National Academies Press: OpenBook

Estimating Eligibility and Participation for the WIC Program: Final Report (2003)

Chapter: 9. Options for Estimating Eligibility and Participation

« Previous: 8. Estimating WIC Participation Among Eligible People
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 133
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 134
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 135
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 136
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 137
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 138
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 139
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 140
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 141
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 142
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 143
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 144
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 145
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 146
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 147
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 148
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 149
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 150
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 151
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 152
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 153
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 154
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 155
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 156
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 157
Suggested Citation:"9. Options for Estimating Eligibility and Participation." National Research Council. 2003. Estimating Eligibility and Participation for the WIC Program: Final Report. Washington, DC: The National Academies Press. doi: 10.17226/10804.
×
Page 158

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

9 Options for Estimating Eligibility and Participation The panel was asked to evaluate current methods used to estimate WIC eligibility and participation and to offer recommendations to im- prove these estimates. This chapter presents two estimation strategies that summarize our recommendations for ways to improve estimates of the numbers of eligible individuals, as well as an approach to predict the num- ber of participants. ALTERNATIVE STRATEGIES FOR PREDICTING ELIGIBILITY Estimating the number of eligible individuals is central to prediction of the number of WIC participants. The estimates are needed to make budget requests and to calculate coverage rates. The panel proposes two options for estimating of the number of WIC-eligible individuals. The first option continues to employ the Current Population Survey (CPS) for an- nual estimates of the eligible population but recommends ways to improve its use. The second option employs the Survey of Income and Program Participation (SIPP) to estimate WIC eligibility. The CPS Option The CPS has both advantages and disadvantages for estimating eligi- bility. The chief advantage is the regular and quick release of the CPS from 133

134 ESTIMATING ELIGIBILITYAND PAR TICIPATION FOR THE TIC PROGRAM the Census Bureau for public use. The data are collected in March, and the public use file is regularly released in the fall of the same year. The CPS has a relatively large sample that ensures adequate sampling rates for infants and children. It has numerous questions pertaining to income sources and participation in other government programs. Given its long history, there is a good deal of experience with the survey. The survey design also makes it relatively easy to access and use. However, as the report has discussed, the CPS has flaws in terms of its ability to estimate eligibility for WIC. The primary disadvantage is that it does not measure monthly income flows for the family. As Chapter 5 documented, the use of annual income instead of the conceptually more appropriate monthly income results in substantial underestimation of the numbers of eligible infants and children. A second disadvantage of the CPS is that it is impossible to identify, even indirectly, women who are pregnant. Moreover, the CPS does not identify which household members receive WIC benefits. If the March CPS is used to prepare eligibility and participation esti- mates, it is the panel's view that improvements can be made to the current USDA methodology. The improvements offer a better use of the data con- tained in the CPS and an approach to account for the lack of monthly data in the CPS. Currently, USDA uses the following information from the CPS to de- . . .. .. ... termlne Income ellglblllty: · Age (used to identify infants and children), · Family relationship (used to identify foster children), 1 · Size of census family and state of residence (used to determine the appropriate poverty threshold),2 and Family income (used to determine whether the individual meets the income eligibility limits of the program). However, the CPS contains significantly more data on individuals that could be used to improve the determination of eligibility. In the case of infants and children, the CPS contains reported participation in the means- 1Foster children are assumed to be WIC eligible regardless of the income of the foster family. 2A census family is defined to be all individuals who live together in a household who are related by blood or marriage.

OPTIONS FOR ESTIMATING ELIGIBILII~YAND PARTICIPATION 135 tested programs (Temporary Assistance for Needy Families or TANF, food stamps, and Medicaid) used to determine adjunctive eligibility for WIC. Underreporting of program participation is a problem, as reported partici- pation is less in the CPS than the number of participants reported from administrative records. Despite this shortcoming, using reported program participation to account for those adjunctively eligible is preferable to the very small adjustment for adjunctive eligibility that is currently made. Table 9-1 documents that ignoring the reported participation inTANF, food stamps, and Medicaid can have a significant effect on the estimates of the number of infants and children. The row labeled "USDA methodol- ogy" presents the estimates of the annual number of eligible infants and children using the current USDA methodology for calendar years 1994 to 1999. The next row, labeled "Using reported enrollment," continues to use the current USDA methodology but also counts as eligible any infant or child who reported enrollment in TANF, food stamps, or Medicaid. The use of reported enrollment in means-tested programs provides a direct method to identify WIC eligible infants and children who are adjunctively eligible. For infants, the impact of the use of reported enrollment has TABLE 9-1 Current Population Survey (CPS) and Transfer Income Microsimulation Model (TRIM) Estimates of Eligibility Year 1994 1995 1996 1997 1998 1999 Infants USDA methodology 1,628 1,669 1,620 1,543 1,492 1,470 Using reported enrollment 1,867 1,905 1,931 1,817 1,777 1,799 TRIM imputed CPS 2,231 NA 2,357 2,170 2,130 2,133 Multipliera 1.195 NA 1.221 1.194 1.200 1.186 Children USDA methodology 7,350 6,963 6,893 6,813 6,375 6,076 Using reported enrollment 8,407 7,560 7,890 7,486 7,263 7,173 TRIM imputed CPS 8,701 NA 8,341 7,821 7,678 7,398 Multipliera 1.035 NA 1.057 1.045 1.057 1.031 Multiplier computed as "TRIM imputed CPS" estimate divided by "Using reported enrollment" estimate. The panel did not obtain TRIM data for 1995. Estimates are in 1,OOOs. NA= Not available.

