the National Economic Crime Victim Care Unit and inform any improvements to the service.
To understand wider perceptions of support and advice needed to protect against fraud and cyber crime.

The research was split into a qualitative phase, consisting of depth interviews with victims and quantitative phase, consisting of 2 large-scale telephone (CATI) surveys. The first survey included a sample of victims of fraud or cyber crime who had reported their crime to Action Fraud between July 2017 and May 2018 and were in receipt of National Economic Crime Victim Care Unit (NECVCU) services. The second survey included a sample of victims of fraud or cyber crime who reported their crime to Action Fraud between July 2017 and May 2018 but did not receive any NECVCU service. All participants were aged 18 and over, residing in England and Wales. Fieldwork was conducted between February 2019 and May 2019.

It should be noted that since the research was conducted, findings were fed in to help improve NECVCU service delivery and NECVCU has been gradually rolling out amongst police force areas. Since this time, changes in delivery and terminology used in the NECVCU service may have changed. Details provided in this report were correct as at the time of the research. Further background is outlined in the ‘Experiences of victims of fraud and cyber crime’ report that sets out the keys findings from this research.

Action Fraud and NECVCU

Since 2013 Action Fraud has been the national reporting centre for fraud and cyber crime. The service is run by City of London Police and works alongside the National Fraud Intelligence Bureau (NFIB) who are responsible for assessing crime reports and determining whether reports contain intelligence value and are viable for investigation. A part of Action Fraud’s role is to provide help and support to victims.

At the time of the research, the Action Fraud National Economic Crime Victim Care Unit (NECVCU) was a proof of concept model where a team of specialist advocates working within the City of London Police (CoLP) delivered tailored services to primarily vulnerable victims of fraud and cyber crime. The NECVCU was established to help fill a gap in service provision from the Victims Code of Practice (VCOP) and improve the ‘Prepare’ response to victims of fraud and cyber crime. The principles of the NECVCU model are:

to make people feel safer
enable them to cope and recover from fraud and cyber crime victimisation
prevent them from becoming a repeat victim of fraud or cyber crime
to improve trust and confidence in Action Fraud/the police.

After an initial pilot in London beginning in 2014, in July 2017, the Home Office committed to funding an extension to the NECVCU pilot to 2 police forces outside of London (West Midlands and Greater Manchester) to look at its applicability for wider roll out. Two further forces joined the NECVCU pilot after the survey started but did not participate in the survey.

The NECVCU model supports individuals who report a fraud or cyber crime to Action Fraud but may not receive any focussed service or tailored support from services, including the police, because their cases are not disseminated to forces for further investigation.

The research aimed to evaluate victims’ satisfaction with different elements of the service and support provided by NECVCU, Action Fraud, the police and other support services. Beyond this, it captures specific information about the harms victims experienced and circumstances that might have affected their level of satisfaction. Such information will help to identify the key drivers of satisfaction with the NECVCU service and support services more widely and thus inform strategies to improve services.

A data protection impact assessment (DPIA) was completed by the Home Office in collaboration with City of London Police (CoLP) and Ipsos. A data processor agreement (DPA) was then produced and signed by CoLP and Ipsos. Contact details for victims were provided to Ipsos by the CoLP which cover England and Wales. Victims were then contacted by Ipsos by post and given an opportunity to opt-out of the research. Those who did not opt-out were then contacted for interview by telephone.

This report describes the methodology and processes employed to conduct the quantitative and qualitative elements of the project. This report does not contain findings from the research.

Question 2

2.  Research materials

Accepted Answer

2.1 Survey development

The following table provides a broad overview of the themes covered in the survey questionnaires which were developed in conjunction with the Home Office and City of London Police (CoLP). The final survey questionnaire is provided at Appendices C and D in the Victim survey technical report - appendices.

Table 2.1: Overview of survey questionnaires

Section	Description	Group
Introduction to the survey	This section covered the aims of the research, how participants could access the privacy notice and was the point at which the interviewers gained consent from participants to take part in the interview. As the survey is about the participant’s experiences of NECVCU and Action Fraud, if participants did not remember receiving advice or support from NECVCU or Action Fraud either on behalf of themselves or someone else, the interview was terminated at this point.	All participants
Details of the crime and experience	The section included an open text question asking for participants to briefly describe the crime they were a victim of. Participants are then asked whether/how often they have been a victim of cyber crime, any types of fraud or any other traditional crime.	All participants
Harms	This section covered how likely participants thought they were to be a victim of crime and how much money, if any, was taken and retrieved, as well as other financial impacts such as having to change bank accounts and borrowing money. Types of non-financial impacts were explored as well as a range of health and emotional impacts and the duration of these impacts. This section also explored the overall effect of the crime at the time of the crime and at the time of the survey.	All participants
NECVCU support	This section focused on the participant’s experience of the NECVCU service, including when and how participants reported the crime to Action Fraud and details about their contact from NECVCU /Action Fraud/City of London Police and their satisfaction with this service. In this section participants were also asked about which agencies they had contacted on the advice of NECVCU, for example, Age UK Victim Support, Trading Standards, Local Authorities, MIND and Samaritans.	NECVCU victims questionnaire only
Non-NECVCU support	This section covered non-NECVCU support and the events which took place after the crime was reported, how the crime was reported, the communication type between the participant and Action Fraud and what information they received from Action Fraud. If participants recalled receiving a letter/letters from Action Fraud after reporting the crime, there were follow up questions about this support, and how useful it was. A ‘police only’ section was included for victims who had contact with the police or their case was taken forward for police investigation with follow up questions on satisfaction, type of contact etc. If a participant did not have contact with the police, a non-police support section asked about other types of support received outside of Action Fraud or the police.	Action Fraud victims questionnaire only
Re-victimisation and behaviour change	In the NECVCU questionnaire, this section covered behaviours which participants may have started as a result of the advice received from NECVCU and whether they have been re-victimised.	NECVCU victims questionnaire only
Awareness of organisations	In the Action Fraud victims questionnaire this section covered which types of organisations participants were aware of prior to the crime and after the crime, for example, Cyber Aware.	Action Fraud victims questionnaire only
Future support needs	This section asked which organisations participants would contact if they were to be victims of the same crime again. If participants would not contact Action Fraud or any other support organisation/agency they were asked the reason(s) for this. The response codes were tailored between the 2 questionnaires in some parts of this section based on the advice given to victims for different services, for example, NECVCU mentions the Samaritans but the Action Fraud victims survey did not.	All participants with slight variation in response codes
Vulnerability	Participants were asked how vulnerable they felt at the time of the crime and how vulnerable they felt at the time of the interview. As a sensitive question, interviewers were advised to exercise caution.	All participants
Socio-demographic information	This included standard socio-demographic information such as work status, gender, chief income earner, ethnicity and their postcode. Postcode was asked in order to analyse answers with people in the same area and compare them to other parts of the country.	All participants

2.2 Question types

The majority of questions were pre-coded. For pre-coded questions, a list of pre-defined responses to the question, called the ‘code frame’, were displayed for interviewers to select from. Some pre-coded questions allowed only one response, where answer options were mutually exclusive, whereas others allowed multiple responses, where more than one answer could logically be provided to the question.

