Article Figures & Data
Tables
U.S. IL and MD IL MD Reported food stamp receipt rate 5.89% 4.73% 5.50% 2.97% Reported average received ($) 108.05 94.66 111.87 54.87 Age under 18 0.17% 0.17% 0.18% 0.14% Age 18–39 33.02% 33.71% 34.79% 31.20% Age 40–64 46.33% 45.61% 44.42% 48.36% Education less than high school 15.37% 13.35% 13.81% 12.28% Education high school 30.43% 30.01% 31.10% 27.49% Education some college 26.80% 25.09% 26.32% 22.24% Black 9.18% 13.51% 10.64% 20.14% Other nonwhite 28.82% 28.30% 28.08% 28.81% Hispanic 7.89% 5.63% 6.59% 3.42% Poverty rate 13.40% 11.80% 12.71% 9.67% Notes: Receipt and poverty rates are in terms of households, demographics in terms of household heads. Average received is unconditional annual dollars per household. For the demographics, the omitted category is “65 and older” for age, “college degree or more” for education, and “white” for ethnicity. All estimates use sampling weights and are based on the entire 2002–2005 CPS sample (and therefore slightly differ from the estimated receipt rates in Table 1).
Unweighted Observation Counts Receipt Rate False Negative Rate False Positive Rate Recipient Nonrecipients Report Admin. Report Admin. Report Admin. (1) (2) (3) (4) (5) (6) (7) (8) Panel A: Full Sample ACS 1,145 1,508 29,156 28,793 5.69% 7.49% 33.08% 0.73% CPS 640 1,104 10,275 9,811 5.20% 8.69% 48.98% 0.84% SIPP 817 793 10,138 10,162 6.14% 5.95% 22.82% 1.64% Panel B: Imputed Observations ACS 256 219 215 252 64.93% 53.39% 3.21% 28.44% CPS 62 108 331 285 12.07% 22.54% 79.63% 9.65% SIPP 130 108 712 734 14.20% 11.00% 28.83% 7.16% Notes: The false negative rate is the fraction of true recipient households with receipt not recorded in the survey. The false positive rate is the fraction of true nonrecipient households recorded as recipients in the survey. In the SIPP, we collapse receipt to the wave level. Columns 5–8 use household weights adjusted for incomplete linkage. The (weighted) imputation rates are 1.9, 3.6, and 7.3 percent in the ACS, CPS, and SIPP, respectively.
- Table 3
Determinants of False Negatives, Probit Average Derivatives, Households with Income Less Than Twice the Poverty Line
ACS CPS SIPP Illinois Maryland Illinois Maryland IL&MD Single, no children –0.0862
(0.0716)
0.0437
(0.0877)
–0.1312*
(0.0779)
0.0558
(0.1755)
–0.1644**
(0.0662)
Single, with children –0.0802
(0.0539)
0.1203
(0.0753)
–0.0227
(0.0620)
–0.0323
(0.1203)
–0.1426***
(0.0481)
Multiple adults, no children –0.1036
(0.0857)
–0.0135
(0.1067)
–0.0245
(0.0739)
0.0668
(0.1416)
0.0323
(0.0843)
Number of members 18 or older –0.0248
(0.0341)
0.0405
(0.0363)
0.0391
(0.0371)
0.0370
(0.0794)
0.0159
(0.0210)
Number of members under 18 –0.0306
(0.0264)
–0.0185
(0.0329)
–0.0230
(0.0224)
–0.0968
(0.0616)
–0.0361*
(0.0219)
Number of members PIKed 0.0308
(0.0268)
0.0358
(0.0333)
–0.0171
(0.0194)
0.0484
(0.0433)
–0.0060
(0.0205)
Age ≥50 0.1514***
(0.0513)
0.1319**
(0.0663)
0.0881*
(0.0525)
–0.1418*
(0.0832)
–0.0044
(0.0347)
Male 0.0877*
(0.0356)
–0.0335
(0.0483)
–0.0603
(0.0446)
0.0195
(0.0858)
–0.0798**
(0.0359)
Less than high school 0.0688
(0.0431)
0.0659
(0.0589)
–0.0695
(0.0479)
–0.0620
(0.1111)
–0.1572***
(0.0517)
High school graduate –0.0001
(0.0425)
0.1147**
(0.0576)
–0.0293
(0.0463)
–0.0002
(0.0926)
0.0425
(0.0327)
College graduate and beyond 0.2197***
(0.0745)
–0.0586
(0.1201)
0.0373
(0.1103)
–0.0295
(0.1223)
–0.0460
(0.0463)
White –0.0897**
(0.0368)
–0.1110***
(0.0422)
–0.0503
(0.0415)
–0.0509
(0.0810)
–0.0672*
(0.0355)
Employed 0.0428
(0.0273)
Unemployed –0.0206
(0.0554)
–0.2504***
(0.0668)
0.0396
(0.0664)
0.0235
(0.1532)
Not in labor force –0.0077
(0.0404)
–0.0627
(0.0513)
0.0199
(0.0447)
–0.0074
(0.0832)
Income divided by poverty line 0.0010***
(0.0003)
0.0008**
(0.0004)
0.0010**
(0.0004)
–0.0003
(0.0008)
–0.0082
(0.0304)
Disabled –0.0637*
(0.0386)
–0.0333
(0.0584)
n.d.
