Collecting the Data Tim Vizard, Office for National Statistics.

Slides:



Advertisements
Similar presentations
Family Resources Survey Data Collection Methods Jo Maher (National Centre for Social Research) Tom Howe (Office for National Statistics)
Advertisements

4. NLTS2 Data Sources: Parent and Youth Surveys. 4. Sources: Parent and Youth Surveys Prerequisites Recommended modules to complete before viewing this.
ALB-3 Paper 1 Headline Measures and Business Intelligence Data Pack.
Brian A. Harris-Kojetin, Ph.D. Statistical and Science Policy
ESS Mixed mode experiment ESRA Conference, Ljubljana, July 2013 Alun 18 July 2013.
Mental Health Survey 2015: Webinar 14 th January 2015.
What’s new in the Child Poverty Unit – Research and Measurement Team Research and Measurement Team Child Poverty Unit.
Scot Exec Course Nov/Dec 04 Survey design overview Gillian Raab Professor of Applied Statistics Napier University.
1 Data Linkage for Educational Research Royal Statistical Society March 19th 2007 Andrew Jenkins and Rosalind Levačić Institute of Education, University.
The Challenge of Non- Response in Surveys. The Overall Response Rate The number of complete interviews divided by the number of eligible units in the.
© 2012 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.
3.14 X AXIS 6.65 BASE MARGIN 5.95 TOP MARGIN 4.52 CHART TOP LEFT MARGIN RIGHT MARGIN ©TNS 2013 Are ‘better’ interviewers more successful at.
McGraw-Hill/Irwin Business Research Methods, 10eCopyright © 2008 by The McGraw-Hill Companies, Inc. All Rights Reserved. Chapter 9 Surveys.
Volunteers and Volunteering Organisations:
South East London Homeless Health Needs Audit
Miami, Florida, 9 – 12 November 2016
Research Methods for Business Students
Chapter 1 Introduction and Data Collection
Checking and Advice Service
The Association of Exposure to Adverse
30 Hour Free Childcare Application Information
Context for the experiment?
2007 Household Travel Survey
Evaluating Survey Data Collection Methods
Market Research Unit 5 - slide 13.
WHO The World Health Survey General Introduction
Results from a survey of UK patients at risk of bowel cancer, their experiences and information preferences. Selina Goodman, Heather Skirton, Ray Jones.
RESEARCH METHODS Lecture 20
SAMPLING OF PARTICIPANTS
Taken from Irwin Mitchell 2017
Workshop A2: Non-response Challenges in GPS Surveys
Improving the Lives of Callers: Call Outcomes and Unmet Needs
LISA, Anticipating the Next Generation of Longitudinal Data
Section 3: Sweep implementation
The second wave of the new design of the Dutch EU-SILC: Possibilities and challenges Judit Arends.
Performance Task Overview
Section 2: Longitudinal study samples
LISA, Anticipating the Next Generation of Longitudinal Data
The European Statistical Training Programme (ESTP)
Schools that work for everyone
School Sport Survey and FE Sport and Active Lifestyles Survey 2018
Recording and monitoring uptake of JE vaccine
Multi-Mode Data Collection Approach
Active Lives: Children and Young People Survey
HOW TO SCREEN PATIENTS AND BOOK THE BASELINE VISIT APPOINTMENT?
How we use Your Health Records
Section 1 The beginning: aims, objectives and feasibility
Chapter 12: Other nonresponse correction techniques
Immediate activity.
The Computer-Assisted Personal
PRODCOM SURVEY IN THE UNITED KINGDOM
Schools that work for everyone
Business Statistics: A First Course (3rd Edition)
Development and Testing
Salah Merad Methodology Division, ONS
Multi-Mode Data Collection Approach
Field procedures and non-sampling errors
Implementing mixed mode questionnaire in FI-SILC
Andrew Jenkins and Rosalind Levačić
Mean vs Median Sampling Techniques
Changes in the Canadian Census of Population Program
Mental Health of Children and Young People
Primary Partnership Headteachers Meeting
Multi-Mode Data Collection
Deciding the mixed-mode design WP1
When do I need to register by?
Chapter 5: The analysis of nonresponse
RESEARCH METHODS Lecture 20
GDPR Information and Consent
When do I need to register by?
Presentation transcript:

Collecting the Data Tim Vizard, Office for National Statistics

Data Collection Aim: Collect data about 9,500 children and young people Multi-Informant Approach, with parents, children and teachers invited to take part. Different modes depending on age of sampled child. Interviews were conducted via: Computer Assisted Personal Interview (CAPI) Computer-Assisted Self Interview (CASI). Paper / Online Questionnaire (Teachers) Aim: Collect data on 9,500 children and young people. Multi-Informant Approach - with parents, children and teachers invited to take part. Different modes depending on age of sampled child. Interviews were conducted via: Computer Assisted Personal Interview (CAPI) Computer-Assisted Self Interview (CASI). Paper / Online Questionnaire (Teachers)