Pre-coded questions sometimes included an ‘other-specify’ option, which allowed participants to provide answers other than those included in the pre-coded lists. There were 14 ‘other specify’ options in the NECVCU survey and 12 in the Action Fraud victims survey. If participants gave an answer that was not in the pre-defined list, interviewers typed in their verbatim response. ‘Other specify’ responses were analysed and added to existing codes where possible. Where there were a sufficient number of responses to create a new code, this was added to the data tables.

The survey also included one open-ended question in each of the surveys, where the script did not attempt to anticipate answers which participants might give. Instead, interviewers typed participants’ verbatim answers into the CATI system.

2.3 Questionnaire review

Following the drafting of the questionnaires by Ipsos and the Home Office (with input from key stakeholders from CoLP), the questionnaires were cognitively tested with 12 fraud and cyber crime victims in London between 10 to 20 December 2018. The sample of participants were split as follows across range of demographics including age, gender and crime type:

7 who received NECVCU support
5 who did not receive NECVCU support

As a result of the cognitive testing, several changes were made to the questionnaires including:

The initial question (Question 1 in both NECVCU and Action Fraud victims surveys), which asked participants to describe the nature of the crime they were a victim of and reported to Action Fraud, was changed to an open response for the interviewer to type the answer.
Following feedback from 4 interviews, a preamble was added prior to asking participants about the financial impacts of the crime (Question 4a in both surveys), as participants during the cognitive testing felt this was a sensitive question for the beginning of the questionnaire. The preamble was tested at later rounds of cognitive testing and the remaining 8 participants did not raise any issues with the wording even when probed.
An answer code at Question 5 (which asked participants to consider the emotional impacts of the crime) was shortened from ‘violent thoughts and anger’ to simply ‘anger’ as participants felt it was possible to experience one without the other.
Answer codes at Question 18 and Question 27 in the NECVCU survey were shortened and instructions were added to interviewers to read out definitions of types of support if necessary. This was because participants during the cognitive interviews found the long series of read outs frustrating.
In the Action Fraud victims survey, an additional question was added prior to Question 26 (Question 26a) to establish whether or not participants contacted anyone else for support or information other than their contact with Action Fraud. Only participants who said ‘yes’ were routed to Question 26b, which listed the different organisations. This shortened the questionnaire and avoided participants who had not sought advice elsewhere having to answer Question 26b.

2.4 Questionnaire amendments

Amendments were also made after the questionnaire had been in field after feedback from the telephone interviewers and a higher than anticipated interview length. The surveys were in field for 4 days with 87 Action Fraud completed and 82 NECVCU victim surveys completed. For example, interviewers felt the financial impact questions on how much money was taken from the victims were still too sudden and felt participants would be more open if they had more time to warm up. Further preamble was therefore included to remind participants that answering questions was optional and that the information collected would be kept in the strictest confidence and to let the interviewer know if they wanted to stop at any time.

Amendments were also made to the initial screener question on the NECVCU questionnaire asking whether people recalled receiving support from the NECVCU. The initial stages of fieldwork showed a high drop out at this stage (question S2) so an additional prompt was added to provide further clarity that we wanted to speak to all victims who had reported to Action Fraud. The changes are summarised below, with the final questionnaire provided at Appendices C and D in the Victim survey technical report - appendices.

Original text and prompts at S2

S2. Do you remember receiving advice, information or support from the Economic Crime Victim Care Unit? This was likely through a phone call but may have been via an email or letter.

ECVCU is a support and advice service for victims of fraud and cyber crime
this may have been up to 18 months ago
if you would like to verify the research, please do contact ECVCU on 0207 164 8222 which will be manned 7:30am to 6pm or visit Action Fraud

Additional prompt added

An additional prompt was added to the above: ‘We are interested in speaking to anyone who reported they were a victim of fraud or cyber crime’.

In addition, both questionnaires were shortened after the survey was in field in order to reduce the average interview length and increase response rates. After 4 days of fieldwork the NECVCU survey was running at an average length of 35 minutes versus an intended length of 20 minutes. The Action Fraud victim survey was running at an average length of 36 minutes compared with an intended length of 25 minutes. These changes were made based on the feedback received from the interviewing team about which questions participants were taking longest on, as well as time stamp data received from the survey team. These cuts included:

in the NECVCU survey, removing an initial question (Question 3) asking about how likely participants saw themselves to be a victim of fraud or cyber crime
in the NECVCU survey, removing a question about how participants were first contacted by NECVCU and an open follow up question asking them to remember the nature of the call
in both surveys, removing the answer code option of ‘please specify’ at questions Question 4c, Question 5 and Question 6
in both surveys, removing several grid questions at Question 24a (in the NECVCU survey) and Question 29a (in the Action Fraud victim survey) which asked about what steps participants take to protect themselves from fraud and cyber crime; the options removed were: back up data on all devices; use password/passcode/PIN to unlock smart phones or tablets; share personal information on social media and use credit cards for purchases over £100
in both surveys, removed demographic questions about living arrangements, household occupants and participants’ self-assessed health

2.5 Qualitative materials

The discussion guides used for the depth interviews were developed in conjunction with the Home Office and were broken down into the following sections:

1. Introduction: Introduced the research and explained anonymity, confidentiality, disclosure policy, how the data will be used, for instance. Discussed general work and life circumstances to provide useful background and establish rapport.

2. The crime and support journey mapping: Interviewees talked through their victim journey which was mapped by the interviewer and checked for accuracy. This section involved a discussion of the following:

the crime and what happened
their reaction/who they contacted
when they contacted Action Fraud
when NECVCU responded and type of support received
pre-victim journey and vulnerability

3. Harms and impact of the crime: Explored in detail the impact of the crime on participant’s life including long and short-term harms:

financial impacts
emotional impacts
health impacts
impacts/wider consequences on their life

4. Support from NECVCU and satisfaction with the service

explored expectations of the service
level and type of support received from NECVCU/police (if Tier 3)
how helpful advice and support was
what other advice or support they felt they needed/wanted
satisfaction with service, timeliness, for instance, and drivers of opinion

5. Behaviour change

what support/advice given by NECVCU participants followed up on and reasons
altered behaviour since NECVCU service
likelihood of being a victim again

Interviewees were given £50 incentive in cash for their time and a support leaflet which included contact details for support charities such as Victim Support as well as signposting to information including the Citizen’s Advice Bureau, Financial Ombudsman, Fraud Advisory Panel and Get Safe Online.

Question 3

3.  Sampling

Accepted Answer

3.1 Quantitative surveys

There were 3 key objectives of the surveys which were dependent on the sampling stage:

Achieve a separate representative sample of victims of fraud and cyber crime in receipt of NECVCU services.
Achieve a separate representative sample of victims of fraud and cyber crime who had reported to Action Fraud but had not received an NECVCU service.
Where possible, ensure sufficient sample was drawn for certain sub-groups to enable analysis at a later stage. These sub-groups included age, gender, crime type, vulnerability and Tier. Larger sample sizes increase the likelihood of being able to identify statistically significant differences within data and as sample size increases, the margin of error decreases. The sampling approaches to both surveys are included in more detail in the following pages.