n.d.
0.0249
(0.0280)
Disabled, not working –0.0382
(0.0465)
0.1179**
(0.0505)
Speaks English only 0.0455
(0.0507)
–0.1448*
(0.0838)
Speaks no or poor English 0.1690***
(0.0454)
Non-U. S. citizen –0.1545***
(0.0327)
0.0697
(0.1011)
–0.2039*
(0.1162)
Rural –0.1000**
(0.0472)
–0.1079**
(0.0476)
–0.0276
(0.0548)
–0.0684
(0.1346)
0.0145
(0.0542)
Reported public assistance receipt –0.2693***
(0.0549)
–0.2453***
(0.0632)
–0.3293***
(0.0722)
n.d.
–0.2283***
(0.0749)
Reported housing assistance receipt –0.0336
(0.0397)
–0.0248
(0.0481)
–0.1753***
(0.0409)
–0.2732***
(0.0871)
–0.1180***
(0.0305)
Food Stamp receipt imputed –0.3115***
(0.0647)
–0.3833***
(0.0899)
0.3580***
(0.0552)
0.1932*
(0.1103)
–0.1029**
(0.0525)
Length of Food Stamp receipt spell –0.0275***
(0.0034)
–0.0384***
(0.0036)
–0.0281***
(0.0051)
–0.0196**
(0.0086)
–0.0452***
(0.0158)
Administrative TANF receipt 0.0658
(0.0446)
0.0273
(0.0514)
0.0986*
(0.0580)
0.2466***
(0.0766)
0.0812
(0.0676)
Linear time trend 0.0222
(0.0157)
0.0980***
(0.0373)
Household had bad data record 0.0936**
(0.0450)
Months since last admin. Food Stamp receipt 0.0414***
(0.0132)
Household in MD 0.0893**
(0.0442)
Observations 789 344 689 136 640 ↵Notes: The unreported omitted family type is multiple adults with children, the education category is some college, the employment category is employed, the race group is nonwhite, and the geographic area is within-MSA. n.d. indicates that the variable was controlled for, but the estimate was not disclosed by the Census Bureau. In the ACS, we also controls for mode of interview (mail-back, CATI, CAPI). For the CPS, samples are pooled across all years (2002–2005). In the SIPP, we also control for year dummies, number of members interviewed, whether there was no interview with the reference person or an interview with a household member with no PIK. “Household had a bad data record” indicates that the reference person refused to answer many related survey questions. For the SIPP, samples include both states, are collapsed to the wave level, and are pooled across all years (IL: 10/2000–10/2004, MD: 10/2000–12/2003). All analyses conducted using household weights adjusted for PIK probability. Standard errors in parentheses. *p<0.1, **p<0.05, ***p<0.01.
- Table 4
Determinants of False Positives, Probit Average Derivatives, Households with Income Less than Twice the Poverty Line
ACS CPS SIPP Illinois Maryland Illinois Maryland IL&MD Single, no children 0.0026
(0.0177)
Single, with children 0.0214*
(0.0124)
Multiple adults, no children –0.0337*
(0.0174)
Number of members 18 or older –0.0024
(0.0034)
0.0053
(0.0050)
0.0092
(0.0067)
–0.0170
(0.0130)
0.0130**
(0.0061)
Number of members under 18 0.0069**
(0.0033)
–0.0020
(0.0036)
0.0044
(0.0049)
–0.0251**
(0.0120)
0.0142**
(0.0065)
Number of members PIKed –0.0085**
(0.0038)
0.0060
(0.0040)
–0.0047
(0.0044)
0.0222*
(0.0118)
–0.0117**
(0.0048)
Age ≥50 –0.0225***
(0.0103)
–0.0063
(0.0075)
–0.0382***
(0.0086)
–0.0010
(0.0147)
–0.0100
(0.0109)
Male –0.0106*
(0.0061)
0.0032
(0.0080)
–0.0130
(0.0104)
0.0106
(0.0094)
–0.0032
(0.0084)
Less than high school 0.0140**
(0.0068)
0.0063
(0.0099)
0.0193
(0.0134)
n.d.