Mode of Data Collection 2 to 4 year olds: Parent interview only (Interviewer administered and self completion) 5 to 10 year olds: Teacher interview (Postal or online questionnaire)

Mode of Data Collection 11 to 16 year olds: Parent interview (Interviewer administered and self completion) Child interview (Interviewer administered and self completion) Teacher interview (Postal or online questionnaire) 17 to 19 year olds: Young person interview (Interviewer administered and self completion) Parent interview (if present at same address)

Interviewing Parents and Children Due to the sensitivity of some questions in the survey, interviewers were encouraged, where possible, to conduct: interviews with parents without the sampled child (or other children) present interviews with the sampled child without parents or other children present Due to the sensitivity of some questions in the survey, interviewers were encouraged, where possible, to conduct: - interviews with parents without the sampled child (or other children) present - interviews with the sampled child without parents or other children present

Supporting participants All interviewers received a Disclosure and Barring Service (DBS) standard level check. Interviewers were provided with safeguarding training tailored to the sensitive nature of the survey. Materials were developed and left with participants to support those affected by survey content. Prior to fieldwork, interviewers were required to complete a Disclosure and Barring Service (DBS) standard level check. This was required due to the sensitive nature of the survey content. In addition to this, interviewers were provided with safeguarding training tailored to the sensitive nature of the survey. Safeguarding is “protecting people’s health, wellbeing and human rights, and enabling them to live free from harm, abuse and neglect” Interviewers provided with clear guidance to follow on the survey. Materials were developed and left with participants to support those affected by survey content.

Supporting participants Here’s an example of our useful contacts leaflet. It explained that “If you have been affected by anything you have discussed during the interview listed below are contact details for organisations that you might find helpful” This included your local GP, NHS services, and charites such as childmind and youngminds.

Response Rates All Ages Issued 18,029 Ineligible 393 (2%)   All Ages Issued 18,029 Ineligible 393 (2%) Eligible households 17,636 Refusals 4,956 (28%) Non-contacts 2,194 (12%) Other unproductive 1,369 (8%) Productive households 9,117 (52%) Full interviews 9,019 Partial interviews 98 WE issue 18029 addresses (15690 main sample and 2339 reserve sample). There were 393 ineligible addresses. These were identified by interviewers following visiting an address (for example, sampled participant had moved and was untraceable). The number of ineligible addresses reported here include 49 addresses which were assumed to be ineligible, however this was not confirmed by an interviewer (and instead a judgement was made from alternative sources of information). This resulted in 17,636 eligible households. Of these, 28% resulted in a refusal ( either to the interviewer, or a telephone call to the office following receipt of an advance letter). 12% of addresses were non-contactable eligible addresses. Although the table shows this figure to be 2194 addresses, in fact there were 2243 addresses where no contact could be made. 49 of these addresses were identified as not-eligible (for example sampled participant had moved and was untraceable) and therefore included in the ineligible total above. A further 8% of addresses were unproductive. Therefore, we achieved a 52% response rate on our eligible sample. (9117 interviews). This included 98 partials interviews - defined as participants completing the interview up to the SDQ section and the first DAWBA module.

Response Rates (BY AGE)

Strategies for improving response Incentives Named sample Updated addresses Re-issues Reserve sample Incentives: An unconditional incentive (£10 post office high street voucher), printed on the advance letter. Children aged 11 to 16 years old who completed an interview were entered into a prize draw to receive a Love2shop gift card of £20, with 50 “winners” were selected at random. Named Sample: Person level sampling frame. Advance letters addressed to parents of the named child. Updated addresses 12% of issued sample had moved address, therefore updated sample information requested from the NHS Patient Register for all participants who had registered an address change with their GP. When the new addresses were: outside of England, interviews were not attempted; in England but outside of the survey postcode sector, interviews attempted subject to availability. Re-issues Where an interview was not achieved, 4% of main sample addresses were reissued. This included addresses where no contact was made or a refusal was given in the initial round of fieldwork. Reserve sample Reserve sample was introduced part-way through fieldwork to increase the number of achieved interviews. The size of the issued reserve sample differed by age group, depending on the response rates by age. The reserve sample resulted in an extension of fieldwork into Sept/Oct for the 2 to 4 and 17 to 19 year olds. Additional interviews with 5 to 16 year olds were mainly completed by June 2018, to ensure comparability with the fieldwork period for this age group in the 2004 survey, enable teacher interviews to take place before the end of the school Summer term.