There were a number of criteria dictating which types of victim, were eligible for inclusion in each survey. The NECVCU survey included cases of victims across London, Greater Manchester and West Midlands who had been in receipt of NECVCU services. The Action Fraud victim survey included a sample of victims in England and Wales of fraud or cyber crime who reported their crime to Action Fraud between July 2017 and May 2018. This was the largest time frame of available data at the time of sampling. The NECVCU sample was made up of Tier 1, Tier 2 and Tier 3 victims^{[footnote 1]}. A brief summary of how Tiers are defined is included below. To note, this report refers to these different forms of support as ‘Tiers’, but since the research was undertaken they are more commonly called ‘Levels’.

Tier 1 - is for victims who do not identify themselves as vulnerable at the reporting stage and whose cases have not been disseminated to the police for investigation. They receive a call from a victim contact advisor (VCA) at the Action Fraud call centre offering guidance and signposting.

Tier 2 - this is the main element of the NECVCU for vulnerable victims and is provided by a dedicated unit set up within the CoLP staffed with victim care reviewers (VCR). Tier 2 is for victims who self-identify as vulnerable at the reporting stage and who, following a further manual vulnerability triage assessment (which took place centrally at CoLP), are deemed eligible for the Tier 2 service. They provide enhanced support and guidance to vulnerable victims and make further assessments about their vulnerability to ensure they receive appropriate victim care.

Tier 3^{[footnote 2]} - at the time of the survey, this element was sitting only within local police forces at West Midlands and Greater Manchester. The forces receive referrals from the Tier 1 and Tier 2 advisors when they deem that face-to-face support is required in order to provide appropriate support to the victim. This may be an immediate call for service or a visit within a 48 to 72 hour period. The set-up of this response and who delivers the support varies between forces.

3.2 Sample cleaning

All sample files were originally provided by CoLP in an Excel format with the files split out into single files by month that the victims reported. This data was extracted from their ‘Know Fraud’^{[footnote 3]} case management system with files split out by ‘vulnerable’ and ‘non-vulnerable’ cases. In total 11, ‘non-vulnerable’ files were sent and 11 ‘vulnerable’ files were sent. Victims in receipt of NECVCU services were able be assigned a Tier based on their vulnerability (either self-assessed or by a member of the NECVCU team) - for Action Fraud victims the ‘vulnerable/not-vulnerable’ flag in the original files was used as the most suitable proxy for Tier. The first stage of processing the sample files was to combine the month data to produce 2 files, one for victims listed as vulnerable and one for non-vulnerable^{[footnote 4]}. This was then followed by further splitting out of files between NECVCU pilot and non-pilot forces, disseminated and non-disseminated cases to produce an NECVCU and Action Fraud (that is, non-NECVCU) database. This process is discussed in more detail and illustrated in the diagram on the following page:

Figure 3.1: Sample cleaning steps diagram

The NECVCU and Action Fraud / non-NECVCU datasets were then matched with data from Know-Fraud datasets using NFRC^{[footnote 5]} numbers to provide more complete information on cases. At the time of research, there was no way of filtering out NECVCU cases from Know Fraud and data matching was used to highlight cases which were genuine but not included in NECVCU data as well as NECVCU cases where victims had not received a service. During this stage, 2 files were created ahead of the cleaning stage. Both files then went through a multi-stage process to ensure only eligible leads with matched Know Fraud and NECVCU data were included in the sample.

Table 3.1: NECVCU leads removed at each cleaning stage

Total available leads	36,250

Data matching with Know Fraud and NECVCU systems	Leads lost
No and NAs from ‘on level one’ (Those who were on the database but whose case had not yet been processed)	35.9%
Not complete in Tier 1 ‘report status’ (Those where the call centre attempted a contact but it was unsuccessful or If not complete contact was made with the victim)	14.6%
Total leads available	17,944

Eligibility checking stage	Leads lost
Tier 2 listed as not on crossdata (Cross data is the database used by NECVCU to record details of victim cases if not on crossdata this indicates the victim did not receive an NECVCU service)	52.3%
Tier 2 listed as unable to contact	6.1%
Total leads available	7,465

Manual data cleaning stage	Leads lost
Those aged 17 and under removed	0.5%
Unusable telephone numbers	2.8%
Business addresses or no valid postcode	0.5%
Leads used for cognitive testing	0.5%
Flagged as unsuitable for interview by City of London Police	1.9%
Total leads available	6,993

Total sample leads	6,993
Total leads as % of leads provided	19%

Table 3.2: Non-NECVCU leads removed at each cleaning stage

Total available leads	167,280

Cleaning stage	Leads lost
Non England and Wales	7.1%
No valid postcode	2.3%
Those aged 17 and under and no date of birth	19.8%
Unusable telephone numbers	16.0%
Deceased	0.01%
Business addresses	0.2%
Total leads available	91,370

Total usable leads	91,370
Usable leads as % of leads provided	55%

In total, 203,171 leads were provided by CoLP which, after cleaning resulted in 91,370 available leads for the Action Fraud survey and 6,993 available leads for the NECVCU survey.

3.3 Sample checking

After the samples had been cleaned, they were checked by a second Ipsos researcher before being signed off. Specifically, it was checked that:

names, addresses, NFRC numbers and telephone numbers from the original file matched those in the clean sample; several records from different parts of the original file were checked in this way
those aged 17 and under were not included
the value in the ‘case date’ field corresponded to the original file name, that is, the month the case was reported
no blank or ‘don’t knows/not available’ cells were included in key fields such as name, address, telephone number
participants’ gender and name corresponded
the first address column contained no police/security addresses, and that no address columns contained text such as ‘unknown’ or ‘N/K’ (which would be printed onto the opt-out letter if not removed)

3.4 Sample selection

For the NECVCU sample a census approach was adopted with 6,993 participants being drawn for the sample. An initial mailing of 4,281 was issued (tranche 1) with an anticipated response rate of 28% to achieve 1,200 completed interviews. Each participant was sent a letter informing them of the survey along with the chance to opt-out. After a week of interviewing it was established that the survey was longer than anticipated and the response rate was also lower than anticipated (around 13%). Following conversations with the Home Office the remaining sample was issued (an additional mailout of 2,712) and the target number of completed interviews was revised down to 800.

For the Action Fraud victim survey a disproportionate sampling approach was used. Given the increased number of cases available compared with the NECVCU sample, it was decided to boost ‘equivalent’ Tier 2, cyber and disseminated cases to enable more robust analysis of these sub-groups. The sample was stratified by Tier (all available Tier 2 equivalent cases were selected in order to maximise the ability to provide analysis of vulnerable groups within the sample whilst acknowledging this was only a proxy and not directly equivalent to the flag in the NECVCU sample) and crime type (all cyber cases were selected) with a random selection made from the remaining cases which were stratified by police force and month reported, to ensure an even distribution across these factors. An initial mailout of 5,189 letters were issued with a target number of completes of 1,800 (an anticipated response of 35%). As with the NECVCU survey the response rate was lower than anticipated (23%) and the survey was also longer than expected. A second tranche of sample was issued (2,400 letters) bringing the total mailout to 7,589 participants. The letter included details on how to opt-out of the survey via phone, letter or email. The profiles of the cleaned sample and the final sample selection are included below.