0.0091
(0.0089)
High school graduate –0.0032
(0.0085)
0.0111
(0.0126)
–0.0001
(0.0117)
0.0008
(0.0079)
0.0086
(0.0081)
College graduate and beyond –0.0051
(0.0127)
White –0.0239***
(0.0071)
–0.0082
(0.0083)
0.0046
(0.0098)
0.0094
(0.0096)
–0.0207**
(0.0083)
Employed –0.0054
(0.0066)
–0.0261*
(0.0151)
–0.0016
(0.0117)
0.0012
(0.0089)
–0.0003
(0.0070)
Income divided by poverty line –0.0001***
(0.0000)
–0.0001
(0.0001)
–0.0003***
(0.0001)
–0.0002**
(0.0001)
–0.0257***
(0.0069)
Disabled 0.0076
(0.0084)
–0.0069
(0.0084)
n.d.
n.d.
0.0441***
(0.0073)
Disabled, not working 0.0159*
(0.0082)
0.0226**
(0.0097)
Speaks no or poor English 0.0027
(0.0125)
Non-U.S. citizen –0.0129
(0.0132)
Rural –0.0051
(0.0088)
n.d.
0.0120
(0.0095)
Reported public assistance receipt 0.0442***
(0.0091)
0.0622***
(0.0186)
0.0957***
(0.0197)
0.0872***
(0.0332)
0.1020***
(0.0210)
Reported housing assistance receipt 0.0108
(0.0070)
0.0007
(0.0081)
0.0571***
(0.0146)
–0.0032
(0.0116)
0.0523***
(0.0089)
Food Stamp receipt imputed 0.0700***
(0.0110)
0.0447***
(0.0139)
0.0544***
(0.0113)
0.0443***
(0.0156)
0.0470***
(0.0130)
Administrative TANF receipt –0.1093***
(0.0422)
Linear time trend 0.0018
(0.0047)
0.0000
(0.0056)
Household in MD –0.0309*
(0.0163)
Observations 3,357 1,455 1,462 504 2,333 ↵Notes: The unreported omitted education category for the false negative probits is some college or more, the race group is nonwhite, the employment category is not employed, and the geographic area is within-MSA. n.d. indicates that the variable was controlled for, but the estimate was not disclosed by the Census Bureau. In the ACS, we also controls for mode of interview (mail-back, CATI, CAPI). For the CPS, samples are pooled across all years (2002–2005). In the SIPP, we also control for year dummies, number of members interviewed, whether there was no interview with the reference person or an interview with a household member with no PIK, and whether the reference person had a bad data record (refused to answer many related survey questions). For the SIPP, samples include both states, are collapsed to the wave level, and are pooled across all years (IL: 10/2000–10/2004, MD: 10/2000–12/2003). All analyses conducted using household weights adjusted for PIK probability. Standard errors in parentheses. *p< 0.1, **p<0.05, ***p<0.01.
- Table 5
Determinants of Reported and Administrative Food Stamp Receipt, Probit Marginal Effects, Linked Households with Income Less than Twice the Poverty Line
ACS CPS SIPP Illinois Maryland Illinois Maryland IL&MD (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Dependent variable Survey
Report
Admin.
Receipt
Survey
Report
Admin.
Receipt
Survey
Report
Admin.
Receipt
Survey
Report
Admin.
Receipt
Survey
Report
Admin.