136 ESTIMATING ELIGIBILI>~D PAR TICIPATION FOR THE ~CPROG~ steadily grown from 14 to 22 percent during this six-year period. For chil- dren, the percentage increase in the number eligible has been erratic but averages 13 percent. The CPS gives only annual instead of the more program-relevant monthly income. To account for how monthly income and certification periods affect eligibility estimates, we propose that a multiplier (or propor- tional adjustment factor) be developed that could be applied to the esti- mates of the number of eligible infants and children. The panel estimated eligibility for WIC using monthly income for 1994-1999 based on Transfer Income Microsimulation data (TRIM), which imputes CPS reports of annual income into monthly measures. (The TRIM model and the estimates of income-eligible infants and children are discussed in Appendix C.) These estimates were used to assess the size of the impact that a monthly income measure has on estimates of eligibility and how stable that impact is over time. The stability of the multipliers is the major factor in deciding whether to use it. The row labeled "TRIM imputed CPS" in Table 9-1 reports the num- ber of infants and children estimated to be eligible from CPS data that have imputed monthly income created by the TRIM model. The next row in the table, labeled "Multiplier," contains the ratio of the eligibility estimates based on the TRIM model's imputed monthly income relative to the CPS- based estimate using annual income plus those who report participation in means-tested programs (row labeled "Using reported enrollment". This multiplier is intended to adjust the annual estimates of eligibility to ac- count for variation in monthly income. This adjustment factor appears to be quite stable for both infants and children for the five years reported. The multiplier for infants ranges between 1.19 and 1.22 percent. Averaged over all the years, the multiplier is 1.20. The multiplier for children ranges from 1.03 to 1.06 and is, on average, 1.05.3 3The TRIM model estimates based on the CPS cannot fully account for the impact that certification periods will have on eligibility estimates. With the TRIM data, it is possible only to partially simulate the role that certification periods play in the eligibility process over the course of the year for children. In this case, children are assumed to have 6 months of WIC eligibility if they have 1 to 7 months of income that fall below 185 percent of the federal poverty guidelines. Otherwise they are simulated to have 12 months of eligibility during the year. Infants are assumed to have 12 months of eligibility if they have at least 1 month of income less than 185 percent of poverty.

OPTIONS FOR ESTIMATING ELIGIBILII~YAND PARTICIPATION 137 While this multiplier would address some of the major shortcomings presented by the use of the annual CPS data, it is far from a perfect solu- tion. Although the TRIM model is routinely used by some government agencies to analyze other transfer programs aimed at the low-income popu- lation such as TANF and Medicaid, the validity of the imputed monthly income amounts have not been recently examined.4 Reliance on imputed monthly income amounts may produce unreliable estimates of the appro- priate multiplier. The stability and accuracy of these two multipliers would need to be assessed periodically.5 As discussed in Chapter 5, the SIPP data provide a more reliable source of information on monthly income, and hence the SIPP data would be a preferable data source to construct a multiplier to be applied to the base CPS estimates of income eligibility estimated from annual income and re- ported enrollment in means-tested programs.6 If SIPP is used to construct the multiplier, it is important that estimates appropriately account for WIC certification periods and include those infants and children who would be eligible based on their annual income and reported participation in means- tested programs. Because the panel had only two years of SIPP data with which to con- struct a multiplier, its stability could not be examined. If USDA decides to use SIPP to create this multiplier, it should be examined more fully. 4The last formal evaluation of the TRIM model's imputation of monthly income was performed in 1990. The results of this evaluation are reported in Long (1990). After this evaluation, several modifications to the imputation procedures were adopted that appear in the current version of the TRIM model. While TRIM is operated and maintained by the Urban Institute through contracts with the U.S. Department of Health and Human Services (DHHS), TRIM data files are publicly available through the Urban Institute web site free of charge. Given the amount of data imputation and the fact that the public release of the data must be approved by DHHS, the release of these files occurs much later than the release of the March CPS file for that same year. However, for the purpose of checking the stability of the multiplier for monthly income, certification periods, and adjunctive eligibility, the delay in the release of the TRIM data should not pose a problem for this purpose. 6For this option, the panel is proposing that the March CPS continue to be utilized to produce annual estimates of WIC eligibility to which a multiplier based on an analysis of the SIPP data would be applied. The second option proposed is based on the yearly use of SIPP data, to which no multiplier would be applied.

138 ESTIMATING ELIGIBILITYANDPARTICIPATIONFOR THE WICPROGRAM Regardless of whether this multiplier is produced from the TRIM im- puted CPS data or SIPP data, it should be examined at least once every five years to determine whether its use continues to be appropriate. The amount of month-to-month income variability could increase or decrease, render- ing the constant multiplier as inaccurately reflecting current income dy- namics. Furthermore, the multiplier could also lead to inaccuracies in esti- mation if rules for means-tested programs, especially Medicaid, change. In the case of pregnant women, the proposed CPS option partially accounts for the differences in income prior to the birth of the child and during the postpartum period. The current USDA methodology assumes that if an infant is income eligible for WIC, then the mother would have been income eligible during her pregnancy. However, in Chapter 6, the panel cites evidence that income variability during pregnancy reduces the percentages of women who are income eligible during pregnancy. The panel proposed lowering the adjustment factor of 0.75 that reflects that a preg- nant mother is, at most, eligible for nine months (.75 = 9/12) to 0.533 to account for the differences in income variability during pregnancy and the first year postpartum. However, the panel did not examine whether poten- tial differences in the impact of adjunctive eligibility for pregnant and post- partum women would significantly differ from those of infants. Accounting for monthly income and adjunctive eligibility are high pri- orities for improving CPS-based estimates of eligibility. In this report, the panel recommends several additional adjustments to current methods: · Adjust the CPS weights for the undercount of infants and overcount of children (Chapter 41. To estimate the number of income-eligible postpartum women from CPS-based estimates (both breastEeeding and nonbreastEeeding), continue to use the current adjustment factor of 0.9844 to account for multiple births and infant and fetal deaths (Chapter 61. To obtain the number of income-eligible pregnant women, apply an adjustment of 0.533 (instead of the 0.75 factor) to the number of income-eligible infants (Chapter 61. Use more recent data to estimate breastSeeding rates and duration among income-eligible women less than 12 months postpartum. Apply them to the estimates of income-eligible postpartum women to determine the number breastfeeding and nonbreastfeeding (Chapter 61.