Table 3.3: Profile of original, cleaned and final sample selection for NECVCU and Action Fraud surveys by Crime Type, Tier and Police Force Area

	NECVCU				Action Fraud
Crime Type	Original sample profile %	Cleaned sample profile %	Final selected sample %	n	Original sample profile %	Cleaned sample profile %	Final selected sample %	n
Fraud	91%	92%	92%	6,437	92%	91%	64%	4,824
Cyber	9%	8%	8%	556	8%	9%	36%	2,765
Tier 1 (or equivalent)	67%	84%	85%	5,914	55%	52%	47%	3,566
Tier 2 (or equivalent)	33%	16%	15%	1,079	20%	23%	25%	1,898
Disseminated - No	-	-	-	-	75%	74%	72%	5,464
Disseminated - Yes	-	-	-	-	25%	26%	28%	2,125
WMP Police Force Area	17%	21%	13%	918	1%	1%	2%	142
GMP Police Force Area	20%	28%	17%	1,220	1%	1%	1%	111
MPS/CoLP Police Force Area	63%	51%	29%	2,062	4%	4%	7%	503
Other Police Force Area	-	-	-	-	94%	94%	90%	6,833
Total			100%	6,993			100%	7,589

Checks were undertaken to ensure the demographic profiles between the original and final sample files including age, gender and ethnicity were broadly in line. Additional checks were also undertaken, for example, within the original sample files, victims had been assigned an ‘impact level’ as a means of categorising cases by severity. Victims self-defined their own impact level upon reporting the crime to Action Fraud, they could select from options ranging from ‘no impact’ or ‘minor’; to more ‘severe’ or ‘significant’ impacts, as set out in Table 3.4.

The profiles of the original sample, the final selected sample and the completed survey profile are included below and show the profiles are broadly in line across both surveys. This provided an additional check against bias in those responding to the survey, that is, that less/more severe cases were less/more likely to respond which could have impacted on survey findings. More detail on how weighting is applied to ensure the sample is representative of each population is included later in this report.

Table 3.4: Profile of original, cleaned and final sample selection for NECVCU and Action Fraud surveys by impact level

	NECVCU				Action Fraud
NECVCU Impact Level	Original sample profile %	Cleaned sample profile %	Final selected sample %	n	Original sample profile %	Cleaned sample profile %	Final selected sample %	n
Concerned about the fraud but it has not impacted on health or financial well being	34%	35%	39%		37%	37%	42%
Minor - only a small impact on either health or financial well being	22%	22%	23%		23%	22%	21%
Other	4%	4%	2%		5%	3%	3%
Severe - have received medical treatment as a result of this crime and/or at risk of bankruptcy	7%	7%	7%		7%	7%	6%
Significant - impacting on health or financial well being	33%	31%	29%		29%	31%	28%

3.5 Sampling qualitative depth interviews

The qualitative research focused on recipients of NECVCU support to allow probing around the interaction with the NECVCU service, whilst also allowing the scope to discuss broader experiences of fraud/ cyber crime and associated harms.

A total of 16 face to face qualitative depth interviews were conducted with victims in receipt of NECVCU support to gain a deeper understanding of the support service delivery and the broader impact that fraud and cyber crime has on victims.

The interviews were carried out both with a subset of people in the clean sample who had received NECVCU support. A recruitment screener was used to identify participants with relevant experience of the NECVCU service, those who were satisfied and dissatisfied and had been victims of fraud and cyber crime. The screener also allowed a mix of participants across demographics, Tier 1 and 2 location, with a range of vulnerabilities and perceptions of the services.

Interviews lasted approximately 60 minutes (though would not be limited if the participant wished to continue talking), giving the moderator enough time to build a bond with participants and the ability to explore their experience in detail.

3.6 Recruitment quotas and screening

Fieldwork for the depth-interviews took place between March and April 2019.

The sample was organised to provide a broad mix of demographics across the Action Fraud database for those in receipt of NECVCU services.

Sixteen in-depth interviews with NECVCU victims aged 18 years or over.

The number of available cyber cases was much lower than cases of fraud and subsequently cyber crime victims were much more difficult to reach.

One interview was conducted before it was decided that Tier 3 cases should not be included due to concerns about the potential impact on the victim of discussing the crime they had been a victim of.

Table 3.5: Profile of qualitative depth interview participants

Type of crime	Vulnerability of victims	Location	Gender	Tiers
15 Fraud	8 Vulnerable	11 London	8 Female	15 Tier 2
1 Cyber case	8 Non-vulnerable	3 Birmingham	8 Male	1 Tier 3
		2 Manchester

Question 4

4.  Fieldwork

Accepted Answer

Survey fieldwork took place between February 2019 and May 2019.

4.1 Opting-out provision

An ‘opt-out’ approach was also adopted for the survey elements of the research: this required Ipsos (acting on behalf of the Home Office) to alert potential participants that a research project was being undertaken and they would be allowed approximately 2 weeks to opt-out of taking part (see Appendices A and B in the Victim survey technical report - appendices. The letters contained information on what the survey would involve, why the survey was being conducted and how participants could find out further information on the survey. Participants would only reply if they did not wish to be interviewed. Those who opted-out via email, telephone or post were removed from the sample, while those who didn’t opt-out remained in the sample to be contacted.

Using an opt-out approach has benefits in research terms, as it is likely to result in a larger and less biased sample than using an opt-in method. This was particularly important given the available sample. Those who opt-in to a survey are also unlikely to be representative of the population, and this method therefore leads to a sampling bias which is less pronounced with an opt-out methodology. While those who opt-out of a survey may also be unrepresentative of the population (for example those who are most dissatisfied with a service), the number of opt-outs was relatively low (around 4%), and there remained a much lower risk of sample bias than with an opt-in approach.

4.2 Briefing interviewers

All interviews were carried out by fully trained and supervised interviewers who had wide experience in conducting research on sensitive subjects, and among similar audiences (such as research among victims on other projects). Interviewers were given a comprehensive briefing by senior researchers from the Ipsos project team before interviewing began. This briefing ensured that:

interviewers understood the background of the study, and the aims of the research
interviewers were fully familiar with the questionnaires, in particular the format and structures followed
interviewers were aware of the screening questions
interviewers were aware of potentially sensitive questions, and would reassure participants that all information gathered is confidential and would not be released to a third party
interviewers understood that some participants may become distressed during the interview - for instance, the interview may trigger painful memories related to the crime; interviewers were provided with detailed instructions and information about the survey and an escalation procedure to follow if they had concerns about a participant’s wellbeing which was developed with Action Fraud; interviewers were also were instructed to provide reassurance, provide victim and witness support helplines if appropriate, and if necessary, terminate the interview

Given the length and complexity of the face to face briefing, interviewers were also given written instructions outlining important points covered in the briefing to serve as a reference.

