Professional Documents
Culture Documents
ComparingFedSurveys 2009
ComparingFedSurveys 2009
Comparing Federal
Government Surveys
that Count Uninsured
People in America
UPDATEDSEPTEMBER22,2009
ThenumberofuninsuredAmericansislargeandgrowingovertime,yettherecontinuestobedebateabout
exactlyhowmanyAmericansareuninsured.Researchersusedatafromseveraldifferentnationalsurveysto
estimatethenumberofuninsuredpeopleinAmericaanddiscrepanciesintheirestimatesarecommon.This
briefcomparesestimatesofhealthinsurancecoveragefromfournationalhealthsurveyssponsoredbythe
federalgovernment,identifiesthedifferencesintheirestimatesofuninsurance,andidentifiesthreecommon
themes:(1)allthesurveysreportverylargenumbersofAmericanslivingwithouthealthinsurance;(2)all
surveysshowthatthesenumbershaverisenoverthedecade,particularlyforadults;and(3)childrensrates
ofuninsurancedeclinedearlierinthedecade,largelyduetogrowthofpublicprogramssuchastheState
ChildrensHealthInsuranceProgram(SCHIP),butinrecentyearshasstayedsteadyorgrownslightly
(dependingonthesurvey).
Thesefourfederalsurveys,CPS,NHIS,MEPS,andACSwereeachdesignedtocollectspecificinformationfor
differentresearchandpolicypurposes.TheCPSisamonthlylaborforcesurveyandthehealthinsurance
questionsareaddedattheendofanannualsupplementtothemonthlyCPS.Conversely,theNHISandMEPS
aredesignedspecificallytocollecthealthinformationfromtheU.S.population.Collectinginformationon
healthinsurancecoverage,thetypeofcoverage,andlackofcoveragearehighpriorityquestionsinthese
surveys.TheACSisdesignedtoprovidecommunitiesthroughouttheU.S.withuptodateinformationonkey
demographicsandpolicyrelevantdata.TheACSisanongoingsurveythattheCensusBureauhasusedto
replacetheformerlongformdecennialquestionnaire.
UPDATEDSEPTEMBER22,2009
TABLE 1: Trends among the Surveys in the Number of People under 65 Years of
Age Who are Uninsured for Entire Year and Point in Time (in millions): 2007
Survey
CPS
45.0
N/A
CPS Adjusted*
42.9
N/A
NHIS
30.4
43.3
MEPS
39.9
53.5
Source:CurrentPopulationSurvey,2008withestimatestabulatedbytheStateHealthAccessDataAssistanceCenter;Cohenetal.,2009,
HealthInsuranceCoverageTrends,19592007:EstimatesfromtheNationalHealthInterviewSurvey;ChuandRhoades,2009,The
UninsuredinAmerica,19962008:EstimatesfortheU.S.CivilianNoninstitutionalizedPopulationunderAge65,StatisticalBrief#259,
AHRQ.
*TheseadjustedestimateshavebeendevelopedbySHADACtoaccountforhistoricalmethodologicalchanges.TheCPSdataare
reweightedandadjustedtoaccountforhistoricalchangesinthesurveysmethodology.Theseestimatesdonotcorrespondcompletelyto
estimatespublishedbytheCensusBureau,however,theycanprovideamoreaccurateassessmentofcoverageestimatesovertime.
Therangeofthepointintimeestimatesismuchsmaller(from43.3to53.5people)relativetothefullyear
estimates.ThefactthattheCPSfullyearuninsuredestimatefallswithintherangeoftheothertwosurveys
pointintimeestimatesmakesittemptingtointerprettheCPSestimateasactuallyapointintime
measure.4TreatingtheCPSaspointintimemeasureorasafullyearuninsuredmeasurecanleadtovastly
differentestimatesofthesizeofimportantpolicyrelevantgroupssuchasthenumberuninsuredpeoplewho
areeligibleforpublichealthinsurancecoverage.Thispartiallyexplainsvariationsinestimatesofthenumber
ofeligiblebutuninsuredkidsintheU.S.5
REFERENCE PERIOD
CPSrespondentsareexpectedtoremembercoveragetheyhadupto15monthspriortotheinterview,
whereastheACSaskscurrent(pointintime)coverage.TheNHISandMEPShaveamuchshorterrecall
periodaswellasapointintimemeasure.Asaresult,CPSrespondentstendtohavemorerecallerror(e.g.,
theyforgetaboutcoveragetheymayhavehadandtheymayreportcoveragetheyhaveatthetimeofthe
interviewratherthanthepriorcalendaryear).7
UPDATEDSEPTEMBER22,2009
mostsignificantamountofrecallerrorwithMedicaidenrolleesreportingasuninsured.Surveysusingpoint
intimemeasuresofcoverageresultinmodestupwardbiastouninsuranceestimates.8
Second,whilesomepeoplewithcoveragemayreportthattheyareuninsured,itisalsopossiblethatsome
uninsuredpeoplereporthavingcoverage.Therearetwolikelyreasonsforthesefalsereports.Firstisthe
tendencyofrespondentstochooseasociallyacceptableresponse.Havinghealthinsuranceisasocially
acceptableanswer,anduninsuredpeoplemayfeelpressuretoansweryestohavingcoverage.Second,
codingerrorsfavorareportofcoveragebecauseitonlytakesonepositiveresponsetoatypeofhealth
insuranceforarespondenttobeconsideredinsured.Itispossiblethatanyerrorsduetounderreportingare
offsetsomewhatbythepotentialcorrespondingtendencytofalselyreportcoverage.