OPTIONS FOR ESTIMATING ELIGIBILII~YAND PARTICIPATION 139 · If the nutritional risk screen is no longer used to determine WIC eligibility, then no adjustment to account for the percentage of the income-eligible population that is at nutritional risk should be made (Chapter 71.7 Box 9-1 compares the current ancl proposed methods for estimating eligi- bility for infants ancl chilclren. Box 9-2 makes the same comparison for pregnant women ancl Box 9-3 for postpartum women. The CPS option attempts to overcome data deficiencies by using con- stant adjustment factors. The accuracy of these adjustments may decline over time. Furthermore, the method implicitly assumes that the multiplier for income variability ancl for acljunctive eligibility for infants applies uni- formly to the variability ancl acljunctive eligibility of pregnant ancl postpar- tum women (e.g., the effects of income variability on estimates of eligible infants is the same as the effects of income variability on eligibility esti- mates for pregnant ancl postpartum women). A preferable option is to use more appropriate data so that adjustment factors would not be neeclecl. As we have previously notecl, SIPP contains many features that are useful for estimating eligibility. The SIPP Option . . A second option for estimating eligibility is to use SIPP data from waves covering the period of time for which eligibility needs to be pre- clictecl. Box 9-4 contains the steps that would be used to estimate eligibility using this SIPP option. Monthly income measures would be employocl to . ..... . . . . .,` . . c Determine ellglulllty In a given month, ant ~ appropriate certl~lcatlon perl- ocls could also be constructecl. Reportecl enrollment in TANF, food stamps, ancl Meclicaicl could also be used to account for acljunctive eligibility. Be- cause of underreporting of program participation, the number acljunctively eligible may still be unclerstatecl. To correct for this, program enrollment, 7If USDA does not drop the nutritional risk screen for determining eligibility, then the panel's lower bound estimates of the prevalence of nutritional risk among the income-eligible population should be used to estimate eligibility. These lower bound estimates are: 100 per- cent for breastfeeding postpartum women, 97 percent for pregnant women, 97 percent for infants, and 99 percent of children ages 2 to 5.

140 ESTIMATING ELIGIBILITYAND PAR TICIPATION FOR THE ~CPROGR~

OPTIONS FOR ESTIMATING ELIGIBILITYAND PARTICIPATION 141 especially Medicaid enrollment, could be imputed to match control totals from administrative data, as the TRIM model does with the CPS.8 The SIPP data provide other advantages over the CPS. Instead of in- 8It should be noted that just because the enrollment counts match administrative totals does not mean that the imputation process correctly assigns participation to the individuals in the survey who indeed participated in the means-tested programs but did not report participation. In particular, errors in the imputation process could assign participation to too many or too few individuals with incomes over 185 percent of federal poverty guidelines. If this is the case, the imputation process will create biases in the estimates of the number of eligible individuals.

142 ESTIMATING ELIGIBILITYAND PAR TICIPATION FOR THE ~CPROGR~

OPTIONS FOR ESTIMATING ELIGIBILITY AND PARTICIPATION 143 Derring the number of pregnant women and their monthly income through the use of adjustment factors based on the estimates of the number of in- come-eligible infants, SIPP data allow one to observe income over the course of a woman's pregnancy. Eligibility of postpartum women can also be directly observed in SIPP. Furthermore, SIPP also specifies which house- hold members receive WIC benefits, which helps in estimating adjunctive eligibility. SIPP, like the CPS, does not provide direct information on breastEeeding status of mothers, so that the adjustment factors for the rate and duration of breastEeeding status will need to continue to be used if the SIPP is used to estimate eligibility. SIPP does have some limitations relative to the CPS. First, given the complexity of the data, the public release of SIPP lags that of the CPS. Second, using monthly income instead of annual income requires more data in order to accurately model certification periods. For example, in