Interviewers were also supplied with copies of the opt-out and opt-in letters sent to participants, the privacy notice and the victim contact process map so they were aware of this procedure and the channels interviewees may have gone through. All interviewers received information on reducing harm to interviewees and the escalation procedure to follow if they had concerns about an interviewee.

4.3 Supervision and quality control

Ipsos is a member of the Interviewer Quality Control scheme (IQCS). Members of this scheme are required to follow set procedures in respect of interviewer recruitment, training, supervision and participant/data validation to ensure that all data is collected ethically and to a high standard. Each year Ipsos is inspected to ensure these standards are being met. Comprehensive records are also kept on interviewers’ ability, technique and work attitude.

During fieldwork, quality was assured through constant supervision and monitoring of interviewers. Monitoring involved listening-in to the interview as well as following it on screen. This enabled project supervisors to assess interviewers’ accuracy at recording information as well as hearing how well the interview was conducted. Interviewers were then shown the supervisor’s comments and, if necessary, assistance was given where there was evidence of weakness. Overall a minimum of 10% of the interviews were monitored.

As a further quality control, Ipsos executives would listen to several anonymised recordings of interviews that had taken place, across a variety of participant categories, to get a broad appreciation of how the project was running. Ipsos executives would watch out in particular for any changes to the questionnaire that had been made recently, interviewer performance, participant interest, and any areas where improvements could be made to improve questionnaire flow and participant understanding.

At the end of the interview, participants were offered the name of the executive at Ipsos Telephone Services and the appropriate telephone number (which could be called free of charge), plus the freephone number for the Market Research Society, who were able to verify that all of Ipsos’s work is confidential.

4.4 Response rates

Response rates were calculated for the NECVCU and Action Fraud victim surveys. Tables 4.1 and 4.2 illustrate these response rates, as well as the steps taken in their calculation. A more detailed description of response rates follows the tables.

Table 4.1: Breakdown of response for NECVCU survey

Outcome	Total	% total mailout
Tranche 1 mailout	4,281
Tranche 2 mailout	2,712
Total mailout	6,993	-
Opt outs	298	4
Refusals	1,958	28
Unusable numbers	1,450	21
Completed interviews	753	11

Table 4.2: Breakdown of response for Action Fraud survey

Outcome	Total	% total mailout
Tranche 1 mailout	5,189
Tranche 2 mailout	2,400
Total mailout	7,589	-
Opt outs	289	4
Refusals	1,742	23
Unusable numbers	1,500	20
Completed interviews	1,697	22

The response breakdowns detailed above show how many successful interviews were obtained as a proportion of all leads who received a letter informing them that they may be called to take part in a survey. Partial completes, that is, those who started the survey but dropped out totalled 271 for the NECVCU survey and 381 for the Action Fraud survey.

The overall response rate for the surveys is an adjusted response rate and corrects for the fact that some leads opt-out of the survey at the opt-out stage, some telephone numbers are unusable and some victims refused to take part when called. The rate of opt-outs was low for both surveys with 4% writing, emailing or calling to opt-out of the survey. Around 1-in-5 telephone numbers were unusable (21% for NECVCU, 20% for Action Fraud). The proportion of victims refusing to take part when contacted by the telephone centre was higher for the NECVCU survey (28%) than the Action Fraud victim survey (23%).

The adjusted response rates, that is, the total number of completes divided by the total eligible sample were 23% for the NECVCU survey and 42% for the Action Fraud victim survey.

4.5 Length of interview

For the NECVCU questionnaire, the expected interview length was 15 minutes and the final average interview length was 29. For the Action Fraud victim questionnaire the anticipated questionnaire length was 20 minutes and the final average length was 27 minutes.

The length of any one interview varied depending on the types of emotional or physical harms the participant had experienced, whether they remembered receiving a letter from Acton Fraud about their report and whether they had been contacted by the police in relation to their case.

Question 5

5.  Data processing, analysis and weighting

Accepted Answer

5.1 Data quality procedures

After interviewing was completed, data tables were produced, breaking down the results by key variables (for example, age, gender, social grade, cyber crime or fraud, Police Force Area, Tier, Satisfaction with NECVCU service among others). A number of quality procedures were then carried out to ensure that the data was accurate.

Two weeks after fieldwork started, a CATI topline (which shows the raw data for numbers of participants answering each question and giving each response) was checked to ensure the questionnaire routing was functioning correctly. This check followed the same procedure as checking the dummy topline, however at this stage the responses comprised of real data.

Halfway through fieldwork, a set of ‘interim tables’ were produced. These comprised a data table for each question of the NECVCU and Action Fraud victim questionnaires showing the results (both raw numbers of responses, and percentages). These data tables were first checked against a CATI topline to ensure consistency. They were then checked against the “Table Specification” document, which defines, consistent with the questionnaire, the routing at each question, as well as any other requirements, such as including combinations of responses (for example, ‘Satisfied’ = ‘Completely satisfied’ + ‘Very satisfied’ + ‘Fairly satisfied’). Finally, the tables were checked to ensure that responses summed to 100% (in the case of single code questions), that the raw numbers of responses summed to the total for each question, and that the total number of responses was fully accounted. Finally, the data was checked for sense.

The final data tables were further checked to ensure that the total number of responses matched the number of achieved interviews reported from the project manager at the telephone centre.

After the final tables had been checked, an SPSS file was produced and comprehensively checked. Key stages included checking all variables, including derived variables, were present in the file; checking all data labels and codes were correct; checking that single-coded and multi-coded variables were entered as such; checking by running frequency tables that the data in the SPSS file exactly matched the data in the tables; and checking that the routing at each question was correct by using SPSS syntax files which included the routing specifications for each question.

5.2 Data outputs

The main outputs of the surveys were as follows:

SPSS files holding the complete set of data for all participants for the NECVCU survey, the Action Fraud victim survey and a combined SPSS file on the combined emotional, financial and health impacts across both surveys.

Topline documents in Excel format. These documents summarised the results for each question in the NECVCU and Non- NECVCU surveys.

Three sets of data tables were delivered to the Home Office for the NECVCU survey, the Action Fraud victim survey as well as a combined dataset used for the harms report. An additional set of tables were produced for the Home Office to show only the cyber crime victims in the sample. The tables show data broken down by a number of variables derived from the questionnaire and the sample (for example, crime type, Tier), showing the statistical significance between these groups for each breakdown. Statistical significance testing on the data tables was carried out at the 5% risk level. This means that for a given result there was a 5% risk of concluding that a difference exists where there was no actual difference. The significance testing used in the data tables were t-test’s on column proportions (and/or means). Base sizes under 30 were not tested for statistically significant differences.

Selected data table outputs reflecting key findings for the NECVCU sample, the Action Fraud sample and the combined sample can be found in the published accompanying data tables.