Receipt
Single, no children 0.0670
(0.0320)
0.1164a
(0.0361)
0.0861
(0.0461)
0.1485
(0.0515)
–0.0119
(0.0256)
0.0001
(0.0386)
–0.0687
(0.0511)
–0.0229
(0.0623)
0.0505
(0.0223)
0.0329
(0.0278)
Single, with children 0.1076
(0.0247)
0.1429a
(0.0272)
0.1083
(0.0351)
0.1965b
(0.0389)
0.0547
(0.0214)
0.1333b
(0.0308)
0.0133
(0.0437)
0.0775
(0.0491)
0.1262
(0.0149)
0.1280
(0.0185)
Multiple adults, no children 0.0696
(0.0344)
0.0959
(0.0392)
0.0547
(0.0500)
0.0975
(0.0547)
0.0192
(0.0226)
0.0664
(0.0346)
–0.0509
(0.0413)
0.0235
(0.0560)
0.0186
(0.0245)
–0.0064
(0.0307)
Number of members under 18 0.0188
(0.0099)
–0.0066b
(0.0145)
0.0202
(0.0144)
0.0027
(0.0191)
0.0227
(0.0058)
0.0309
(0.0087)
0.0235
(0.0117)
0.0541a
(0.0181)
0.0199
(0.0065)
0.0052
(0.0113)
Number of members 18 or older 0.0027
(0.0111)
–0.0201a
(0.0138)
0.0039
(0.0174)
0.0153
(0.0208)
–0.0069
(0.0104)
0.0128
(0.0143)
–0.0213
(0.0258)
0.0055
(0.0246)
0.0155
(0.0074)
–0.0121c
(0.0096)
Number of members PIKed 0.0145
(0.0076)
0.0692c
(0.0131)
0.0165
(0.0118)
0.0612c
(0.0183)
0.0048
(0.0036)
0.0281c
(0.0069)
Age 16–29 –0.0208
(0.0231)
–0.0055
(0.0264)
0.0274
(0.0300)
0.0141
(0.0332)
–0.0111
(0.0198)
–0.0378
(0.0291)
–0.0086
(0.0287)
–0.0428
(0.0431)
0.0298
(0.0159)
0.0210
(0.0205)
Age 30–39 0.0061
(0.0221)
0.0061
(0.0262)
–0.0386
(0.0288)
–0.0454
(0.0323)
–0.0118
(0.0194)
0.0040
(0.0280)
–0.0285
(0.0257)
–0.0043
(0.0419)
0.0338
(0.0150)
–0.0151c
(0.0188)
Age 50–59 –0.0981
(0.0261)
–0.0405b
(0.0294)
–0.0315
(0.0366)
–0.0375
(0.0369)
0.0016
(0.0228)
0.0287
(0.0369)
0.0249
(0.0291)
0.0382
(0.0461)
0.0112
(0.0171)
–0.0269b
(0.0190)
Age 60–69 –0.1144
(0.0278)
–0.0806
(0.0320)
–0.0856
(0.0358)
–0.0702
(0.0384)
–0.0110
(0.0240)
–0.0625
(0.0353)
0.0372
(0.0344)
–0.0052
(0.0519)
–0.0039
(0.0175)
–0.0075
(0.0219)
Age ≥70 –0.1641
(0.0278)
–0.1619
(0.0321)
–0.1346
(0.0359)
–0.1354
(0.0386)
–0.1313
(0.0254)
–0.1579
(0.0352)
–0.0714
(0.0353)
–0.1675a
(0.0599)
0.0080
(0.0174)
0.0316
(0.0206)
Less than high school 0.0648
(0.0184)
0.0687
(0.0218)
0.0739
(0.0237)
0.1089a
(0.0271)
0.0503
(0.0165)
0.0455
(0.0248)
–0.0056
(0.0262)
0.0073
(0.0405)
0.0100
(0.0122)
0.0077
(0.0155)
High school graduate 0.0239
(0.0186)
0.0318
(0.0212)
0.0130
(0.0232)
0.0510
(0.0255)
0.0266
(0.0158)
0.0409
(0.0236)
0.0031
(0.0241)
–0.0085
(0.0360)
0.0137
(0.0114)
0.0387a
(0.0139)
College graduate and beyond –0.0584
(0.0313)
–0.0569
(0.0329)
0.0114
(0.0361)
–0.0147
(0.0407)
–0.0892
(0.0267)
–0.1557
(0.0442)
0.0191
(0.0300)
–0.0420
(0.0510)
–0.0223
(0.0152)
0.0004
(0.0187)
White –0.0380
(0.0178)
–0.0801c
(0.0191)
0.0055
(0.0187)
–0.0355b
(0.0211)
–0.0211
(0.0133)
–0.0762c
(0.0196)
0.0048
(0.0182)
–0.0118
(0.0261)
–0.0767
(0.0100)
–0.1069c
(0.0112)
Employed –0.0380
(0.0164)
–0.0217
(0.0188)
–0.0488
(0.0227)
–0.0078a
(0.0247)
–0.0399
(0.0141)
–0.0665
(0.0207)
–0.0391
(0.0191)
–0.0633
(0.0280)
–0.0146
(0.0092)
–0.0275
(0.0112)
Income divided by poverty line –0.0007
(0.0100)
–0.0007
(0.0001)
–0.0010
(0.0001)
–0.0013b
(0.0001)
–0.0009
(0.0002)
–0.0015c
(0.0001)
–0.0003
(0.0002)
–0.0003
(0.0001)
–0.0841
(0.0002)
–0.1028a