144 ESTIMATING ELIGIBILI>~D PAR TICIPATION FOR THE ~CPROG~ order to determine whether an 11-month-old infant is eligible that month, the monthly income from the previous 10 months is required (e.g., an infant may not be eligible in her 11th month, but she may have been eli- gible in a previous month and certified as eligible for the next 12 months). To determine whether a women is pregnant in December would require up to eight months of data in the next calendar year. Estimating eligibility from SIPP for any calendar year will require data from the waves for the year of interest, all of the waves from the preceding year, and at least three waves from the following year. Even if the timing of the release of SIPP closely matched that of the CPS, there would still be a wait of a year in order to estimate eligibility and participation for the same year from both SIPP and the CPS. Both of these factors would lengthen the forecast pe- riod the time period between when data are available for analysis and the time for which budget decisions are being made and hence, potentially . . . increase pre( action error. Another potential problem with SIPP's longitudinal data is cumulative attrition over waves of interviews, although response rates for each wave are high. Evidence from previous SIPP panels suggests that attrition is more likely to occur among young adults, males, minority groups, never-married people, people with incomes below the poverty level, and people with low educational attainment (Lames et al., 19941. Since the Census Bureau is well aware of the problem of attrition, SIPP makes a number of reasonable efforts to reduce this type of nonresponse. Weights are designed to reduce nonresponse bias and, through posts/ratification, are made to resemble the month-by-month U.S. population by age, race, and gender. Comparison of Options Table 9-2 presents estimates of the number of infants, children, and pregnant women who are eligible for WIC and coverage rates for these groups computed by using the current USDA methodology and the panel's proposed methodologies for CPS and SIPP options.9 We focus on the years 9We have chosen not to present results for breastfeeding and postpartum women since we do not make specific recommendations pertaining to the proportional adjustment factors to be used for breastfeeding rates less than 6 months and greater than 6 months. These two rates are needed to estimate eligibility for these two groups under both the CPS and SIPP options. The estimates based on the current USDA method do not make an adjustment for the prevalence of nutritional risk among the income-eligible populations.

OPTIONS FOR ESTIMATING ELIGIBILITY AND PARTICIPATION 145 TABLE 9-2 Eligibility and Coverage Rate Estimates of Infants, Children, and Pregnant Women Year 1994 1995 1996 1997 1998 1999 Infants USDA methodology Eligibility (in 1,000) 1,628 1,669 1,620 1,543 1,492 1,470 Coverage rate 116% 115% 119% 127% 133% 136% CPS option estimates Eligibility (in 1,000) 2,298 2,345 2,377 2,236 2,187 2,215 Coverage rate 78% 79% 77% 83% 86% 86% SIPP option estimates Eligibility (in 1,000) 2,493 2,368 Coverage rate 75% 80% Children USDA methodology Eligibility (in 1,000) 7,350 6,963 6,893 6,813 6,375 6,076 Coverage rate 58% 67% 72% 75% 78% 80% CPS option estimates Eligibility (in 1,000) 8,785 7,900 8,245 7,823 7,590 7,496 Coverage rate 36% 44% 45% 49% 49% 49% SIPP option estimates Eligibility (in 1,000) Coverage rate Pregnant Women 9,383 9,039 41% 41% USDA methodology Eligibility (in 1,000) 1,202 1,232 1,196 1,139 1,102 1,085 Coverage rate 67% 66% 69% 74% 78% 78% CPS option estimates Eligibility (in 1,000) 1,206 1,230 1,247 1,173 1,147 1,162 Coverage rate 66% 66% 66% 72% 75% 73% SIPP option estimates Eligibility (in 1,000) Coverage rate 1,465 1,329 58% 65%

146 ESTIMATING ELIGIBILITYAND PAR TICIPATION FOR THE ~CPROGR~ of 1997 and 1998, since estimates using all three data sets are available only for those two years. For infants in 1997, the panel's CPS option results in a 45-percent increase in eligibility estimates over estimates based on current methodol- ogy. The SIPP option results in a 62-percent increase in eligibility esti- mates. In 1998, the CPS option results in a 46-percent increase in the number of infants estimated to be eligible for WIC, while the SIPP option results in a 59-percent increase in the number of infants estimated to be eligible. For children, the CPS option results in a 15-percent increase in eligi- bility estimates over estimates based on the current methodology, while the SIPP option results in a 38-percent increase in eligibility estimates for 1997. In 1998, the CPS option results in a 19-percent increase in the number of infants estimated to be eligible for WIC, while the SIPP option results in a 42-percent increase in the number of infants estimated to be eligible. For pregnant women, the CPS option yields roughly the same number of eligible women in 1997 and 1998. Employing the SIPP data, there are significantly more eligible pregnant women. In 1997,29 percent more preg- nant women are estimated to be eligible. The 1998 SIPP estimates show a 21 percent increase in the number of eligible pregnant women compared with estimates based on the current method. These considerably greater estimates of eligibility compared with esti- mates based on the current methodology translate into coverage rates un- der 100 percent. Using the panel's proposed CPS-based estimates, the cov- erage rates for infants range from 77 percent in 1996 to 86 percent in 1998 and 1999. Coverage rates for children range from 36 percent in 1994 to 49 percent in 1997-1999. SIPP-based coverage rate estimates for infants range from 75 to 80 percent and for children are 41 percent. CPS-based coverage rates for pregnant women, which range from 66 to 75 percent, are only slightly lower than those estimated using the current USDA methodology. SIPP-based coverage rates for pregnant women are 58 percent in 1997 and 65 percent in 1998. The SIPP coverage rate estimates are lower than those using the CPS- based option, which is a reflection of the larger numbers of eligible indi- viduals estimated by SIPP data. The panel examined the differences be- tween the CPS-based option estimates and those derived from the SIPP data. The majority of the difference was due to differences in monthly income from the two data sources. The smaller degree of variability in the