Conclusion
ThenumberofuninsuredintheU.S.willneverbeexactlydetermined.Surveysproducedifferent
estimatesbecausetheyeachhandlecomplexsurveyandestimationissuesinslightlydifferentways.While
researchcanfocusonimprovingsurveysandestimationmethodologies,westillareleftwithjustthat:
estimates.Nonetheless,thesesurveysaretheonlytoolsavailabletoestimatethenumberofuninsured.The
bottomlineisthattherearetensofmillionsofuninsuredpeopleinthiscountry.Thenumberofuninsured
childrendeclinedearlierinthedecadeinallofthesurveys,butinrecentyearshasstayedsteadyinNHIS,
droppedintheCPS,andgrownintheMEPS.ThefirstyearofACSestimates,reflectingcalendaryear2008,
areconsistentwiththeCPS.
Graph1showsestimatesofchildrenunder18yearsofageinAmericawhoareuninsuredforanentireyear
fromtheCPS,CPSestimatesadjustedtoaccountforhistoricalchangesinthesurveysmethodology,andthe
pointintimeestimatesfromMEPSandNHIS.Theindividualestimatesvaryforreasonsdiscussedabove,yet
UPDATEDSEPTEMBER22,2009
theyallshowsimilartrends.Thenumberofuninsuredchildrenshowsadropinallthesurveysfrom2001
2004whilethenumberofuninsuredadultsisincreasingrapidlyinallthreesurveys(seeGraph2).For
children,theMEPSsurveyshowsanincreaseinthenumberofuninsuredthrough2008,whiletheNHIS
showsthenumberholdingsteadyandtheCPSshowsarecentdecline.Weareconductingfurtheranalysis
intowhythethreesurveysaredivergingforestimatesofcoverageamongchildren.
Asdetailedanalyseshavedemonstratedelsewhere,thedecreaseincoverageamongadultsismainlydueto
droppingprivatehealthinsurancecoverage.Forchildrenoverthistimeframe,therehasbeenbothan
increaseinthenumberenrolledinpublicprogramsandadecreaseinthenumberenrolledinprivate
coverage,sotheoverallnumberofuninsuredhasnotchangedsignificantlyordroppedslightly.12Focusingon
thedifferencesandmethodsusedtoproducethosenumbersisimportant,butshouldnotdistractpolicy
makersfromtheneedtoaddressissuesofcostandaccessforthosewithouthealthinsurancecoverageinthe
UnitedStates.
GRAPH 1: Trends among the Surveys in the Number of Children (under 18 years) who are
Uninsured at All Year (CPS) and Point-in-Time (MEPS and NHIS) (in millions)
12
CPS
CPSAdjusted*
10
MEPS
NHIS
ACS
6
2000
2001
2002
2003
2004
2005
2006
2007
2008
Note:SlightlynewmethodswereimposedintheNHISandCPS2005uninsuredestimates,resultinginfeweruninsured.
Source:CurrentPopulationSurvey,20012009withestimatestabulatedbytheStateHealthAccessDataAssistanceCenter;Cohenetal.,Health
InsuranceCoverage:EarlyReleaseofEstimatesfromtheNationalHealthInterviewSurvey,2008;andMEPSHConlinetables,Table5(multiple
years),http://www.meps.ahrq.gov/mepsweb/data_stats/quick_tables_results.jsp?component=1&subcomponent=0&year=
1&tableSeries=4&searchText=&searchMethod=1&Action=Search
*TheseadjustedestimateshavebeendevelopedbySHADACtoaccountforhistoricalmethodologicalchanges.TheCPSdataare
reweightedandadjustedtoaccountforhistoricalchangesinthesurveysmethodology.Theseestimatesdonotcorrespondcompletelyto
estimatespublishedbytheCensusBureau;however,theycanprovideamoreaccurateassessmentofcoverageestimatesovertime.2008
adjustedestimatesarenotyetavailable.