(0.0083)
Disabled 0.0906
(0.0182)
0.0774
(0.0209)
0.0773
(0.0235)
0.0933
(0.0249)
0.0466
(0.0451)
0.0377
(0.0719)
0.1046
(0.0629)
0.0022a
(0.0867)
0.1220
(0.0116)
0.1550b
(0.0143)
Disabled, not working 0.0271
(0.0193)
0.0086
(0.0224)
0.0093
(0.0242)
0.0465
(0.0266)
Speaks English only 0.0343
(0.0207)
0.0850c
(0.0245)
0.0716
(0.0306)
0.0772
(0.0393)
Speaks no or poor English 0.0393
(0.0138)
0.0814c
(0.0167)
Rural 0.0293
(0.0191)
0.0458
(0.0189)
0.0499
(0.0183)
0.0491
(0.0225)
0.0275
(0.0167)
0.0383
(0.0262)
0.0495
(0.0278)
0.0682
(0.0388)
0.0386
(0.0122)
0.0384
(0.0160)
Reported public assistance receipt 0.3189
(0.0240)
0.2386b
(0.0315)
0.3020
(0.0324)
0.3728
(0.0408)
0.2179
(0.0268)
0.2077
(0.0432)
0.1934
(0.0327)
0.2246
(0.0590)
0.1676
(0.0244)
0.0883c
(0.0262)
Reported housing assistance receipt 0.1461
(0.0184)
0.1811b
(0.0217)
0.1021
(0.0198)
0.1337
(0.0241)
0.1517
(0.0147)
0.1999
(0.0243)
0.1378
(0.0221)
0.1593
(0.0364)
0.1522
(0.0116)
0.1355
(0.0132)
Linear time trend 0.0039
(0.0053)
0.0180a
(0.0079)
–0.0002
(0.0096)
0.0329b
(0.0164)
2001 –0.0373
(0.0124)
–0.0687b
(0.0152)
2002 –0.0360
(0.0133)
–0.0517
(0.0164)
2003 –0.0228
(0.0137)
–0.0581b
(0.0180)
2005 0.0188
(0.0167)
–0.0207c
(0.0195)
Household in MD –0.0493
(0.0146)
–0.0008c
(0.0161)
Observations 4,591 4,146 1,945 1,799 2,981 2,151 808 640 4,177 2,973 p-value admin equal survey (jointly) <0.0001 0.0004 <0.0001 0.0085 <0.0001 ↵Notes: For each survey, the first column contains estimates from a probit model using reported receipt as the dependent variable. The second column estimates the same model using the administrative receipt measure as the dependent variable. Superscript a, b, and c in these columns indicate that a χ2-test whether the survey and administrative estimates are equal rejects at the 10, 5, and 1 percent level, respectively. Standard errors of the estimates are in parentheses. All analyses conducted using household weights adjusted for PIK probability. All demographic characteristics refer to the reference person. The unreported omitted family type is multiple adults with children, the age group is 40–49, the education group is some college, the race group is nonwhite, the employment group is not employed, and the geographic area is within MSA. For the CPS, samples are pooled across all years (2002–2005). For the SIPP, samples include both states, are collapsed to the wave level, and are pooled across all years (IL: 10/2000–10/2004, MD: 10/2000–12/2003).
Additional Files
Free alternate access to The Journal of Human Resources supplementary materials is available at https://uwpress.wisc.edu/journals/journals/jhr-supplementary.html
Jump to section
- Article
- ABSTRACT
- I. Introduction
- II. Misreporting in Survey Data
- III. Data and Linkage
- IV. Agreement between Survey and Administrative Reports
- V. What Affects the Agreement between the Survey Reports and the Administrative Records?
- VI. The Effect of Survey Error on Estimates of Program Receipt
- VII. Conclusions
- Footnotes
- References
- Figures & Data
- Supplemental
- Info & Metrics
- References