Appendix 1.Understanding and Using ACS Single-Year and Multiyear EstimatesWhat Are Single-Year and MultiyearEstimates?Understanding Period EstimatesThe ACS produces period estimates of socioeconomicand housing characteristics. It is designed to provideestimates that describe the average characteristics ofan area over a specific time period. In the case of ACSsingle-year estimates, the period is the calendar year(e.g., the 2007 ACS covers January through December2007). In the case of ACS multiyear estimates, theperiod is either 3 or 5 calendar years (e.g., the 2005–2007 ACS estimates cover January 2005 throughDecember 2007, and the 2006–2010 ACS estimatescover January 2006 through December 2010). The ACSmultiyear estimates are similar in many ways to theACS single-year estimates, however they encompass alonger time period. As discussed later in this appendix,the differences in time periods between single-yearand multiyear ACS estimates affect decisions aboutwhich set of estimates should be used for a particularanalysis.While one may think of these estimates as representingaverage characteristics over a single calendar year ormultiple calendar years, it must be remembered thatthe 1-year estimates are not calculated as an average of12 monthly values and the multiyear estimates are notcalculated as the average of either 36 or 60 monthlyvalues. Nor are the multiyear estimates calculated asthe average of 3 or 5 single-year estimates. Rather, theACS collects survey information continuously nearlyevery day of the year and then aggregates the resultsover a specific time period—1 year, 3 years, or 5 years.The data collection is spread evenly across the entireperiod represented so as not to over-represent anyparticular month or year within the period.Because ACS estimates provide information aboutthe characteristics of the population and housingfor areas over an entire time frame, ACS single-yearand multiyear estimates contrast with “point-in-time”estimates, such as those from the decennial censuslong-form samples or monthly employment estimatesfrom the Current Population Survey (CPS), which aredesigned to measure characteristics as of a certaindate or narrow time period. For example, Census 2000was designed to measure the characteristics of thepopulation and housing in the United States basedupon data collected around April 1, 2000, and thus itsdata reflect a narrower time frame than ACS data. Themonthly CPS collects data for an even narrower timeframe, the week containing the 12th of each month.Implications of Period EstimatesMost areas have consistent population characteristicsthroughout the calendar year, and their periodestimates may not look much different from estimatesthat would be obtained from a “point-in-time” surveydesign. However, some areas may experience changesin the estimated characteristics of the population,depending on when in the calendar year measurementoccurred. For these areas, the ACS period estimates(even for a single-year) may noticeably differ from“point-in-time” estimates. The impact will be morenoticeable in smaller areas where changes such as afactory closing can have a large impact on populationcharacteristics, and in areas with a large physical eventsuch as Hurricane Katrina’s impact on the New Orleansarea. This logic can be extended to better interpret 3year and 5-year estimates where the periods involvedare much longer. If, over the full period of time (forexample, 36 months) there have been major orconsistent changes in certain population or housingcharacteristics for an area, a period estimate for thatarea could differ markedly from estimates based on a“point-in-time” survey.An extreme illustration of how the single-year estimatecould differ from a “point-in-time” estimate within theyear is provided in Table 1. Imagine a town on the Gulfof Mexico whose population is dominated by retireesin the winter months and by locals in the summermonths. While the percentage of the population in thelabor force across the entire year is about 45 percent(similar in concept to a period estimate), a “point-intime” estimate for any particular month would yieldestimates ranging from 20 percent to 60 percent.Table 1. Percent in Labor Force—Winter t.Nov.Dec.202040606060606060503020Source: U.S. Census Bureau, Artificial Data.Appendix A-1U.S. Census Bureau, A Compass for Understanding and Using American Community Survey Data