OPTIONS FOR ESTIMATING ELIGIBILITY AND PARTICIPATION 147 TRIM-imputed monthly income, compared with the SIPP data, resulted in fewer children being found eligible during the course of the year (a fuller description of this reconciliation is found in Appendix C). Of course, even with these lower coverage estimates, there may still be ineligible people participating in WIC. If so, true coverage rates could be even lower. We further emphasize that these lower estimates of coverage rates are due to increases in eligibility estimates after accounting for monthly income and adjunctive eligibility. They use the same participation levels obtained from administrative data. PREDICTING WIC FULL-FUNDING PARTICIPATION Each year, USDA submits to Congress a budget requesting funds for the WIC program. In recent years, the administration has submitted to Congress a budget requesting sufficient funds for the WIC program so that every eligible individual who wishes to participate may enroll in the pro- gram in other words, fully fund the program. To estimate the level of full funding for the WIC program, the primary question is how many of the eligible individuals will choose to participate. If waiting lists for the WIC program, which would deny eligible individuals from receiving benefits, have not occurred, then one would be tempted to conclude that the fund- ing had been adequate to meet the congressional desire to fully fund WIC.10 If the absence of waiting lists indicates that full funding has been achieved, any future changes in the funding level of the program would reflect antici- pated changes in the number of individuals eligible for WIC or changes in the rate by which individuals chose to participate. Concluding that full-funding levels are achieved if waiting lists are not needed may not be appropriate. This conclusion assumes that a family's decision to participate in the WIC program is not influenced by adminis- trative practices of the local WIC programs. Chapter 8 indicated that the family's decision to participate was influenced by the amount of informa- tion they have about the program, the level of benefits they can expect to receive, and the costs of acquiring these benefits. Even with an absence of waiting lists, a family's decision to participate may also depend on the Thor purposes of this discussion, we assume that there are no excess funds of the pro- gram and the program does not serve individuals who are not eligible.

148 ESTIMATING ELIGIBILITYAND PAR TICIPATION FOR THE ~CPROGR~ amount of program outreach, the proximity of WIC offices, their hours of operation, and other administrative practices of local WIC offices. The panel heard testimony from several state directors during its first phase of work. One state WIC director indicated that, for many years, the WIC program in her state concentrated outreach efforts on women and infants, since they were concerned that funds would be insufficient to serve chil- dren too. However, once it became evident that funding was adequate to serve children, the state then began to concentrate outreach on children. States direct funds to local offices, which use funding targets. State agencies can lose future funding if they fail to meet the state-specified targets. The incentives to raise or lower participation in the program to hit these fund- ing targets may be a potent force in determining the actual rate of partici- pation (coverage rate) in a year. These examples and observations have led us to conclude that the rate at which eligible individuals participate in the WIC program (participation rate) should be viewed as much as a policy choice as it is a reflection of individual behavior to participate. To understand the consequence of viewing the participation rate as a policy choice, consider the context of how budget requests for the WIC program are created. Assume that there is a four-year lag between the year the data used for the prediction budget submission were collected and the year for which the budget request is being made. For example, in the pro- cess of preparing the WIC budget request for 2003, assume that USDA employs data that reflect the demographic and economic characteristics of individuals in 1999. Using these data and the methodology described ear- lier in the chapter, USDA would first estimate the number of eligible indi- viduals in the categories of infants, children, and pregnant and postpartum women in 1999. The next step is to project forward the number of eligible individuals from 1999 to 2003.11 Finally, USDA would explicitly make a judgment about an appropriate participation rate among eligible individu- als, which is then the policy goal for the program in 2003. In setting these goals, USDA should take into account current coverage rates of the various groups and the likelihood that changes in administrative practices can in- IIHistorically, USDA has assumed that there is no change in the number of eligible individuals over the four-year period. Some adjustments could be made for changes in demo- graphic factors such as birth rates, mortality rates, and immigration, but if economic condi- tions and other program rules do not change eligibility during a four-year period, it would be a reasonable decision not to adjust the eligibility estimates from 1999 to 2003.

OPTIONS FOR ESTIMATING ELIGIBILI~YAND PARTICIPATION 149 fluence participation among eligible individuals.12 Assuming that additional administrative changes and program outreach efforts aimed at increasing participation levels will have a diminishing return on increasing WIC par- ticipation, and given the relatively low benefit values and the inherent stigma that some recipients attach to receiving public assistance benefits, it is likely that the full-funding participation rate (FFPR) is substantially be- low 100 percent. Furthermore, it is likely that the FFPR will differ across the five demographic eligibility groups of pregnant women, infants, chil- dren, and breastEeeding and nonbreastEeeding postpartum women. In the process of making a budget request, policy makers will have set their goal for participation in the program and will assess whether more or less funding is needed for the program, as compared with previous years. Coverage rates are used to make this assessment. For example, when pre- paring the budget request for 2003, policy makers would have estimates of both the number of eligible individuals in 1999 (E1999), and the number of participants in 1999 (P1999), which comes from administrative data. From these two pieces of information they could compute the WIC coverage rates in 1999 (CR1999) as: CR1999= Pl999 1999 For example, if USDA employs the CPS-based option for estimating eligi- bility, it would have estimated 2,215,000 eligible infants (Table 9-2), im- plying that 86 percent of eligible infants were served during 1999. The USDA's method to estimate the number of eligible infants who will partici- pate assumed an 80 percent participation rate. For the time being, let us assume that this 80 percent participation rate was actually the policy goal for the program that is, the goal was to allocate funds to serve 80 percent of those who were eligible.13 Because the estimated coverage rate exceeds 12Empirical studies on the decision to participate in WIC similar to the ones reported in Chapter 8 or perhaps a study that interviews people eligible for WIC who chose not to participate could inform these types of decisions to be made by USDA and the Congress. 13One interpretation of the USDA assumption that 80 percent of eligible individuals will participate is that the assumption was more of a statement of a policy goal than a "behav- ioral" prediction about actual participation in the program.