5.3 Key Drivers Analysis

Key Drivers Analysis (KDA) is a type of regression analysis that shows how much of the variation in the responses to a single question (the dependent variable) can be explained by other information collected from the participant (the drivers). Whilst KDA does not show any causal path from one variable to another, it shows the relative importance of each driver in explaining the variation in the dependent variable, allowing the drivers with the strongest impact to be identified.

A KDA was run on the data to discover which harms experienced were most closely related to:

still feeling affected by the crime
still feeling vulnerable

The analysis was run using Shapley values. This approach entails running a multiple regression (ordinary least squares) and calculating the contribution of each explanatory variable in the model to the variation in the dependent variables, as measured by the overall R-squared. The R-squared is a broad measure of the model’s goodness of fit; it is usually a number between 0 and 1 and the closer it is to 1 the better the model is able to explain the dependent variable. The potential drivers included a range of questions relating to the effects of being a victim of crime. A full list of potential drivers is given in Table 5.1 below.

Table 5.1: Potential drivers for ‘still feeling vulnerable after the crime’ or ‘still feeling affected by the crime’

Type of harm	Question wording
Financial	Q.4ca (I borrowed money from friends or family) Which, if any of the following things happened to you as a result of the crime?
Financial	Q.4cb (I took out a loan from a bank/credit lender/other) Which, if any of the following things happened to you as a result of the crime?
Financial	Q.4cc (I had to change my bank accounts or details) Which, if any of the following things happened to you as a result of the crime?
Financial	Q.4cd (My retirement plans or pension was affected) Which, if any of the following things happened to you as a result of the crime?
Financial	Q.4ce (My credit rating was affected) Which, if any of the following things happened to you as a result of the crime?
Financial	Q.4cf (I had to pay to clean-up my computer/tablet/smartphone) Which, if any of the following things happened to you as a result of the crime?
Financial	Q.4cg (I incurred other bank charges or overdraft fees) Which, if any of the following things happened to you as a result of the crime?
Financial	Q.4ch (I took time off work) Which, if any of the following things happened to you as a result of the crime?
Financial	Q.4ci (I lost my business or my job) Which, if any of the following things happened to you as a result of the crime?
Financial	Q.4cj (I lost my home) Which, if any of the following things happened to you as a result of the crime?
Emotional	Q.5a (I experienced stress) Which if any of the following, happened to you as a result of the crime you experienced?
Emotional	Q.5b (I experienced anxiety) Which if any of the following, happened to you as a result of the crime you experienced?
Emotional	Q.5c (I experienced fear) Which if any of the following, happened to you as a result of the crime you experienced?
Emotional	Q.5d (I experienced anger) Which if any of the following, happened to you as a result of the crime you experienced?
Emotional	Q.5e (I felt alone) Which if any of the following, happened to you as a result of the crime you experienced?
Emotional	Q.5f (I felt embarrassed/ ashamed /self-blame or similar) Which if any of the following, happened to you as a result of the crime you experienced?
Health	Q.6a (Difficulty sleeping/fatigue) Which, if any, of the following things happened to you as a result of the crime?
Health	Q.6b (Change in appetite/weight loss/weight gain) Which, if any, of the following things happened to you as a result of the crime?
Health	Q.6c (Stress-related illness/condition) Which, if any, of the following things happened to you as a result of the crime?
Health	Q.6d (Panic or anxiety related illness) Which, if any, of the following things happened to you as a result of the crime?
Health	Q.6e (Depression) Which, if any, of the following things happened to you as a result of the crime?
Health	Q.6f (Self-harm) Which, if any, of the following things happened to you as a result of the crime?
Health	Q.6g (Suicidal thoughts) Which, if any, of the following things happened to you as a result of the crime?
Other	Q.8a (Your relationships with family and/or friends) How much, if at all, did the crime negatively impact the following things?
Other	Q.8b (Your sense of safety offline) How much, if at all, did the crime negatively impact the following things?
Other	Q.8c (Your sense of safety online) How much, if at all, did the crime negatively impact the following things?
Other	Q.8d (Your willingness to use the internet to perform necessary tasks) How much, if at all, did the crime negatively impact the following things?
Other	Q.8e (Your willingness to use social media) How much, if at all, did the crime negatively impact the following things?
Other	Q.8f (Your trust in other people) How much, if at all, did the crime negatively impact the following things?

The analysis discovered that the strongest driver of still feeling affected by the crime is whether trust in other people has been affected as a result of the crime. The strongest driver, the driver that explains most of the variation in the dependent variable, of still feeling vulnerable is whether victims experienced fear as a result of the crime. This suggests that a general sense of still feeling affected is closely related to whether people’s day-to-day trust in others has been impacted while feelings of vulnerability are more closely linked to the sense of fear experienced at the time of the crime.

Any potential drivers that were not significantly related to the outcome have been excluded from the model, since they would explain <1% of the variation in the outcome. The full set of potential drivers were entered into a stepwise regression model and any drivers with p-values lower than 0.05 were removed from the analysis as this indicates they have a weaker relationship with the dependent variable. The remaining drivers were included in the final model used for the KDA.

The full model outputs are provided in Tables 5.2 and 5.3 below.

Table 5.2: Model output for still feeling affecting by the crime

Drivers	Coef.	Standard error	t	P>t	Lower 95% CI.	Upper 95% CI.	Std. domin.
Q.8f (Your trust in other people) How much, if at all, did the crime negatively impact the following things?	0.17	0.02	7.16	0	0.13	0.22	15%
Q.6a (Difficulty sleeping/fatigue) Which, if any, of the following things happened to you as a result of the crime?	0.18	0.06	3	0	0.06	0.3	11%
Q.8a (Your relationships with family and/or friends) How much, if at all, did the crime negatively impact the following things?	0.18	0.03	5.73	0	0.12	0.24	11%
Q.6d (Panic or anxiety related illness) Which, if any, of the following things happened to you as a result of the crime?	0.22	0.07	3.38	0	0.09	0.35	7%
Q.8d (Your willingness to use the internet to perform necessary tasks) How much, if at all, did the crime negatively impact the following things?	0.08	0.03	3.05	0	0.03	0.13	9%
Q.4cd (My retirement plans or pension was affected) Which, if any of the following things happened to you as a result of the crime?	0.39	0.09	4.32	0	0.21	0.57	6%
Q.8b (Your sense of safety offline) How much, if at all, did the crime negatively impact the following things?	0.08	0.03	2.94	0	0.03	0.13	7%
Q.5e (I felt alone) Which if any of the following, happened to you as a result of the crime you experienced?	0.12	0.06	1.96	0.05	0	0.23	6%
Q.6e (Depression) Which, if any, of the following things happened to you as a result of the crime?	0.19	0.07	2.66	0.01	0.05	0.33	9%
Q.8c (Your sense of safety online) How much, if at all, did the crime negatively impact the following things?	0.05	0.02	1.99	0.05	0	0.09	6%
Q.4cg (I incurred other bank charges or overdraft fees) Which, if any of the following things happened to you as a result of the crime?	0.19	0.08	2.34	0.02	0.03	0.34	3%
Q.5c (I experienced fear) Which if any of the following, happened to you as a result of the crime you experienced?	0.11	0.05	2.14	0.03	0.01	0.22	5%
Q.5f (I felt embarrassed/ ashamed /self-blame or similar) Which if any of the following, happened to you as a result of the crime you experienced?	0.09	0.04	1.92	0.05	0	0.17	4%
Q.5d (I experienced anger) Which if any of the following, happened to you as a result of the crime you experienced?	0.11	0.05	2.14	0.03	0.01	0.21	2%
Constant	0.52	0.06	8.18	0	0.39	0.64	-