(encompassing 2005–2009) for all geographic areas—down to the tract and block group levels. Whileeventually all three data series will be available eachyear, the ACS must collect 5 years of sample beforethat final set of estimates can be released. This meansthat in 2008 only 1-year and 3-year estimates areavailable for use, which means that data are onlyavailable for areas with populations of 20,000 andgreater.The important thing to keep in mind is that ACSsingle-year estimates describe the population andcharacteristics of an area for the full year, not forany specific day or period within the year, while ACSmultiyear estimates describe the population andcharacteristics of an area for the full 3- or 5-yearperiod, not for any specific day, period, or year withinthe multiyear time period.Release of Single-Year and Multiyear EstimatesNew issues will arise when multiple sets of multiyearestimates are released. The multiyear estimatesreleased in consecutive years consist mostly ofoverlapping years and shared data. As shown in Table2, consecutive 3-year estimates contain 2 years ofoverlapping coverage (for example, the 2005–2007ACS estimates share 2006 and 2007 sample data withthe 2006–2008 ACS estimates) and consecutive 5-yearestimates contain 4 years of overlapping coverage.The Census Bureau has released single-year estimatesfrom the full ACS sample beginning with data fromthe 2005 ACS. ACS 1-year estimates are publishedannually for geographic areas with populations of65,000 or more. Beginning in 2008 and encompassing2005–2007, the Census Bureau will publish annualACS 3-year estimates for geographic areas withpopulations of 20,000 or more. Beginning in 2010,the Census Bureau will release ACS 5-year estimatesTable 2. Sets of Sample Cases Used in Producing ACS Multiyear EstimatesYear of Data ReleaseType of estimate20082009201020112012Years of Data 7–20092008–20102009–20115-yearestimatesNot AvailableNot Available2005–20092006–20102007–2011Source: U.S. Census Bureau.Differences Between Single-Year and Multiyear ACS Estimatessingle year is the midyear of the ACS multiyear period(e.g., 2007 single year, 2006–2008 multiyear).CurrencyFor example, suppose an area has a growing Hispanicpopulation and is interested in measuring the percentof the population who speak Spanish at home. Table 3shows a hypothetical set of 1-year and 3-year estimates. Comparing data by release year shows that foran area such as this with steady growth, the 3-yearestimates for a period are seen to lag behind the estimates for the individual years.Single-year estimates provide more current information about areas that have changing population and/orhousing characteristics because they are based on themost current data—data from the past year. In contrast,multiyear estimates provide less current informationbecause they are based on both data from the previousyear and data that are 2 and 3 years old. As noted earlier, for many areas with minimal change taking place,using the “less current” sample used to produce themultiyear estimates may not have a substantial influence on the estimates. However, in areas experiencingmajor changes over a given time period, the multiyearestimates may be quite different from the single-yearestimates for any of the individual years. Single-yearand multiyear estimates are not expected to be thesame because they are based on data from two different time periods. This will be true even if the ACSReliabilityMultiyear estimates are based on larger sample sizesand will therefore be more reliable. The 3-year estimates are based on three times as many sample casesas the 1-year estimates. For some characteristics thisincreased sample is needed for the estimates to bereliable enough for use in certain applications. Forother characteristics the increased sample may not benecessary.A-2 AppendixU.S. Census Bureau, A Compass for Understanding and Using American Community Survey Data

Table 3. Example of Differences in Single- and Multiyear Estimates—Percent of PopulationWho Speak Spanish at HomeYear of datarelease1-year estimates3-year estimatesTime periodEstimateTime 00516.82003–200515.9Source: U.S. Census Bureau, Artificial Data.Multiyear estimates are the only type of estimatesavailable for geographic areas with populations of lessthan 65,000. Users may think that they only need touse multiyear estimates when they are working withsmall areas, but this isn’t the case. Estimates for largegeographic areas benefit from the increased sampleresulting in more precise estimates of population andhousing characteristics, especially for subpopulationswithin those areas.In addition, users may determine that they want to usesingle-year estimates, despite their reduced reliability,as building blocks to produce estimates for meaningful higher levels of geography. These aggregations willsimilarly benefit from the increased sample sizes andgain reliability.the estimates. All of these factors, along with anunderstanding of the differences between single-yearand multiyear ACS estimates, should be taken into consideration when deciding which set of estimates to use.Understanding CharacteristicsFor users interested in obtaining estimates for smallgeographic areas, multiyear ACS estimates will be theonly option. For the very smallest of these areas (lessthan 20,000 population), the only option will be touse the 5-year ACS estimates. Users have a choice