150 ESTIMATING ELIGIBILI~YANDPARTICIPATIONFOR THE WICPROGRAM the full-funding participation rate and there were no waiting lists for this group, a reasonable inference is that the program was fully funded in 1999. To make budgetary decisions, estimates of the number of eligible in- fants using 1999 data would be used to forecast the number of eligible infants for 2003. Ideally, in making this forecast, one would want to take into account changes in the population, the economy, and eligibility rules (not only for WIC but also for other means-tested programs that affect adjunctive eligibility) over the four-year period. Trying to account for the impact of these factors may improve the expected accuracy of the forecast. However, the potential variability of the forecast errors may be high. Be- cause modeling these potential changes could introduce error, it may be better (based on mean square error criteria) to forecast no change in the number of eligible infants (or any group) over the four-year period.14 His- torically, USDA has made this judgment and assumed that the number of eligible individuals does not change over the four-year period (E2003=E19991. Once the number of eligible infants in 2003 is forecasted, the next step is to estimate the number of full-funding participants. If (as our example suggests) the program was determined to be fully funded for infants, the same percentage of eligible infants should be expected to participate in 2003 if the program's administrative practices and other factors that affect participation were not changed. If this is the desired full-funding participa- tion level, USDA should use the following formula to estimate participa- tion in 2003: . FFPR x E2003 = CR1999 x E2003 This is, simply, the coverage rate in 1999 multiplied by the number of infants estimated to be eligible in 2003. However, since the USDA implic- itly assumes that the number of eligible infants does not change over the forecasting period (from 1999 to 2003), the above expression can be re- written as FFPR x E2003 = CR1sss x E2003 = E x E1999 P1999 999 14Assume that the true change in the number of eligible infants is ~ while the predic- tion of the change based on a forecasting model is D = ~ + £ where £ is the error in the forecast and its expected value is O and variance is O2. Predicting no change will be better than modeling the change on a mean squared error basis if ^2 is less than O2, in other words, if the variability in the forecast errors is expected to be greater than the change or bias in predicting no change.

OPTIONS FOR ESTIMATING ELIGIBILII~YAND PARTICIPATION 151 which is the number of participating infants in 1999. Now consider the case of children. Employing the CPS option, we estimate that in 1999 7,496,000 children would be eligible. In 1999, 3,673,040 children were served by WIC. This corresponds to a 49-percent coverage rate in 1999. Given this lower coverage rate, USDA may conclude that more eligible children could be served through greater administrative efforts. Assume that USDA chooses a policy goal (FFPR) of serving 60 percent of the eligible children. Since the most current coverage rate is less than the full-funding participation rate, the estimate of the number of chil- dren eligible and likely to participate under full funding would equal: FFPR x E2003 = FFPR x Elggg The USDA would request funding to serve 4,498,000 children (0.60 x 7,496,000), which is 22 percent greater than the 1999 figure. The paragraphs above outline the strategy the panel recommends to estimate the number offull-funding participants. Within this strategy, there are two possible methods to be used which one is chosen depends on whether policy makers decide that full-funding participation levels have been achieved or not: · If the FFPR has been achieved, then last year's participation levels can be used to estimate next year's participation levels. · If the FFPR has not been achieved, then the desired FFPR can be multiplied by the estimated number of eligible persons in the eligi- bility category.15 This strategy can be represented by the following equation, assuming that there is a four-year time difference between the year for which the budget is prepared (year t) and the most recent year for which the number of eligible individuals can be estimated (year t - 41: 15A variant of this method is to use a weighted average of coverage rates for the past three years, presuming that this would be a more stable estimate of the coverage rate experi- enced over the time period. This variant is discussed below.

152 ESTIMATING ELIGIBILII~YANDPARTICIPATIONFOR THE WICPROGRAM Panel's Alternative Strategy: If CRt_4 2 FFPR then FFPt = CRt-4 X Et_4 = Pt_4 If CRt_4 < FFPR then FFPt = FFPR x Et_4 This method is based on the assumption that coverage rates do not exceed 100 percent. If coverage rates for any group begin to exceed 100 percent, then USDA should undertake an investigation as to whether eligibility is being significantly understated or whether there has been a substantial in- crease in the number of participants who are ineligible for the program. It is likely that some categorical groups may have achieved the full-funding level, while others have not. Thus, the assessment of which estimator to use should be conducted separately for each categorical eligibility group. It is instructive to compare the USDA methodology for predicting the number of participants for budget requests to the panel's proposed strategy. The USDA strategy for forecasting the number offully funded participants (FFP) is equal to: Current USDA Method: FFPt = 0.80 x E' = 0.80 x E' 4 recalling that the USDA assumes that eligibility does not change over the four-year prediction period. Table 9-3 examines the percentage difference between the predicted number of participants (made using the USDA's current methodology to estimate eligibility and participation) and the actual number of partici- pants (from administrative records). We call this percentage difference the "prediction error rate." The use of the actual number of participants is relevant only if the program is considered to be fully funded. For the sake of comparison we assume that the program has been fully funded from 1996. The results of this comparison are presented in Table 9-3 in the panel labeled "USDA methodology." The percentage difference between the ac- tual number of participants and the estimated number of participants for infants ranges from negative 26 to negative 39 percent, indicating that the estimates are smaller than the actual number who are served. For postpar- tum women, the prediction error rate is also negative, ranging from 15 percent in 1996 to 36 percent in 2001. Instead of getting better over time, the prediction of the number of participants in these categories is getting worse. However, the predicted number of children and pregnant women exceeded the actual numbers of participants. While the prediction error for children is roughly constant over this period, USDA predictions for preg-