Model R-squared = 0.43

The model generates a coefficient (coef) for each driver that summarises the relationship between the driver and the dependent variable. The coefficients range between -1 and 1, with a positive value indicating a positive relationship between the driver and the outcome^{[footnote 6]} (that is, a participant agreeing that their trust in people had been negatively impacted were also more likely to agree that they were still feeling affected by the crime). Each coefficient has a corresponding standard error, and the results from a t-test with p-values (P

The model produced an R-squared statistic of 0.43, suggesting that the model explains 43% of the variation in the responses to the question feeling affected still by the crime. A score of 43% can be considered a fairly strong model for identifying the key drivers for how a participant feels affected by crime. The remaining 57% of variation in participants’ feelings about the crime is accounted for by variables outside the model, i.e. variables not measured in the survey or other collected data.

Table 5.3: Model output for still feeling vulnerable due to the crime

Drivers	Coef.	Standard error	t	P>t	Lower 95% CI.	Upper 95% CI.	Std. domin.
Q.5e (I felt alone) Which if any of the following, happened to you as a result of the crime you experienced?	0.27	0.06	4.84	0	0.16	0.38	12%
Q.8b (Your sense of safety offline) How much, if at all, did the crime negatively impact the following things?	0.12	0.03	4.36	0	0.06	0.17	14%
Q.6d (Panic or anxiety related illness) Which, if any, of the following things happened to you as a result of the crime?	0.27	0.07	3.98	0	0.14	0.4	13%
Q.5c (I experienced fear) Which if any of the following, happened to you as a result of the crime you experienced?	0.23	0.05	4.46	0	0.13	0.33	16%
Q.8c (Your sense of safety online) How much, if at all, did the crime negatively impact the following things?	0.06	0.02	2.97	0	0.02	0.1	13%
Q.6g (Suicidal thoughts) Which, if any, of the following things happened to you as a result of the crime?	0.35	0.14	2.51	0.01	0.08	0.62	5%
Q.5f (I felt embarrassed/ ashamed /self-blame or similar) Which if any of the following, happened to you as a result of the crime you experienced?	-0.14	0.04	-3.33	0	-0.22	-0.06	1%
Q.4cd (My retirement plans or pension was affected) Which, if any of the following things happened to you as a result of the crime?	0.23	0.1	2.28	0.02	0.03	0.43	6%
Q.4ca (I borrowed money from friends or family) Which, if any of the following things happened to you as a result of the crime?	-0.18	0.08	-2.34	0.02	-0.33	-0.03	1%
Q.4cc (I had to change my bank accounts or details) Which, if any of the following things happened to you as a result of the crime?	0.12	0.06	2.19	0.03	0.01	0.23	2%
Q.5a (I experienced stress) Which if any of the following, happened to you as a result of the crime you experienced?	0.11	0.05	2.44	0.02	0.02	0.21	8%
Q.4cb (I took out a loan from a bank/credit lender/other) Which, if any of the following things happened to you as a result of the crime?	-0.22	0.11	-1.94	0.05	-0.45	0	1%
Q.8a (Your relationships with family and/or friends) How much, if at all, did the crime negatively impact the following things?	0.06	0.03	1.85	0.05	0	0.13	9%
Constant	1.23	0.06	19.51	0	1.11	1.36	-

Model R-squared = 0.19

The model produced an R-squared statistic of 0.19, which means the model explains 19% of the variation in the question about whether the participant still feels vulnerable. This is lower than the model for still feeling affected, however, it still gives us a good indication of what other attitudes are strongly linked to these feelings of vulnerability.

5.4 CHAID analysis

In addition to the KDA, a CHAID (Chi-Squared Automatic Interaction Detector) analysis was conducted to explore which demographic variables were the best predictors of 2 target variables, namely, whether a victim still felt affected by the crime or whether a victim still felt vulnerable after the crime. A CHAID analysis is an algorithm used for discovering relationships between a target variable and predictor variables and is useful when looking for patterns in datasets as the relationships can be easily visualised.

CHAID is used to identify sub-groups of participants in a dataset who are similar in terms of a key survey characteristics, in this instance, whether the victim still felt affected by the crime or whether the victim still felt vulnerable.

CHAID identifies which variables in the dataset have the strongest relationship with the key survey characteristic, it then splits the data into sub-groups based on these variables. The output is a family tree type structure where successive branches indicate where the data has been split to produce the sub-groups. These final sub-groups are used to summarise, and therefore to better understand, the relationships between the various characteristics and the target variables.

The demographic variables used in the CHAID were:

crime type
gender
age
previous victim
social grade
ethnicity
qualifications

The analysis showed that the chosen target demographic characteristics were not able to discriminate well between the outcome categories because of the weak relationship between demographic characteristics and the outcomes. In other words, many of the demographic variables performed poorly when used to predict the dependent variable because they had a weak relationship that was not statistically significant when measured using a chi-squared test (the p-values were smaller than 0.05). However, it should be noted that sub-group analysis throughout this report suggests that certain types of victims are more likely to experience specific harms (particularly females, ethnic minority, cyber victims).

Figure 5.1: CHAID tree - ‘still feeling affected by the crime’

Figure 5.2: CHAID tree - ‘still feeling vulnerable after the crime’

5.5 Weighting the data

The main purpose of the project was to provide robust measures to help understand the experiences and satisfaction with the NECVCU service as well as the experiences of victims not in receipt of the service, but who may have had support from Action Fraud or the police. In order to do this, the results of the two surveys had to be representative of their respective populations. As a result, certain types of victim and witness were over- or under-represented in the raw results. Weighting is applied to data when we are not confident that we have interviewed a sample that is truly representative of the survey population. Weighting the data ensured that results were representative of the true population of victims.

A weight is a multiplicative factor assigned to each individual’s responses, which, while not changing their answer, gives it relatively more or less importance in the results, with a neutral weight being 1. For example, if cyber victims were under-represented in a sample, and fraud victims were over-represented, cyber victims would be given a weight of more than 1, while fraud victims would be given a weight of less than 1.

For both datasets, data was weighted to the original corresponding sample profile (NECVCU and Action Fraud) by crime type (fraud/cyber), police force area and Tier (or equivalent). For the combined dataset an additional weight was applied to factor for the NECVCU/Action Fraud splits, that is, the Action Fraud / non-NECVCU population was much larger than the NECVCU population.

