2016 S 2016 Sample Redesign of the l R d i f th National Health - PowerPoint PPT Presentation

2016 S 2016 Sample Redesign of the l R d i f th National Health Interview National Health Interview Survey Survey Chris Moriarity Van Parsons Chris Moriarity, Van Parsons National Center for Health Statistics 2015 Joint Statistical Meetings 2015 Joint Statistical Meetings Summer Conference Preview/Review August 3, 2015

Presentation outline Presentation outline National Health Interview Survey (NHIS) sample design periods Timeline for the 2016 NHIS sample p redesign Decision-making process for the year 2016 NHIS "base sample" y p allocation

NHIS sample design NHIS sample design periods p Each sample design period is ~10 years long based on information years long, based on information from previous decennial census Most recent completed period: 1995-2005, based on 1990 Census Current period: 2006-2015, based on Census 2000 on Census 2000

Some historic NHIS sample design features Emphasis on producing precise national estimates - sample national estimates - sample allocation by state approximately proportional to state population proportional to state population size Most NHIS sample designs (i (including current one) have l di t ) h sampled in all U.S. States and D.C.

Motivation for periodic p NHIS sample redesigns Reselect sample areas to take account of changes in the account of changes in the distribution of the U.S. population Accommodate changes in survey design features and analytical d i f t d l ti l objectives

Recent NHIS sample p redesigns: relatively minor changes i h 1995-2005 design: began using 1995 2005 d i b i screening as part of the mechanism to oversample black mechanism to oversample black and Hispanic persons Current 2006-2015 design: g expanded oversampling to include Asian persons

Timeline for the 2016 NHIS sample redesign 2012: interagency group (Census Bureau NCHS) created a milestone Bureau, NCHS) created a milestone schedule for the redesign 2013-2015: workgroups at Census g Bureau/NCHS implementing the redesign January 2016: implementation January 2016: implementation

2016 NHIS sample p redesign: several major changes h Build in more flexibility to Build in more flexibility to increase/decrease overall sample and/or shift sample allocations by and/or shift sample allocations by State from year to year, if desired (lead time required to implement) (lead time required to implement) N New source of sample addresses f l dd

NHIS sample address p source, 1985-2015 Sample addresses for the 1985-94, 1995-2005 and current NHIS 1995-2005, and current NHIS sample designs have come primarily from field listing primarily from field listing Th The Census Bureau can share C B h NHIS sample addresses with NCHS NCHS uses the addresses for other surveys, record linkage, etc.

2016 NHIS sample p design: change in sample address source l dd Other demographic surveys conducted by the Census Bureau conducted by the Census Bureau moving toward using the Master Address File (MAF) as the main ( ) source of sample addresses NCHS has been sharing costs for NHIS field listing; NCHS cannot NHIS field listing; NCHS cannot afford the full cost of listing

2016 NHIS sample p design address source: address list(s) dd li t( ) NCHS does not want to use MAF addresses for NHIS because MAF addresses for NHIS because MAF addresses are confidential, they cannot be shared NHIS will use one or more NHIS will use one or more commercial address lists as main sample address source beginning sample address source, beginning in 2016

Issues to address when using a commercial address list dd li t Accuracy, geocoding quality, coverage of areas like college coverage of areas like college dormitories, etc. How to do periodic updates How to select annual samples that do not overlap with previous ones

2016 NHIS sample p design will still include some field listing fi ld li ti Rural route/highway contract route/P.O. Box addresses not route/P.O. Box addresses not suitable for personal visit interviewing g Situations like large apartment Situations like large apartment buildings with a single mail delivery point: listing likely needed delivery point: listing likely needed at least part of the time

NHIS sample redesign: p g milestones reached Primary sampling units (PSU) defined PSU sample selected defined, PSU sample selected National address list has been acquired; preliminary quality y y assessment completed Identified areas where field listing is required for year 2016 NHIS is required for year 2016 NHIS

NHIS sample redesign: p g current work College dormitory sample frame Developing methodology for within PSU sampling stages within-PSU sampling stages Doing necessary field work preparation (listing, training p p ( g g personnel, etc.)

NHIS sample redesign: p g future milestones Implementing the year 2016 NHIS sample allocation sample allocation Monitoring/evaluating the initial Monitoring/evaluating the initial year 2016 implementation Planning for changes/adjustments g g j identified by monitoring/evaluation Developing estimation structures

Planning for year-to-year g y y flexibility in allocation "Base sample" to remain at about ~35,000 completed household 35,000 completed household interviews annually ~25,000 completed household interviews to remain stable interviews to remain stable, allocated proportional to state population each year population each year remaining ~10,000 flexible

The year 2016 NHIS y sample allocation Lead time of at least one year, in general, to decide on a given general, to decide on a given year's allocation of the flexible ~10,000 completed household 10,000 completed household interviews More than one year of lead time for year 2016 sample allocation year 2016 sample allocation decision, given this is first year of new sample design new sample design

Year 2016 NHIS base sample allocation options ti Extensive research conducted on Extensive research conducted on various state-level sample allocations that retained ~25 000 allocations that retained ~25,000 completed household interview stable core stable core F Four options presented to the ti t d t th NCHS Office of the Director in O t b October 2014 2014

Year 2016 NHIS sample p allocation Option #1 Similar to historic NHIS sample allocations - goal is to maximize allocations goal is to maximize precision of national-level estimates estimates Allocate entire sample proportional Allocate entire sample proportional to state population One-year estimates for 21 states; three-year estimates for 40 states

Year 2016 NHIS sample p allocation Option #2 Modify Option #1 by reducing sample in 40 most populous states sample in 40 most populous states to increase sample in 10 least populous states and DC to ~250 populous states and DC to 250 completed household interviews Enables three-year estimates for all 50 states and DC all 50 states and DC Slight loss of national precision

Year 2016 NHIS sample p allocation Option #3 Modify Option #1 by reducing sample in 25 most populous states sample in 25 most populous states to increase sample in 25 least populous states and DC to ~525 populous states and DC to 525 completed household interviews Enables two-year estimates for all 50 states and DC 50 states and DC Noticeable loss of precision

Year 2016 NHIS sample p allocation Option #4 Modify Option #1 by reducing sample in 11 most populous states sample in 11 most populous states to increase sample next 30 populous states to ~650 completed populous states to 650 completed household interviews Enables one-year estimates for 41 states states Loss of precision (less than #3)

Year 2016 NHIS sample p allocation decision Option #2 Retains precision of national-level estimates as highest priority estimates as highest priority Enables three-year estimates for all 50 states and DC

Year 2016 NHIS sample p augmentation On March 23, 2015, we learned that funds are being provided for year funds are being provided for year 2016 NHIS sample augmentation Major challenge to implement the augmentation simultaneously with t ti i lt l ith keeping the overall sample redesign on schedule d i h d l Recently finished implementation

Summary The NHIS undergoes periodic The NHIS undergoes periodic sample redesigns every ~10 years Next NHIS sample design will be implemented in January 2016 implemented in January 2016 Year 2016 NHIS base sample Year 2016 NHIS base sample allocation decision has been made Funds provided in March 2015 for year 2016 sample augmentation year 2016 sample augmentation