OPTIONS FOR ESTIMATING ELIGIBILII~YAND PARTICIPATION 153 TABLE 9-3 Comparison of Methods for Predicting Participation Levels: Percentage Difference Between Estimated Number of Participants and the Actual Number of Participants USDA Methodology (Assume 80% of eligible persons participateja Pregnant Postpartum Infants Children Women Women Total 1996 -27% +18% +16% -15% +3% 1997 -26% +21% +17% -17% +4% 1998 -34% +18% +4% -27% -2% 1999 -33% +14% +8% -27% -4% 2000 -35% + 17% +5% -30% -4% 2001 -39% +14% +2% -36% -8% Alternative Method Recommended by the Panel if the Full-Funding Goal Is Achieved (Use administrative counts from time t- 3 to predict participation in t)6 Pregnant Postpartum Children Women Women Total 1996 -5% -24% -6% -28% -18% 1997 -4% -17% -6% -19% -13% 1998 -4% -7% -6% -13% -6% 1999 -4% +1% -2% -9% -2% 2000 -2% +8% +1% -5% +3% 2001 -2% +4% +4% -8% +1% aPercentage difference between estimates of the number of participants based on USDA:s method for estimating the number of eligible people who will participate (assuming an 80 percent participation rate) and the actual number of participants from administra- tive records. Percentage difference between estimates of the number of participants and actual num- ber of participants when estimates assume that the number of predicted participants will be equal to the last year's number of participants. nant women have improved. If one is not concerned with estimating par- ticipation within eligibility categories but only with the total counts of participants, the USDA predictions are surprisingly good. In 1996 and 1997, USDA's predicted number of participants exceeded actual numbers of participants. However, since 1997, the USDA methodology has led to a growth in the underestimate of the actual number of participants.

154 ESTIMATING ELIGIBILI~YANDPARTICIPATIONFOR THE WICPROGRAM In recent years, USDA has not used the 80-percent participation as- sumption to estimate the number of eligible people who will participate. Rather, it has been the department's practice to fund WIC at a level that will serve 7.5 million people. Figure 9-1 shows that this has been a success- ful way to "predict" the number of WIC participants. The number of WIC participants has remained steady for the past five years. Even during a time of historic economic expansion and declining participation in other social welfare programs, the level of WIC participation has remained fairly con- stant. This may be an indication that the number of participants is a policy choice, one that can, in effect, be set at a level that meets goals for the program. If it is determined that the FFPR has not been achieved, it may be desirable to use the weighted average of the past three years' coverage rates to multiply by the estimated number of eligible persons in the category (instead of using just the last available year's coverage rate). In other words, with the panel's alternative strategy, CROP would be replaced by a coverage rate that is a weighted average of coverage rates of the previous three years (coverage rates from years t- 4, t- 5 and t- 61. This weighted average would presumably be a more stable estimate of the actual coverage rate, although this presumption should be explored. The second panel of Table 9-3 present the prediction error under the alternative strategy outlined above to estimate the number of participants when the full funding goal has been achieved. For illustrative purposes, we assume that the most recently available survey data are from four years 4,500 4,000 (n Q .O . _ ~ 2,500 o ~ 2,000 id 3,500 3,000 1 ,500 1 ,000 500 o 1996 1997 1998 Year 1 999 2000 FIGURE 9-1 Number of WIC participants 1996-2000, by category. Infants I Children Pregnant women Postpartum non- breastfeeding i Postpartum breastfeeding

OPTIONS FOR ESTIMATING ELIGIBILII~YAND PARTICIPATION 155 prior to the year for which we wish to predict. Furthermore, we assume that the most recent administrative data are from three years prior to the year for which we are predicting. The panel labeled "method if full-funding is achieved" uses administrative counts of the number of participants from administrative records to predict participation.16 This method would be used if policy makers judged that the FFPR has been achieved. The num- bers reported in the panels are the percentage difference between predic- tions of participants and the actual number of participants from adminis- trative records three and four years later. The relative accuracy of the alternative strategy depends on the stabil- ity of the number of participants over a four- or three-year period. Table 9- 4 presents annual growth rates in the number of WIC participants by cat- egory. Over the period since 1994, the average annual growth in the number of infant and pregnant women participants has been relatively moderate. Consequently, the use of past participation in the program for these groups has led to rather "accurate" forecasts of the number of participants, espe- cially in comparison to the strategy employed by USDA. For both children and postpartum women, the annual growth in their participation did not fall below 5 percent until 1997 for children and a year later for postpartum women (Table 9-41. In previous years, the growth in the participation of these groups meant that using past participation as an indicator of future participation would lead to substantial underestimates of participation. The alternative forecast strategy tends to understate the number of participants in future years. This is reflected in the results pre- sented in Table 9-3 for the time period of 1996 to 1998 for children. For postpartum women, it is not until the growth in this group has slowed that this approach becomes accurate. Here the alternative strategy does no bet- ter than the current USDA method. However, when growth in the partici- pation of these groups declines, the strategy of using past participation to predict future participation significantly improves the accuracy of the esti- mates. 16We could not provide estimates of the variation using a weighted average of past coverage rates since we did not have sufficient past years' data available to construct the weighted averages for all the years from 1996 to 2001. Recall when making a prediction for year t, the most recent data available are assumed to be four years prior. Hence to construct even a three-year weighted average of past coverage rates, we would need data from six years prior to the year we are predicting. Given that we had coverage rates from 1994 to 1999, we could have provided estimates only for 2000 and 2001.

