COLLECTING DATA ON VIOLENCE AGAINST CHILDREN THROUGH SURVEYS Claudia Cappa, Senior Adviser, Data and Analytics Section, UNICEF
OVERVIEW 1. REVIEW OF AVAILABLE COUNTRY STUDIES 2. RECCOMENDATIONS ON HOW TO ADDRESS DATA NEEDS THROUGH SURVEYS
REVIEW OF AVAILABLE COUNTRY STUDIES
Elements covered in the assessment Commissioning and implementing agencies Definitions and indicators Sample designs Research protocols Ethical protocols Field coordination Quality control and data processing
Basic characteristics 34 studies had a specific focus on VAC (stand alone studies) 8 studies were general surveys with modules or questions on VAC - 4 national surveys 4 part of international programs (MICS, DHS, GSHS, HBSC) 34 were meant to be representative at the national level
Years and frequency of implementation Among the 42 national studies : 12 conducted in 2008 alone 13 conducted after 2008, the latest in 2017 17 studies conducted before, the earliest in 2002-2003 34 studies were conducted just once
Number of national studies by region Coverage by region Number of national studies by region Note: This table does not included countries that collected data on VAC through MICS, DHS, GSBS or HBSC.
Commissioning and implementation Of the 42 national studies identified, most were commissioned by government agencies (21), followed by NGOs (10), international organizations (11) and academic institutions Most of the studies conducted before 2006 were undertaken by research institutions or NGOs. During and after 2006, 33 studies were conducted with governments’ participations Most studies conducted by consultants
Some information on the study design Most studies were household surveys, few school based Full questionnaire available for only 15 of the studies In 16 studies self-administered questionnaire, 13 interviews, rest unknown Respondents: children from age 5 (mostly adolescents) and adults (as victims and as perpetrators)
Variations in age of target population from selected surveys on VAC Country Survey name Target population Botswana, Cambodia, Haiti, Honduras, Lao PDR, Kenya, Malawi, Nigeria, Rwanda, Tanzania, Uganda, Zambia, Zimbabwe Violence against Children Surveys All children and young adults aged 13-24 Chile Maltrato Infantil y Relaciones Familiares en Chile Children attending grade 8 India Study on Child Abuse in India Children aged 5-18 and young adults aged 18-24 Moldova Violence against Children in the Republic of Moldova Children attending grades 5 to 12 (aged 10-18) and primary caregivers Swaziland Violence against Children Survey Girls and young women 13-24 Philippines National Baseline Study on Violence against Children All children and young people aged 13-24 The Balkans The Balkan Epidemiological Child Abuse and Neglect study Children from the general school-going population of 11, 13 and 16-year olds UK Child Abuse and Neglect in the UK Today Children aged 11-17, parents/guardians of children under age 10, and young adults aged 18-24
Definitions and indicators Vast majority of the surveys identified in this review used their own definitions (15) In a few cases, definition used reflected national legal framework (3) Frequent references to the CRC (24) Some references to the WHO definitions of violence and abuse (5) Rationale for selection the target population or definitions not given in most cases (17)
Questionnaire design Majority of surveys developed/used own tools 12 studies relied on modified versions of the CTS or ICAST Limited information on how the tools were developed and selected Cognitive testing= no information/not done Field testing of the questionnaire prior to survey implementation = no information/not done Pilot testing in 27 cases
Types of violence covered
Types of violence covered Physical abuse Corporal punishment Sexual abuse Emotional abuse Neglect Bullying No. of Studies X 4 1 2 3
Questionnaire content Content: experience of violence (35), attitudes (4), perpetuation (3) Most surveys collected information of lifetime experience of violence, 12 surveys collected information on recent experience of violence (12 months, 6 months or last month) Risk/protective factors: data collected in 3 cases only
Implementation Length of training for the field teams unknown: 23 studies two weeks: 4 studies 3 to 7 days: 8 1 or 3 days: 5 Content of the training largely unknown Interviewers’ profiles: unknown for 35 studies, general interviewers for most countries, social workers in 1 countries Teams composition and size: largely unknown (information available for 8 studies only)
Main findings Highly fragmented sector Key terms defined on an ad-hoc basis that was unique to each specific study Absence of a clear (theoretical) research framework and of rationale behind chosen study design Most studies conducted only once Limited attention given to ethical protocols
RECCOMENDATIONS ON HOW TO ADDRESS DATA NEEDS THROUGH SURVEYS
Content of the guidelines Considerations for planning (identifying a lead ministry and coordination structure, establishing technical working groups, preparing a budget and timeline) Survey design (target population, definitions, sampling, mode of administration, content of questionnaire) Survey implementation (content of the training, support during field work, field work procedures, obtaining informed consent, conducting the interview, strategies to minimize non-response and managing refusals, quality control) Considerations for data entry and data analysis
Variables Prevalence of violence – life time and/or last 12 months Characteristics of the child and his/her family Frequency of violence Relationship of victim to perpetrator Severity of violence Location of violence Help seeking Health outcomes Risky behaviours Attitudes HIV knowledge and status
Target population Both sexes should be included Lower age limits to be defined on the basis of ethical considerations, legal considerations, and cognitive ability to provide valid and reliable information Upper limit influenced by recall bias and exposure to risk Important to understand and made explicit excluded population due to study design
Comparison of methods to collect data on violence against children Item Interviewer-administered Self-administered Face-to-face interviews Supervised Unsupervised Sample type Probability Non-probability Cost High Medium Low Interviewers required Yes No (assistance personnel only) No Interview length Long Short Response rates Willingness to disclose sensitive information Medium (highly dependent on interviewer)
Comparing results from different modes of data collection in South Africa Two separate surveys: one in schools and another in households Main questionnaire administered by interviewer. Each respondent was also asked to complete short self-administered questionnaire Sexual abuse, higher prevalence rates using the self-administered approach (more than twice among girls, five times among boys) Intimate partner violence: lower prevalence rates using the self-administered approach School survey more likely to produce higher rates than the household survey
Comparing results from different modes of data collection in Lao Anonymous reporting substantially lower reports of physical and emotional violence substantially higher reports of sexual abuse, particularly among females
Sampling designs to protect confidentiality and safety of respondents Only one child interviewed per household to minimize breach in confidentiality Survey for males is conducted in different enumeration areas than the survey for females to protect the confidentiality of respondents and reduce the chance that opposite-sex perpetrators and victims in the same community would both be interviewed
Importance of testing Cognitive testing of questions Validations of responses through triangulation Careful review of refusals, non-responses and missing data Testing should include ethical protocols and all procedures
THANK YOU ccappa@unicef.org