Continuous Surveys: Statistical Challenges and Opportunities Carl Schmertmann Center for Demography & Population Health Florida State University [email protected].

Continuous Surveys: Statistical Challenges and Opportunities

Carl SchmertmannCenter for Demography & Population HealthFlorida State University

[email protected]

Outline

CHALLENGES (long)

Increased Temporal Complexity Increased Sampling Error New Weighting Problems

OPPORTUNITIES (brief, but important)

Sample Size Comparison US CENSUS LONG FORM:

--- 17% / decade

ACS ROLLING SURVEY: 2 per 1000 Households / month 24 per 1000 Households / year 240 per 1000 Households / decade--- 24% / decade

Sampling Differences over Decade

Long Form ACS

Sample Size ≈ 17% ≈ 24%

Taken on… 1 day 3650 days

Released as… 1 dataset 10+ datasets

Simultaneous100% count? YES NO

1. Temporal Complexity

Long Form ACS






What is the Population? 1-Day Census

Population membership is binary: {0,1}

Each individual is IN or OUT

Continuous Survey Population membership is fuzzy:

0 --------------- + ---------------1

Individuals can be MORE IN (more person-days of residence) or MORE OUT (fewer)



J F M A M J J A S O N D ●

Type A 10 10 10 10 10 10 10 10 10 10 10 10 120

Type B 2 2 2 2 10 10 10 10 10 2 2 2 64

● 12 12 12 12 20 20 20 20 20 12 12 12 184

Residents (in 000s)



Type A 10 10 10 10 10 10 10 10 10 10 10 10 120

Type B 2 2 2 2 10 10 10 10 10 2 2 2 64

● 12 12 12 12 20 20 20 20 20 12 12 12 184

Residents (in 000s)

Census Population = 12 000 (83% Type A)



Type A 10 10 10 10 10 10 10 10 10 10 10 10 120

Type B 2 2 2 2 10 10 10 10 10 2 2 2 64

● 12 12 12 12 20 20 20 20 20 12 12 12 184

Residents (in 000s)

An ACS ‘Data Sandwich’ includes samples from all months



Type A 10 10 10 10 10 10 10 10 10 10 10 10 120

Type B 2 2 2 2 10 10 10 10 10 2 2 2 64

● 12 12 12 12 20 20 20 20 20 12 12 12 184

Residents (in 000s)

ACS samples from 184 000 person-months Avg Population: 15 333 (65% Type A)

Characteristics change over the Sampling Period

Persons Age Marital Status Employment Education

Housing Units Vacancy Number of Occupants $ Value


Rolling ‘Population’

Population formed by sandwiching monthly samples is the average frame of a film, not a snapshot

Individuals and housing units with changing characteristics are sampled and caught ‘in motion’.


Reference Period Problems

Many ‘long-form’ questions refer to retrospective periods:

Income in last 12 months Place of residence 1 year ago Child born in last 12 months? Etc.


Time Reference Example

‘2004’ data from 12 monthly samples taken in Jan04…Dec04

Question on fertility in the 12 months prior to the survey, so there are 12 overlapping periods in ‘2004’ data ‘Jan04’ question covers Jan03-Jan04 ‘Feb04’ question covers Feb03-Feb04 etc.


Jan 2004 x x x x x x x x x x x x ● . . . . . . . . . . .Jan 03

Jan 04

Feb 2004 . x x x x x x x x x x x x ● . . . . . . . . . .Mar 2004 . . x x x x x x x x x x x x ● . . . . . . . . . Apr 2004 . . . x x x x x x x x x x x x ● . . . . . . . . May 2004 . . . . x x x x x x x x x x x x ● . . . . . . . Jun 2004 . . . . . x x x x x x x x x x x x ● . . . . . . Jul 2004 . . . . . . x x x x x x x x x x x x ● . . . . . Aug 2004 . . . . . . . x x x x x x x x x x x x ● . . . . Sep 2004 . . . . . . . . x x x x x x x x x x x x ● . . . Oct 2004 . . . . . . . . . x x x x x x x x x x x x ● . . Nov 2004 . . . . . . . . . . x x x x x x x x x x x x ● . Dec 2004 . . . . . . . . . . . x x x x x x x x x x x x ●

1

2

3

4

5

6

7

8

9 10 11 12 11 10 9 8 7 6 5 4 3 2 1

Jan 05



Reference Periods for ‘Last 12 Month’ Questions in 1-year ACS Datasets

Temporal Issues Summarized

‘Data Sandwiches’ contain: New meaning of ‘population’

Units that change over sampling period (moving targets)

Multiple reference periods for retrospective questions


2. Sampling Error

Long Form ACS





2. Sampling Error

Small Samples

More overall data from continuous sampling, but…

1-, 3-, or 5-Year Sandwiches have smaller samples than the single, decennial long form survey more sampling error

in published data

2. Sampling Error

Small Samples

The problem is especially acute for small areas narrow age groups rare subpopulations

e.g., How many unmarried teen births per year in Sevier County, Tennessee?

