280 likes | 629 Views
Response rate comparison of ELSA and HRS. Hayley Cheshire Shaun Scholes Mary Beth Ofstedal. Response Framework . Framework developed by Peter Lynn (ISER) Aim for best practice Main features: Focus on “study” rather than “field” response rate Classify “eligibility” at each wave
E N D
Response rate comparison of ELSA and HRS Hayley Cheshire Shaun Scholes Mary Beth Ofstedal
Response Framework • Framework developed by Peter Lynn (ISER) • Aim for best practice • Main features: • Focus on “study” rather than “field” response rate • Classify “eligibility” at each wave • Cross-sectional and longitudinal rates • unconditional rate (based on eligibility) • conditional rate (based on participation at a prior wave).
Cross-sectional rates • Focus on a particular wave of data collection • Numerator = those who have taken part at a given wave • Provide a “snap shot” of representativeness • What proportion of eligible people took part at this wave? • Measure of wave on wave success • What proportion of people who participated at the previous wave also took part at this wave?
Longitudinal rates • Summarise response over a number of waves • Numerator = those who have participated in every wave • Measure of study representativeness • What proportion of people eligible at all waves were interviewed at all waves? • Measure of success of panel maintenance strategies • What proportion of people who did a wave 1 interview went on to participate in all further waves?
THEORETICAL EXAMPLES OF EACH RESPONSE RATE
Example 1 (CS – unconditional) W1 = 900/1000 W2 = 860/977
Example 2 (CS – conditional) W2|W1 = 850/882
Example 3 (Long – unconditional) W2=850/977
Example 4 (Long – conditional) W2|W1=850/882 W3,W2|W1=750/870
Comparing ELSA and HRS • To understand calculation of response rates: • survey design • eligibility criteria • To understand differences in response rates between studies • look at fieldwork practices
ELSA Eligibility for Wave 1 • Living within the household at the time of the HSE interview in 1998, 1999 or 2001 • Born before 1 March 1952 • Still living at a private residential address in England at the time of the ELSA wave 1 interview in 2002-3
Non-response to Health Survey for England • ELSA response calculations need to factor in non-response to HSE interview • HSE cooperating households • Age information was collected • able to ascertain ELSA eligibility • HSE non-cooperating households • Age information NOT collected • unable to ascertain ELSA eligibility • Estimate eligibility in non-cooperating households • Assume same proportion of people aged 50+ in non-cooperating households as cooperating.
HRS Sample Sources • Original HRS • Screening of 69, 337 household units in 1992. • AHEAD • 1992 household screening operation (for those born between 1914 and 1923, and about half of those born in 1913 or before) • Medicare enrolment files (for the other half of those born in 1913 or before) • War Baby • 1992 household screening exercise • CODA sample • Medicare enrolment file • Early baby boomer • Screening of 38,385 households.
Sources of non-response from HRS Sample Design • 1992 Screening Exercise • 14% were determined to be non-sample • Eligibility NOT established in only 214 of the 59,918 identified households. • Adjustments made for non-response • Medicare Enrolment files • Adjustments made for non-response • 2004 Screening Exercise (Early baby boomer sample) • Eligibility was determined in 91.3% of the screened households.
BACKGROUND TO RESPONSE CALCULATIONS
Who is included in response calculations for ELSA and HRS? • Follow-up eligible households with at least one Wave 1 interview • Eligible individuals not interviewed at baseline are treated as non-respondents • Rates based on age-eligible sample members (excludes younger/older partners) • Productive = full, partial or proxy interview • Apply mortality rate for “unknown eligibles”
Who is included in response calculations for ELSA and HRS wave 2 onwards?
What can we draw from these comparisons? • Impact of sample design on non-response • Impact of fieldwork practices • Mode • Use of proxy interviews • Use of incentives • Cross country differences
THANK YOU WHAT COMMENTS?