SERPM 8 PopSyn 2015 Population and Household Validation and Calibration RTTAC-MS Marty Milkovits, JJ Zang, Jay Evans, David Kurth 9/19/2017

Revision History 9/19/2017 – revisions to recommendations to only impact household and persons composition, but not spatial locations. Added next steps slide to specify that PopSyn III outputs would be summarized following revised control totals. Original version presented to Project Status Meeting on 9/18/2017

Outline Description of validation data sources County total population and households comparison TAZ control total and ACS data comparison by attributes Next steps and recommendation

Description of validation data sources 2015 TAZ population and household control totals PopSyn III Synthetic population Synthetic population and household data based on TAZ level data. 2015 Bureau of Economic and Business Research (BEBR) Population Estimates 2011-2015 American Community Survey (ACS) 2011-2015 PUMS Micro population/household records using 5% of ACS sample

Total Population County County_Name PopSyn TAZ Control BEBR ACS PopSyn vs. TAZ TAZ vs. BEBR TAZ vs. ACS Palm Beach 1,378,427 1,378,417 1,378,806 0.0% Broward 1,809,698 1,809,705 1,827,367 1,843,152 -1.0% -1.8% Miami-Dade 2,587,387 2,587,396 2,653,934 2,639,042 -2.5% -2.0% Total 5,775,512 5,775,518 5,859,718 5,861,000 -1.4% -1.5% PopSyn outputs match TAZ population control total. Broward and Miami-Dade slightly underestimate population compared to BEBR. BEBR and ACS estimates are close.

Total Households County County_Name PopSyn TAZ Control BEBR ACS PopSyn vs. TAZ TAZ vs. BEBR TAZ vs. ACS Palm Beach 574,954 574,466 566,485 534,605 0.1% 1.4% 7.5% Broward 712,214 711,772 670,284 0.0% 6.2% Miami-Dade 913,874 912,963 914,152 842,153 -0.1% 8.4% Total 2,201,042 2,199,201 2,192,409 2,047,042 0.3% 7.4% PopSyn outputs match TAZ households control total. TAZ control totals is close to BEBR estimates. ACS has fewer households (thus larger average hh size) than TAZ or BEBR

County-Level Control Totals TAZ level controls are consistent with BEBR totals BEBR is considered to be more reliable than ACS for the region No changes recommended to total households or total population BEBR does not include estimates by market segment ACS data is most reliable estimates in region Market segment level analysis should first scale ACS data to TAZ (BEBR) county-level control totals

SERPM8 Market Segments TAZ level household and person attributes (control totals):

TAZ control and ACS Comparison Household Variables ACS county total were scaled to match TAZ control for equal comparison between segmented variables. 2011-2015 ACS household income: IN 2015 INFLATION-ADJUSTED DOLLARS, consistent with TAZ 2015 household income dollars. *ACS total scaled to match TAZ TAZ control overestimate 4-persons households.

TAZ control and ACS Comparison Household Variables ACS county total were scaled to match TAZ control for equal comparison between segmented variables. 2011-2015 ACS household income: IN 2015 INFLATION-ADJUSTED DOLLARS, consistent with TAZ 2015 household income dollars. *ACS total scaled to match TAZ

TAZ control and ACS Comparison Household Variables ACS county total were scaled to match TAZ control for equal comparison between segmented variables. 2011-2015 ACS household income: IN 2015 INFLATION-ADJUSTED DOLLARS, consistent with TAZ 2015 household income dollars. *ACS total scaled to match TAZ TAZ control underestimate lowest and highest income households.

TAZ control and ACS Comparison Person Variables Employment status definition (based on PopSyn Python scripts): Full-time: age>=16 & ESR in (1,2,4,5) & WKHP >=35 & WKW in (1,2,3,4) Part-time: age>=16 & ESR in (1,2,4,5) & WKW >=5 or age>=16 & ESR in (1,2,4,5) & WKHP <35 & WKW in (1,2,3,4) Unemployed: age>=16 & ESR in (3,6) ESR 1 Employment status recode b .N/A (less than 16 years old) 1 .Civilian employed, at work 2 .Civilian employed, with a job but not at work 3 .Unemployed 4 .Armed forces, at work 5 .Armed forces, with a job but not at work 6 .Not in labor force WKW 2 Weeks worked during past 12 months b .N/A (less than 16 years old/did not work .during the past 12 months) 1 .50 to 52 weeks 2 .48 to 49 weeks 3 .40 to 47 weeks 4 .27 to 39 weeks 5 .14 to 26 weeks 6 .13 weeks or less WKHP: Usual hours worked per week past 12 months *ACS total scaled to match TAZ * Full-time, part-time estimates were generated using PUMS data to ensure the consistency of worker definition.