oftwo sets of multiyear estimates when analyzing datafor small geographic areas with populations of at least20,000. Both 3-year and 5-year ACS estimates will beavailable. Only the largest areas with populations of65,000 and more receive all three data series.Deciding Which ACS Estimate to UseThree primary uses of ACS estimates are to understand the characteristics of the population of an areafor local planning needs, make comparisons acrossareas, and assess change over time in an area. Localplanning could include making local decisions such aswhere to locate schools or hospitals, determining theneed for services or new businesses, and carrying outtransportation or other infrastructure analysis. In thepast, decennial census sample data provided the mostcomprehensive information. However, the currencyof those data suffered through the intercensal period,and the ability to assess change over time was limited.ACS estimates greatly improve the currency of datafor understanding the characteristics of housing andpopulation and enhance the ability to assess changeover time.Several key factors can guide users trying to decidewhether to use single-year or multiyear ACS estimatesfor areas where both are available: intended use of theestimates, precision of the estimates, and currency ofThe key trade-off to be made in deciding whetherto use single-year or multiyear estimates is betweencurrency and precision. In general, the single-yearestimates are preferred, as they will be more relevantto the current conditions. However, the user must takeinto account the level of uncertainty present in thesingle-year estimates, which may be large for smallsubpopulation groups and rare characteristics. Whilesingle-year estimates offer more current estimates,they also have higher sampling variability. One measure, the coefficient of variation (CV) can help youdetermine the fitness for use of a single-year estimatein order to assess if you should opt instead to use themultiyear estimate (or if you should use a 5-year estimate rather than a 3-year estimate). The CV is calculated as the ratio of the standard error of the estimateto the estimate, times 100. A single-year estimate witha small CV is usually preferable to a multiyear estimateas it is more up to date. However, multiyear estimatesare an alternative option when a single-year estimatehas an unacceptably high CV.Appendix A-3U.S. Census Bureau, A Compass for Understanding and Using American Community Survey Data

Table 4 illustrates how to assess the reliability of1-year estimates in order to determine if they shouldbe used. The table shows the percentage of householdswhere Spanish is spoken at home for ACS test counties Broward, Florida, and Lake, Illinois. The standarderrors and CVs associated with those estimates are alsoshown.In this illustration, the CV for the single-year estimatein Broward County is 1.0 percent (0.2/19.9) and inLake County is 1.3 percent (0.2/15.9). Both are sufficiently small to allow use of the more current singleyear estimates.Single-year estimates for small subpopulations (e.g.,families with a female householder, no husband, andrelated children less than 18 years) will typically havelarger CVs. In general, multiyear estimates are preferable to single-year estimates when looking at estimatesfor small subpopulations.For example, consider Sevier County, Tennessee, whichhad an estimated population of 76,632 in 2004 according to the Population Estimates Program. This population is larger than the Census Bureau’s 65,000population requirement for publishing 1-year estimates. However, many subpopulations within thisgeographic area will be much smaller than 65,000.Table 5 shows an estimated 21,881 families in SevierCounty based on the 2000–2004 multiyear estimate;but only 1,883 families with a female householder, nohusband present, with related children under 18 years.Not surprisingly, the 2004 ACS estimate of the povertyrate (38.3 percent) for this subpopulation has a largestandard error (SE) of 13.0 percentage points. Usingthis information we can determine that the CV is 33.9percent (13.0/38.3).For such small subpopulations, users obtain moreprecision using the 3-year or 5-year estimate. In thisexample, the 5-year estimate of 40.2 percent has anSE of 4.9 percentage points that yields a CV of 12.2percent (4.9/40.2), and the 3-year estimate of 40.4 percent has an SE of 6.8 percentage points which yields aCV of 16.8 percent (6.8/40.4).Users should think of the CV associated with anestimate as a way to assess “fitness for use.” The CVthreshold that an individual should use will vary basedon the application. In practice there will be manyestimates with CVs over desirable levels. A generalguideline when working with ACS estimates is that,while data are available at low geographic levels, insituations where the CVs for these estimates are high,the reliability of the estimates