156 ESTIMATING ELIGIBILI>~D PAR TICIPATION FOR THE ~CPROG~ TABLE 9-4 Annual Growth Rates in the Number of WIC Participants (percentage) Infants Children Pregnant Postpartum Women Women Total 1992 8.0 12.6 NA NA 10.4 1993 3.4 12.8 NA NA 9.6 1994 2.6 13.4 2.6 19.3 9.4 1995 1.7 9.7 1.5 9.4 6.4 1996 0.5 6.1 2.1 7.0 4.3 1997 2.0 3.3 2.3 5.4 3.1 1998 1.1 -2.2 1.3 1.4 -0.5 1999 0.8 -2.1 -1.6 2.6 -0.8 2000 -0.2 -3.2 -0.3 1.0 -1.5 2001 1.4 1.4 -1.7 4.9 1.5 NA= not available. This result is what we would expect. If there is significant growth in any group, then one of two factors have occurred either there has been a growth in the number of eligible individuals or there has been an increase in the rate at which eligible individuals choose to participate in the pro- gram. In the absence of any evidence that eligibility has grown, it must be the case that not all eligible people were participating. Thus, the primary assumption underlying these alternative strategies was not met the pro- gram had not achieved a full-funding rate of participation. In these years, instead of using the actual coverage rate, the implicitly higher full-funding participation rate should have been employed. Consequently, the forecast of participation would have yielded a higher number of participants and reduced the difference between the forecast and the actual number of par- . . tlclpants. The Panel's Recommended Strategy for Estimating Participation After examining this alternative strategy for predicting participation and two of its three variants, the panel recommends that USDA forecast future numbers of participants in the following manner: Explicitly state the rate of participation in the WIC program that is consistent with the policy goal of fully funding the program. This is the full-funding participation rate (FFPR).

OPTIONS FOR ESTIMATING ELIGIBILITY AND PARTICIPATION 157 . During the process of creating a budget request, compute the num- ber of eligible individuals by participant group (infants, children, and pregnant, breastEeeding postpartum, and nonbreastEeeding postpartum women) and their respective coverage rates using con- current administrative data for the actual number of participants (i.e., use administrative data for the same year covered in the survey data that is used to estimate eligibility). Estimates of eligibility should be made using one of the methods outlined by the panel in the first part of this chapter. Policy makers could, as an alternative to setting full-funding participation goals by category, set them by other groups of priority, for example, by those in most need. This could be done within an eligibility category as well (e.g. infants with the lowest income). · Separately for each participant group, determine whether the group's coverage rate exceeds the FFPR. If the coverage rate does exceed the FFPR, then use the most recently available administrative data on the number of participants to estimate the future number of par- . . tlclpants. If the group's coverage rate does not exceed the FFPR, then estimate the number of participants by multiplying the FFPR by the num- ber of eligible individuals from the most recently available data. Alternatively, USDA could construct a three-year weighted average of past coverage rates.18 If the weighted average of coverage rates exceeds the FFPR, then the weighted average of past coverage rates for the group would be multiplied by the most recently available estimate of the number of eligible individuals from the participant group. This recommendation implicitly assumes that the number of eligible indi- viduals and, correspondingly, the number of participants for the year from which there are data is the same as for the year for which participation is being predicted. During this period of time, changes in eligibility could be caused by changes in demographic factors, the economy, or the eligibility rules of WIC or other programs that provide adjunctive eligibility. While it is unlikely that demographic shifts will occur during this relatively short . 180ne suggestion is to use an exponential weighting scheme so that more recent cover- age rates receive higher weights.

158 ESTIMATING ELIGIBILITYANDPARTICIPATIONFOR THE WICPROGRAM time period, the latter two factors could greatly influence future participa- tion in WIC. In the panel's judgment, USDA should explore the accuracy and feasibility of methods to adjust eligibility and participation forecasts to account for such changes. The effects of changes in the eligibility rules for WIC can, in principle, be estimated by "simulating" the new rules with existing data, such as the CPS or SIPP. While this approach will lead to estimates of the number of eligible individuals under the new rules, to forecast the number of partici- pants will be difficult because there is a greater degree of individual choice in the decision to participate. Modeling such behavioral choices is more difficult. If the changes in eligibility rules reduce the size of the eligible population, then future participation rates may be higher than current cov- erage rates. Hence, using the most recent coverage rate may understate the future number of participants. Conversely, rule changes that expand eligi- bility will have the opposite effect. Coverage rates will be much lower be- cause of the increase in eligibility, assuming participation levels do not change. Thus, USDA may want to reassess what it defines or believes to be the FFPR if such changes take place. If the USDA deviates from the use of either the FFPR or the most current coverage rate to forecast future partici- pation, it should explicitly state this deviation in its forecast and the justifi- . ~ . . cation tor c going It. SUMMARY This chapter outlines two methods to improve estimates of eligibility for WIC and proposes a method to forecast participation. The proposed improvements in estimating eligibility seek to better account for income variation and adjunctive eligibility. The proposed improvements for pre- dicting participation seek to reduce the size of errors associated with predic- tion into the future. Regardless of which of the methods are chosen, it will be important to periodically review their performance. The panel encourages USDA to con- tinue the efforts it has made, previous to this panel's formation, to review the methods and their assumptions.

Next: 10. Summary »
Estimating Eligibility and Participation for the WIC Program: Final Report Get This Book
×
Buy Paperback | $61.00 Buy Ebook | $48.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

This report reviews the methods used to estimate the national number of people eligible to participate in the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) under full funding of the program. It reviews alternative data sets and methods for estimating income eligibility, adjunctive eligibility (which occurs when people are eligible for WIC because they are enrolled in other federal public assistance programs) and nutritional risk, as well as for estimating participation if the program is fully funded.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!