Table 5.4: Unweighted and weighted profiles for NECVCU, Action Fraud and combine datasets

	NECVCU		Action Fraud		Combined
	unweighted	weighted	unweighted	weighted	unweighted	weighted
Crime Type
Fraud	91%	92%	65%	91%	73%	91%
Cyber	9%	8%	35%	9%	27%	9%

Tier
Tier 1 (or equivalent)	75%	84%	48%	52%	52%	57%
Tier 2 (or equivalent)	25%	16%	25%	23%	23%	25%
-	-	-	27%	26%	25%	19%

Police Force Area
WMP	19%	21%	-	-	-	-
GMP	29%	28%	-	-	-	-
MPS/CoLP	52%	51%	-	-	-	-
Other	-	-	-	-	-	-
Pilot area	-	-	5%	6%	34%	13%
Non-pilot area	-	-	95%	94%	66%	87%

Survey
NECVCU	-	-	-	-	30%	7%
Non-pilot area	-	-	-	-	69%	93%

5.6 Qualitative analysis

All qualitative data generated via the in-depth interviews was professionally transcribed by an approved supplier before being analysed in Excel. An analysis plan was drafted and shared with the Home Office for sign off prior to commencing analysis. This included a thematic analysis framework to allow for analysis by both case and theme to ensure a rigorous and robust interrogation of the data. Any stimulus materials or data captured outside of the interview recordings (such as pictorial or diagrammatic representations of victims’ journeys) were also be included in analysis and coded accordingly. Verbatim quotes were added to the analysis matrix in order to maintain a constant link and reference to the raw interview data.

5.7 Recall bias

When interpreting the findings, it is important to note that the survey was completed by victims between 9 to 22 months after their initial report to Action Fraud and/or support from the NECVCU. This means that the findings may be subject to some level of recall bias, where respondents do not accurately remember their experiences, or omit some details. It is not possible to determine from the findings how much recall bias affects the overall results.

Question 6

6.  Glossary of terms

Accepted Answer

Action Fraud only support: Those who recall receiving a letter from Action Fraud, who were not contacted by the police and whose case was not disseminated.

Computer Assisted Telephone Interviewing (CATI): computerised interviewing method. The survey questionnaire is programmed into CATI format so that questions and possible responses appear on-screen for interviewers to read. Interviewers then select from the answers displayed by typing numeric codes into their computer. The CATI system automatically routes the interviewer to the appropriate questions.

The Data Protection Act 1998 (DPA): the legal requirements which instruct those using personal details of members of the public on the correct use of personal data.

Disability: Those who have a physical or mental health condition or illness that has lasted or is expected to last 12 months or more which reduces their ability to carry out day to day activities.

Employment: Working - those in full-time or part-time work or self-employed; Not working - those who are not working (including students, retired, seeking work, looking after home, full time carer and voluntary work).

Multi-coded questions: questions to which several responses can be given. Other questions are single-coded questions, where only one response can be given.

Open-ended question: a question which allows participants to answer spontaneously, and where interviewers record their verbatim answers. This differs from a standard pre-coded question, which will anticipate the possible answers of a participant, and where interviewers might prompt participants to select from possible answers.

Opt-Out: a method of gaining participants’ consent to be approached for interview. A letter is sent explaining the purpose of the survey and how it will be conducted; those who do not wish to take part reply to ‘opt-out’ of further participation. Those who do not opt-out implicitly give their permission to be contacted for interview, although they are free to refuse to participate at this stage.

Police contact only: Those who recall being contacted by the police about their case, including disseminated and non-disseminated cases.

Police disseminated: Only includes those whose case was disseminated to police.

Pre-codes: a list of possible responses to a question; interviewers select the pre-code which best summarises the answer given by participants.

Routing: instructions in a questionnaire which direct interviewers to skip forward to other questions, based on the answers at previous questions.

Single-coded question: a question to which only one response can be given, as opposed to multi-coded questions where several responses can be given.

Social grade: based on the socio-economic classification produced by the ONS. AB - Higher and intermediate managerial, administrative and professional occupations; C1 - Supervisory, clerical and junior managerial, administrative, professional occupations; C2 - Skilled manual occupations; DE - Semi-skilled manual occupations, unemployed and lowest grade occupations.

Tier 1: This level is for victims who were not identified via the Action Fraud reporting system as being vulnerable and whose cases were not disseminated to the police for investigation. They receive advice and sign-posting via a short phone call.

Tier 2: Victims identified via the reporting system as vulnerable are sent to the tier 2 service for triage. These cases are checked for further signs of vulnerability. Eligible cases are provided enhanced support and guidance over the phone. Time spent on these cases can go on for some weeks/months depending on the needs of the individual and may trigger a Tier 3 referral.

Tier 3: This is an enhanced level of support provided locally by police forces to victims where concerns remain over a victim’s vulnerability and tends to result in face to face support being provided by the police or by other support services.

Types of support: definitions provided to survey participants to try and achieve consistency in understanding of support provided:

advocacy - includes speaking to different organisations on behalf of people - social services, a housing association or financial institution
crime prevention advice - includes specific guidance about behaviours, techniques or common scams to help people avoid being a victim of fraud or cyber crime
emotional support - includes talking to people about what happened, how they were feeling and what kind of support they needed
immediate support - includes where to go and who to talk to during, or immediately after, the crime occurred in order to stop the crime and prevent further loss or damage
practical support - includes practical advice such as organising a free phone blocker or home improvements
referrals to or information about other support services - includes suggestion that people contact other services such as Age UK, Victim Support, local council, Trading Standards, social services, MIND
welfare checks - including checking with follow up calls or visits to check on the welfare of victims

Vulnerability: The definition for vulnerability used throughout the reports is based on data collected from the survey, that is, whether the victim has self-identified as vulnerable. This was then used in the analysis to compare vulnerable to non-vulnerable victims.

Question 7

7.  Appendices

Accepted Answer

These are provided separately in the Victim survey technical report - appendices.

Those in receipt of NECVCU services were assigned a Tier based on their level of vulnerability. For those respondents in the Action Fraud only sample this was not possible, and so a flag was added into the cleaned sample files (where possible) indicating the likely equivalent Tier the respondent fell into. This was primarily added to use as a check during the sampling phase and was not used during analysis and reporting as it was only a proxy for Tier which could not be applied across the whole sample. ↩
Following cognitive interviewing and one depth interview with a Tier 3 victim it was decided not continue to include Tier 3 victims in the main stage of the research due to concerns about the potential impact on the participant of discussing the crime they had been a victim of. ↩
The Know Fraud system is the main data collection tool for Action Fraud and it holds details of victim reports. It was used to provide the survey samples as it is the most consistent and complete source available. ↩
The Action Fraud definition of vulnerable is self-defined with victims ticking one or more of 3 vulnerability boxes when they made a report to Action Fraud: The 3 pieces of information are (1) Whether they are repeat/prior victims, (2) Whether they are a regular target or (3) Whether they are at risk of losing money. ↩
An NFRC number refers to a crime reference number provided by Action Fraud after a victim has made their report. ↩
Note that the dependent variable and the potential drivers were coded in a consistent manner, meaning a higher value always indicated agreement with the statement. ↩