UPDATEDSEPTEMBER22,2009
GRAPH 2: Trends among the Surveys in the Number of Adults (18-64 years of age) who are
Uninsured at all Year (CPS) and Point-in-Time (MEPS and NHIS) (in millions)
50
CPS
45
CPSAdjusted*
40
MEPS
35
NHIS
30
ACS
25
2000
2001
2002
2003
2004
2005
2006
2007
2008
Note:SlightlynewmethodswereimposedintheNHISandCPS2005uninsuredestimates,resultinginfeweruninsured.
Source:CurrentPopulationSurvey,20012009withestimatestabulatedbytheStateHealthAccessDataAssistanceCenter;Cohenetal.,Health
InsuranceCoverage:EarlyReleaseofEstimatesfromtheNationalHealthInterviewSurvey,2008;andMEPSHConlinetables,Table5(multiple
years),http://www.meps.ahrq.gov/mepsweb/data_stats/quick_tables_results.jsp?component=1&subcomponent=0&year=
1&tableSeries=4&searchText=&searchMethod=1&Action=Search
*TheseadjustedestimateshavebeendevelopedbySHADACtoaccountforhistoricalmethodologicalchanges.TheCPSdataare
reweightedandadjustedtoaccountforhistoricalchangesinthesurveysmethodology.Theseestimatesdonotcorrespondcompletelyto
estimatespublishedbytheCensusBureau;however,theycanprovideamoreaccurateassessmentofcoverageestimatesovertime.2008
adjustedestimatesarenotyetavailable.
Acknowledgements
SupportforthisresearchcamefromtheRobertWoodJohnsonFoundation,theAgencyforHealthcare
ResearchandQuality,andtheAssistantSecretaryforPlanningandEvaluationattheU.S.Departmentof
HealthandHumanServices.Theviewsexpressedinthispaperarethoseoftheauthors,andnoofficial
endorsementbytheAgencyforHealthcareResearchandQualityortheDepartmentofHealthandHuman
Servicesisintendedorshouldbeinferred.
About SHADAC
TheUniversityofMinnestotasStateHealthAccessDataAssistanceCenter(SHADAC)helpsstatesmonitor
ratesofhealthinsurancecoverageandunderstandfactorsassociatedwithuninsurance.SHADACprovides
targetedpolicyanalysisandtechnicalassistancetostatesthatareconductingtheirownhealthinsurance
surveysand/orusingdatafromnationalsurveys.SHADACsworkisfundedbytheRobertWoodJohnson
Foundation.Informationisavailableatwww.shadac.org.
StateHealthAccessDataAssistanceCenter
2221UniversityAvenue,Suite345
Minneapolis,MN55414
Phone(612)6244802
UPDATEDSEPTEMBER22,2009
References
CongressionalBudgetOffice.2003.Howmanypeoplelackhealthinsuranceandforhowlong?EconomicandBudgetIssueBrief.WashingtonDC:
CongressionalBudgetOffice,May2003;FronstinP.2000.Countingtheuninsured:Acomparisonofnationalsurveys.IssueBriefEmployeeBenefit
ResearchInstitute225.WashingtonDC:TheEmployeeBenefitResearchInstitute,September2000;ShortPF.2001.Countingandcharacterizingthe
uninsured.WorkingPaperSeries.AnnArborMI:EconomicResearchInitiativeontheUninsured,December2001;Lewis,K,MRElwood,andJCzajka.
1998.Countingtheuninsured:Areviewoftheliterature.WashingtonDC:TheUrbanInstitute,July1998.
2U.S.CensusBureau.2002.CurrentPopulationSurveytechnicalpaper#63revised(TP63RV);Chapters15and16.WashingtonDC:U.S.CensusBureau,
March2002;DavernM,TJBeebe,LABlewett,KTCall.2003.RecentchangestotheCurrentPopulationSurvey:Sampleexpansion,healthinsurance
verificationandstatehealthinsurancecoverageestimates.PublicOpinionQuarterly67(4):60326;MedicalExpenditurePanelSurvey.Information
availableat:http://www.meps.ahrq.gov/PUFFiles/H82/h82doc.pdf;NationalCenterforHealthStatistics,DivisionofHealthInterviewStatistics.2002.