TAZ control and ACS Comparison Person Variables Employment status definition (based on PopSyn Python scripts): Full-time: age>=16 & ESR in (1,2,4,5) & WKHP >=35 & WKW in (1,2,3,4) Part-time: age>=16 & ESR in (1,2,4,5) & WKW >=5 or age>=16 & ESR in (1,2,4,5) & WKHP <35 & WKW in (1,2,3,4) Unemployed: age>=16 & ESR in (3,6) ESR 1 Employment status recode b .N/A (less than 16 years old) 1 .Civilian employed, at work 2 .Civilian employed, with a job but not at work 3 .Unemployed 4 .Armed forces, at work 5 .Armed forces, with a job but not at work 6 .Not in labor force WKW 2 Weeks worked during past 12 months b .N/A (less than 16 years old/did not work .during the past 12 months) 1 .50 to 52 weeks 2 .48 to 49 weeks 3 .40 to 47 weeks 4 .27 to 39 weeks 5 .14 to 26 weeks 6 .13 weeks or less WKHP: Usual hours worked per week past 12 months *ACS total scaled to match TAZ * Full-time, part-time estimates were generated using PUMS data to ensure the consistency of worker definition.

TAZ control and ACS Comparison Person Variables Employment status definition (based on PopSyn Python scripts): Full-time: age>=16 & ESR in (1,2,4,5) & WKHP >=35 & WKW in (1,2,3,4) Part-time: age>=16 & ESR in (1,2,4,5) & WKW >=5 or age>=16 & ESR in (1,2,4,5) & WKHP <35 & WKW in (1,2,3,4) Unemployed: age>=16 & ESR in (3,6) ESR 1 Employment status recode b .N/A (less than 16 years old) 1 .Civilian employed, at work 2 .Civilian employed, with a job but not at work 3 .Unemployed 4 .Armed forces, at work 5 .Armed forces, with a job but not at work 6 .Not in labor force WKW 2 Weeks worked during past 12 months b .N/A (less than 16 years old/did not work .during the past 12 months) 1 .50 to 52 weeks 2 .48 to 49 weeks 3 .40 to 47 weeks 4 .27 to 39 weeks 5 .14 to 26 weeks 6 .13 weeks or less WKHP: Usual hours worked per week past 12 months *ACS total scaled to match TAZ

Spatial Distribution 5-year ACS data is available at the census tract level Market segment differences vary by super-district Household and population totals vary by super district Key question: ACS or TAZ spatial distribution at super district level? ACS or TAZ market segment distribution? Following tables show effective scaling factors Spatial distribution adjustment Market segment adjustment

Total Households Super District TOT_HH (TAZ) TOT_HH (ACS) Difference (%) Scale Factors Palm Beach North 96,119 88,305 8.8% 0.99 Palm Beach CBD 27,517 24,231 13.6% 0.95 Palm Beach Central 184,034 168,762 9.0% Palm Beach West 11,547 12,142 -4.9% 1.13 Palm Beach South 255,249 241,165 5.8% 1.02 Broward North 239,094 223,545 7.0% Broward CBD 32,590 29,581 10.2% 0.96 Broward Central 170,345 159,996 6.5% 1.00 Broward South-West 172,946 167,063 3.5% 1.03 Broward South-East 96,797 90,099 7.4% Miami-Dade North 221,460 206,628 7.2% 1.01 Miami-Dade CBD 118,177 98,063 20.5% 0.90 Miami-Dade NorthWest 118,790 105,989 12.1% 0.97 Miami-Dade Central 236,911 226,083 4.8% Miami-Dade West 96,192 98,528 -2.4% 1.11 Miami-Dade South 121,433 106,862 -TAZ and ACS estimates differ at super district level. -Scale the TAZ estimates to match ACS households distribution among super district within one county.

Total Population Super District TOT_POP (TAZ) TOT_POP (ACS) Difference (%) Scale Factors Palm Beach North 220,761 214,141 3.1% 0.97 Palm Beach CBD 59,707 58,510 2.0% 0.98 Palm Beach Central 492,255 486,568 1.2% 0.99 Palm Beach West 35,652 43,681 -18.4% 1.22 Palm Beach South 570,042 575,906 -1.0% 1.01 Broward North 588,613 598,395 -1.6% 1.00 Broward CBD 64,041 62,789 0.96 Broward Central 426,887 438,176 -2.6% Broward South-West 503,203 515,186 -2.3% Broward South-East 226,961 228,606 -0.7% Miami-Dade North 635,078 645,805 -1.7% Miami-Dade CBD 230,521 216,276 6.6% 0.92 Miami-Dade NorthWest 370,457 356,664 3.9% 0.94 Miami-Dade Central 648,600 674,802 -3.9% 1.02 Miami-Dade West 309,613 360,229 -14.1% 1.14 Miami-Dade South 393,127 385,266