ACS 2006-2008 says 0 ± 161

2. Sampling Error

2. Sampling Error

St. Johns County, FL 2006 1-Year ACS Data for Males

BELOW POVERTY ABOVE POVERTY POVERTY RATE

AGE Estimate MOE Estimate MOE Percent MOE*

0-4 746 +/-562 3,495 +/-501 17.6 +/-13.3

5 0 +/-300 906 +/-467 0 +/-33.1

6-11 376 +/-363 5,401 +/-769 6.5 +/-6.3

12-14 231 +/-292 2,787 +/-768 7.7 +/-9.7

15 0 +/-300 1,342 +/-460 0 +/-22.4

16-17 0 +/-300 1,995 +/-417 0 +/-15.0

18-24 1,235 +/-655 5,387 +/-878 18.6 +/-9.9

25-34 221 +/-371 10,192 +/-889 2.1 +/-3.6

35-44 202 +/-194 11,558 +/-785 1.7 +/-1.6

45-54 581 +/-399 12,794 +/-807 4.3 +/-3.0

55-64 468 +/-452 10,679 +/-550 4.2 +/-4.1

65-74 245 +/-200 5,825 +/-248 4.0 +/-3.3

*Denominators have MOE≈0 under ACS sampling and weighting design

2. Sampling Error

C24020. SEX BY OCCUPATION – Key West, Florida Data Set: 2006-2008 American Community Survey 3-Year Estimates (http://tinyurl.com/acs-alap)

…etc

Temporal Instability

Teenage Birth Rate in a County

Unfortunate Result

Aggregating over 1+ years of surveys produces datasets that are often

Unfamiliar and difficult to understand

Still too noisy to be useful for planners and researchers

2. Sampling Error

3. Weighting for Non-Response

Long Form ACS





3. Weighting Problems

Weighting

Weighting from

Respondents Total Population

requires Population Control Totals:

(Place x Age x Sex x Race x Ethnicity x …)


Decennial Long Form Sample Control Totals

Measured from a simultaneous

enumeration of the population(Sample & Census on same day)

Only 1 set needed

Sample and Population defined identically (resid. on Census Day)


Continuous Survey

Control Totals Must be estimated (no simultaneous

census)

Many sets needed (2006, 2007, 2006-8, 2007-9, 2008-12, …)

Sample and Population defined differently


ACS Control Totals (Persons)


ACS responses are weighted to match official intercensal estimates by

• Year (1 July midpoint snapshot) • County (sometimes city)• Age• Race• Sex • Hispanic Origin (yes/no)

ACS Control Totals (Persons)


Potential Errors Estimates are Wrong:

Unanticipated internal migration Unanticipated international migration etc

Population Definition don’t match Seasonal fluctuations Different race/ethnic categories



Type A 10 10 10 10 10 10 10 10 10 10 10 10 120

Type B 2 2 2 2 10 10 10 10 10 2 2 2 64

● 12 12 12 12 20 20 20 20 20 12 12 12 184

Census Pop = 12 000 (83% Type A)Average Pop = 15 333 (65% Type A)

If every year looks like this…Intercensal Estim= 12 000 (83% Type A)

Weighting Error Example

ACS weighting to estimates produces:

Popn too small (Census < Avg Pop) Popn too “A” (seasonal Bs missed) Overestimates of vars + correl. with A

(e.g., % with college education) Underestimates of vars - correl. with

A (e.g., % single-parent families)


Opportunities

CensusSurvey

ContinuousSurvey

Frequency

Recency

Sample Error

Familiarity

4. Opportunities

Statistical models that exploit likely cell relationships (over times, ages, sexes, places, variables …) could, in principle

Opportunities

ACS table cells = millions of “seemingly unrelated” maximum likelihood estimates

4. Opportunities

Retain frequency & recency Reduce variance of estimates Recover familiar measures

Conclusion

5. Conclusion

CONTINUOUS SURVEYS like ACS create

Big Problems for producers and users Unfamiliar, temporally complex data Potentially high sample error Technical problems with weighting

Big Opportunities, IF we can develop appropriate statistical models and practices

5. Conclusion

Thanks!

¡Gracias!

Obrigado!

Continuous Surveys: Statistical Challenges and Opportunities Carl Schmertmann Center for Demography & Population Health Florida State University [email protected].

Documents

x x x x x x x x x x

temporal complexity

decade slide

s slide

yesno slide

important slide

type b222210

rolling population population