NationalHealthInterviewSurvey(NHIS)publicusedatarelease.NHISsurveydescription.Hyattsville,MD:CentersforDiseaseControlandPrevention,
December2003.Availableat:ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Dataset_Documentation/NHIS/2002/srvydesc.pdf.
3OthersurveysalsoprovidenationalestimatesbutarenotconductedbythefederalgovernmentsuchastheCommunityTrackingSurveyandtheNational
SurveyofAmericasFamilies.TheBehavioralRiskFactorSurveillanceSystemisconductedbythefederalgovernmentandcontainshealthinsurance
coverageestimateshoweverinmanystateschildrenareexcluded.
4ForadiscussionseeCongressionalBudgetOffice2003;SwartzK.1986.Interpretingtheestimatesfromfournationalsurveysofthenumberofpeople
withouthealthinsurance.JournalofEconomicandSocialMeasurementOct;14(3):23342;Lewis,K,MRElwood,andJCzajka.1998.Countingthe
uninsured:Areviewoftheliterature.WashingtonDC:TheUrbanInstitute,July1998.
5ForadiscussionofthisissueseeHudson,J.L.,andT.M.Selden.2007.ChildrensEligibilityandCoverage:RecentTrendsandaLookAhead.Health
Affairs26(5)w618w629.
6BecausethefirstreleaseofACShealthinsuranceestimateswillhappeninlateSeptember,2009,thisanalysisdoesnotincludeanyreferencetotheACS.
7KlermanJA,JSRingel,andBRoth.2005.UnderreportingofMedicaidandwelfareintheCurrentPopulationSurvey.WorkingPaper.SantaMonicaCA:
RAND,March2005;ShortPF.2001.Countingandcharacterizingtheuninsured.WorkingPaperSeries.AnnArborMI:EconomicResearchInitiativeon
theUninsured,December2001;SudmanS,NBradburn,andSSchwarz.1996.ThinkingaboutAnswers.SanFrancisco:JosseyBass;BhandariS.2004.
Peoplewithhealthinsurance:Acomparisonofestimatesfromtwosurveys.WorkingPaperNo.243.WashingtonDC:U.S.CensusBureau,June2004.
Availableat:http://www.census.gov/dusd/MAB/wp243.pdf;Lewis,K,MRElwood,andJCzajka.1998.Countingtheuninsured:Areviewofthe
literature.WashingtonDC:TheUrbanInstitute,July1998.
8KlermanJA,JSRingel,andBRoth.2005.UnderreportingofMedicaidandwelfareintheCurrentPopulationSurvey.WorkingPaper.SantaMonicaCA:
RAND,March2005;CallKT,GDavidson,ASSommers,RFeldman,PFarseth,andTRockwood.2002.UncoveringthemissingMedicaidcasesand
assessingtheirbiasforestimatesoftheuninsured.Inquiry38(4):396408;DavernM,KTCall,JZiegenfuss,DMcAlpineandTBeebe.2006.Arelow
responserateshazardoustoyourhealth?PaperPresentedattheTelephoneSurveyMethodologyIIConference,MiamiFL,January12,2006.
9DavernM.2005DoesimputationbiasleadtomoreuninsuredintheCurrentPopulationSurveysestimates?PresentedattheAcademyHealth2005
AnnualResearchMeeting,BostonMA,June27,2005.
10OfficeoftheAssistantSecretaryforPlanningandEvaluation(ASPE).2003.Understandingdifferentestimatesofuninsuredchildren:Puttingthe
differencesincontext.U.S.DepartmentofHealthandHumanServices,OfficeoftheAssistantSecretaryforPlanningandEvaluation,May2003.Available
at:http://aspe.hhs.gov/health/reports/uninsur3.htm
11Blumberg,S,JLuke,MCynamon,andMFrankel.2006.RecenttrendsinhouseholdcoverageintheUnitedStates.PresentationattheTelephoneSurvey
MethodsIIconferenceinMiami,Florida,January2006.
12DeNavasWaltc.,B.D.Proctor,andJ.Smith.2007.Income,Poverty,andHealthInsuranceCoverageintheUnitedStates:2006.CurrentPopulation
SurveyReports#P60233.WashingtonDec:U.S.CensusBureau;andHolahan,J.andA.Cook.2008.TheU.S.EconomyandChangesinHealthInsurance
Coverage,20002006.HealthAffairs27(2):w135w144(publishedonline20February2008).