Household Size Adjustments Super District HHSIZE_1 HHSIZE_2 HHSIZE_3 HHSIZE_4PLUS Palm Beach North 1.02 1.03 0.90 Palm Beach CBD 0.99 1.06 0.87 Palm Beach Central 1.01 0.93 Palm Beach West 1.07 1.11 Palm Beach South 0.94 Broward North Broward CBD 0.88 Broward Central 1.04 0.98 0.92 Broward South-West 1.08 Broward South-East Miami-Dade North 1.13 1.05 0.84 Miami-Dade CBD 0.97 1.16 Miami-Dade North West 1.21 Miami-Dade Central 1.10 1.00 Miami-Dade West Miami-Dade South Total 0.91

Household Income Adjustments Super District INCOME_25K INCOME_50K INCOME_75K INCOME_100K INCOME_100KPLUS Palm Beach North 1.03 1.00 0.88 0.97 1.08 Palm Beach CBD 0.92 1.20 0.99 0.98 Palm Beach Central 1.09 1.02 0.90 Palm Beach West 0.86 1.05 1.15 1.39 Palm Beach South 0.94 Broward North 1.04 Broward CBD 1.13 0.87 1.06 Broward Central 0.96 Broward South-West 0.95 Broward South-East 0.93 1.16 Miami-Dade North 0.91 Miami-Dade CBD 1.01 Miami-Dade North West 1.18 Miami-Dade Central Miami-Dade West Miami-Dade South 1.17 Total Scale TAZ estimates to match ACS market segments distribution for each super district.

Number of Workers Adjustments Super District WORKERS_3PLUS Palm Beach North 0.98 0.97 1.09 0.94 Palm Beach CBD 0.96 1.01 1.06 1.02 Palm Beach Central 0.99 Palm Beach West 0.86 1.19 Palm Beach South Broward North 1.00 1.04 Broward CBD 1.13 0.82 Broward Central 1.07 Broward South-West Broward South-East 1.12 Miami-Dade North 1.10 1.05 0.88 0.77 Miami-Dade CBD 1.03 0.87 Miami-Dade North West 1.22 Miami-Dade Central 0.93 Miami-Dade West 1.20 1.08 Miami-Dade South 1.16 Total 0.95

Presence of Children Adjustments Super District CHILDREN_1PLUS Palm Beach North 1.02 0.95 Palm Beach CBD 0.93 Palm Beach Central 1.01 0.98 Palm Beach West 1.08 0.90 Palm Beach South 1.00 Broward North Broward CBD Broward Central 0.97 Broward South-West 1.04 0.94 Broward South-East Miami-Dade North 1.03 Miami-Dade CBD 1.37 Miami-Dade North West Miami-Dade Central 0.99 Miami-Dade West Miami-Dade South Total

Age Adjustments Super District AGE0TO17 AGE18TO24 AGE25TO34 AGE35TO49 AGE50TO64 AGE65TO79 AGE80PLUS Palm Beach North 0.95 0.91 1.07 1.02 1.04 1.16 Palm Beach CBD 0.93 1.11 0.89 0.99 1.03 1.13 Palm Beach Central 0.97 0.94 1.12 Palm Beach West 0.79 0.96 1.19 1.14 Palm Beach South 0.98 1.01 1.00 Broward North 1.05 Broward CBD 1.06 0.77 0.92 1.26 Broward Central 1.08 Broward South-West 1.09 1.10 Broward South-East 0.90 Miami-Dade North Miami-Dade CBD 1.18 Miami-Dade NorthWest Miami-Dade Central Miami-Dade West Miami-Dade South Total

Gender Adjustments Super District MALE FEMALE Palm Beach North 0.99 1.01 Palm Beach CBD Palm Beach Central Palm Beach West 1.12 0.89 Palm Beach South Broward North Broward CBD 1.00 Broward Central Broward South-West Broward South-East Miami-Dade North Miami-Dade CBD Miami-Dade North West Miami-Dade Central Miami-Dade West 1.03 0.97 Miami-Dade South Total

Recommendation Maintain TAZ total household and total person control totals as specified by the T/MPOs T/MPO spatial distribution is more reliable than ACS. Adjust TAZ control totals to match ACS distribution pattern by market segment at super district level ACS market segment distribution is more reliable. Scale factors presented above are preliminary An iterative proportional fitting process is necessary to derive final factors

Next Steps Derive balanced scale factors for household and person compositions Apply scale factors and compare PopSynIII outputs to control totals and ACS Identify segments where revised control totals are not well matched due to correlations in sample population Revise control totals to compensate as necessary