Interview Length (provisional)* IT’s also interesting to look at the interview length for the household interview (with parents and children). This histogram shows the average interview length for the household interview. It’s provisional as it only shows interview length for a selection of achieved interviews, so may not be representative of the whole survey. IT has also been trimmed to remove outliers. However it does show: -The majority of interviews took between 60-90 minutes. Anecdotally, the interviews which took less time were more likely to be conducted with one informant only (parents of 2-16 year olds, or the 17-19 year old on their own). Likely to be households without complex mental health needs. Interviews which too the longest time were anecdotally: More likely to have more than 1 person interviewed. Come from households with more complex mental health needs. Although some of the longest interviews were tiring for interviewers and participants, these were also some of the most important interviews to understanding the complex needs of children and young people. Estimated interview based on selection of completed interviews, not representative of final interview length

Teacher Interview Following completion of the parent / child interview, we also collected information from the teacher. To enable this, we required: Consent from the parent to contact their child’s teacher The name of the school, teacher and headteacher An email address and postal address for the school To reduce the burden on interviewers and participants, a lookup was built into the questionnaire to populate some of this information based on the school name.

Teacher Response Number % of eligible % of invited % of completed   Number % of eligible % of invited % of completed Eligible 5 to 16 year olds 6,665 Consent received 5,930 89% Invited to complete questionnaire 5,718 86% Complete teacher questionnaires 3,595 54% 63% Completed on paper 2,292 64% Completed online 1,303 36% Although interviews were achieved with 6,718 5 to 16 year olds, 6,665 5 to 16 year olds were asked to provide consent to a teacher interview, and subsequently eligible (this reduction was mainly a result of partial interviews and item non-response). Of our eligible sample, consent was received from 89% (5930). Following a review of contact details, 86% of our eligible teacher sample were invited to complete a questionnaire. Questionnaires were returned from 3595 teachers. Of these 3595 teachers, 64% completed the questionnaire on paper, and 36% took part online.

Number of interviews with individuals by age group Although interviewing multiple participants was a strength of this survey, this was not possible in all cases. For 2-4 year olds, we collected information from 1463 parents and young people. For 5-10 year olds, in addition to the 3,597 parent interviews, we achieved an additional 2050 teacher interviews. For 11-16 year olds, we achieved: 3121 parent interviews In addition, 2609 child interviews were conducted. Furthermore 1545 teacher interviews were achieved (although some of teacher interviews may have only included a parent interview without a child interview). As information was not collected for all teachers of children aged 5 to 16, this was accounted for by applying an adjustment factor to minimise bias. Information was not collected from teachers for children aged 2 to 4 and 17 to 19 years old and should be taken into account when comparing rates across age groups. For 17-19 year olds: 936 interviews were conducted with the young person (our primary sampling unit) Additional interviews were conducted with 421 parents.

Item Non-Response 2 to 4 5 to 10 11 to 16 17 to 19 Ethnic group 1 (0%) 0 (0%) Equivalised household income 140 (9.6%) 331 (9.2%) 354 (11.3%) 518 (55.3%) Income related benefits* 133 (9.1%) 260 (7.2%) 335 (10.7%) 577 (61.6%) Disability benefits 135 (9.2%) 273 (7.6%) 338 (10.8%) 576 (61.5%) Neighbourhood deprivation Region (GOR) Family functioning 27 (1.8%) 53 (1.5%) 80 (2.6%) 529 (56.5%) Parent mental health 21 (1.4%) 44 (1.2%) 61 (2%) 525 (56.1%) General health 1 (0.1%) Special educational needs 11 (0.8%) 22 (0.6%) 23 (0.7%) 519 (55.4%) Item non-response introduces problems such as non-response bias; this is where there are differences in the respondents who participated and those who did not. For 2-16 year olds, item non-response was mainly found in Income and benefit questions Additional item non-response was found for family functioning, parental mental health and SEN. For 17-19 year olds, levels of item non-response were much higher: item non-response was over 50% for income, benefits, family functioning / parental mental health and SEN. Reasons for item non-response: Participants were either: Unwilling to provide answers to some questions (for 2-16 year olds). Not asked questions. 17 to 19 were eligible to be interviewed without a parent present, and in these instances some questions related to socioeconomic status and family characteristics were not asked Survey results have excluded cases with item non-response. This assumes that the characteristics of participants who answered each question are the same of those who did not provide an answer. Therefore users were informed they should be mindful of the impact non-response may have on estimates from this survey.