will be improved byaggregating such estimates to a higher geographiclevel. Similarly, collapsing characteristic detail (forexample, combining individual age categories intobroader categories) can allow you to improve the reliability of the aggregate estimate, bringing the CVs to amore acceptable level.Table 4. Example of How to Assess the Reliability of Estimates—Percent of PopulationWho Speak Spanish at HomeCountyEstimateStandard errorCoefficient ofvariationBroward County, FL19.90.21.0Lake County, IL15.90.21.3Source: U.S. Census Bureau, Multiyear Estimates Study data.Table 5. Percent in Poverty by Family Type for Sevier County, TN2000–2004All familiesWith related children under 18 yearsMarried-couple familiesWith related children under 18 yearsFamilies with female householder, no husbandWith related children under 18 years2000–20042002–20042004Total familytypePct. inpovertySEPct. inpovertySEPct. .46.838.313.0Source: U.S. Census Bureau, Multiyear Estimates Study data.A-4 AppendixU.S. Census Bureau, A Compass for Understanding and Using American Community Survey Data

Making ComparisonsAssessing ChangeOften users want to compare the characteristics of onearea to those of another area. These comparisons canbe in the form of rankings or of specific pairs of comparisons. Whenever you want to make a comparisonbetween two different geographic areas you need totake the type of estimate into account. It is importantthat comparisons be made within the same estimatetype. That is, 1-year estimates should only be compared with other 1-year estimates, 3-year estimatesshould only be compared with other 3-year estimates,and 5-year estimates should only be compared withother 5-year estimates.Users are encouraged to make comparisons betweensequential single-year estimates. Specific guidance onmaking these comparisons and interpreting the resultsare provided in Appendix 4. Starting with the 2007ACS, a new data product called the comparison profilewill do much of the statistical work to identify statistically significant differences between the 2007 ACS andthe 2006 ACS.You certainly can compare characteristics for areas withpopulations of 30,000 to areas with populations of100,000 but you should use the data set that they havein common. In this example you could use the 3-yearor the 5-year estimates because they are available forareas of 30,000 and areas of 100,000.As noted earlier, caution is needed when using multiyear estimates for estimating year-to-year changein a particular characteristic. This is because roughlytwo-thirds of the data in a 3-year estimate overlap withthe data in the next year’s 3-year estimate (the overlap is roughly four-fifths for 5-year estimates). Thus,as shown in Figure 1, when comparing 2006–20083-year estimates with 2007–2009 3-year estimates,the differences in overlapping multiyear estimates aredriven by differences in the nonoverlapping years. Adata user interested in comparing 2009 with 2008 willnot be able to isolate those differences using these twosuccessive 3-year estimates. Figure 1 shows that thedifference in these two estimates describes the difference between 2009 and 2006. While the interpretationof this difference is difficult, these comparisons can bemade with caution. Users who are interested in comparing overlapping multiyear period estimates shouldrefer to Appendix 4 for more information.Figure 1. Data Collection Periods for 3–Year an.Dec.2007Jan.Dec.2008Jan.Dec.2009Source: U.S. Census Bureau.Appendix A-5U.S. Census Bureau, A Compass for Understanding and Using American Community Survey Data

Variability in single-year estimates for smaller areas(near the 65,000-publication threshold) and small subgroups within even large areas may limit the ability toexamine trends. For example, single-year estimates fora characteristic with a high CV vary from year to yearbecause of sampling variation obscuring an underlyingtrend. In this case, multiyear estimates may be usefulfor assessing an underlying, long-term trend. Hereagain, however, it must be recognized that because themultiyear estimates have an inherent smoothing, theywill tend to mask rapidly developing changes. Plottingthe multiyear estimates as representing the middleyear is a useful tool to illustrate the smoothing effectof the multiyear weighting methodology. It also canbe used to assess the “lagging effect” in the multiyearestimates. As a general rule, users should not considera multiyear estimate as a proxy for the middle year ofthe period. However, this could be the case under somespecific conditions, as is the case when an area is experiencing growth in a linear trend.As Figure 2 shows, while the single-year estimatesfluctuate from year to year without showing a smoothtrend, the multiyear estimates, which incorporate datafrom multiple years, evidence a much smoother trendacross time.Figure 2. Civilian Veterans, County X Single-Year, Multi