Er the weeks. Looking at the average times spent in primary eating/drinking and secondary PD98059 msds eating (Fig 1), we see that SNAP participants had a relatively short duration average time spent in eating inPLOS ONE | DOI:10.1371/journal.pone.0158422 July 13,7 /SNAP Benefit CycleTable 1. Percent of group that reported no primary eating/drinking or secondary eating on an average day, 2006?8, age 15 and over, by week since SNAP issuance. Week 1 SNAP participants 90 confidence interval N Low-income non-SNAP 90 confidence interval N High income (> 185 poverty threshold) non-SNAP 90 confidence interval N Total population 90 confidence interval N 0.65 [0.26,1.05] 741 1.13 [0.56,1.69] 2,483 0.39 [0.29,0.49] 6,892 0.73 [0.55,0.91] 10,662 Week 2 1.48 [0.41,2.55] 579 0.47 [0.18,0.76] 1,943 0.36 [0.21,0.50] 5,664 0.51 [0.36,0.67] 8,566 Week 3 0.83 [0.12,1.54] 547 1.14 [0.60,1.67] 1,822 0.65 [0.34,0.96] 5,067 0.94 [0.67,1.21] 7,826 Week 4* 1.68 [0.78,2.59] 744 0.58 [0.33,0.84] 2,388 0.46 [0.29,0.62] 6,874 0.67 [0.52,0.82] 10,90 confidence interval in brackets [lower bound, upper bound]. Source: Authors’ estimates using 2006?8 American Time Use Survey and Eating Health Module data. * Week 4 includes calendar month days 29, 30, and 31 for those months that contain those days. doi:10.1371/journal.pone.0158422.tweek 1, a relatively long duration in week 2, then short durations in weeks 3 and 4. However, again due to large standard errors we cannot make a strong statement about eating time duration over the benefit cycle for either SNAP participants or for the low-income non-SNAP group using descriptive statistics. The estimated total time spent eating for SNAP participants in week 2 has a particularly large standard error. This is because this group has disproportionately more respondents who reported very long eating times, that is, more than 600 minutes, which created a dispersed distribution. Such long eating times are typically reported when a respondent RDX5791 supplement attends an event such as a reception, or reports engaging in secondary eating “all day.” This resulted in a large standard error relative to the estimates for the other weeks. Because of large standard errors in the descriptive statistics, another method of analysis is needed to better determine the story told by the time use data as to eating patterns over the benefit cycle.Logistic Regression ModelWe estimated a model of the likelihood of an individual going an entire day without eating. The model developed is the probability of Pr ?1jx??F ?where: yi = 1 no primary eating/drinking or secondary eating occurrences yi = 0 at least one primary eating/drinking occurrence and/or secondary eating occurrence F = logistic cumulative distribution function x = matrix of individual-level variables pi ?Pr i ?1jxi ?if yi ?1 pi ?1 ?Pr i ?1jxi ?if yi ?0 and where the matrix x contains variables pertaining to SNAP, calendar characteristics,PLOS ONE | DOI:10.1371/journal.pone.0158422 July 13,8 /SNAP Benefit CycleFig 1. Time spent in eating (primary eating/drinking and secondary eating), in minutes, on an average day 2006?8, age 15 and over. Week 4 includes calendar month days 29, 30, and 31 for those months that contain those days. Black lines are 90 confidence intervals for each estimate. Source: Authors’ estimates using 2006?8 American Time Use Survey and Eating Health Module data. doi:10.1371/journal.pone.0158422.ghousehold characteristics, individual characteristics, and geographical information. Individua.Er the weeks. Looking at the average times spent in primary eating/drinking and secondary eating (Fig 1), we see that SNAP participants had a relatively short duration average time spent in eating inPLOS ONE | DOI:10.1371/journal.pone.0158422 July 13,7 /SNAP Benefit CycleTable 1. Percent of group that reported no primary eating/drinking or secondary eating on an average day, 2006?8, age 15 and over, by week since SNAP issuance. Week 1 SNAP participants 90 confidence interval N Low-income non-SNAP 90 confidence interval N High income (> 185 poverty threshold) non-SNAP 90 confidence interval N Total population 90 confidence interval N 0.65 [0.26,1.05] 741 1.13 [0.56,1.69] 2,483 0.39 [0.29,0.49] 6,892 0.73 [0.55,0.91] 10,662 Week 2 1.48 [0.41,2.55] 579 0.47 [0.18,0.76] 1,943 0.36 [0.21,0.50] 5,664 0.51 [0.36,0.67] 8,566 Week 3 0.83 [0.12,1.54] 547 1.14 [0.60,1.67] 1,822 0.65 [0.34,0.96] 5,067 0.94 [0.67,1.21] 7,826 Week 4* 1.68 [0.78,2.59] 744 0.58 [0.33,0.84] 2,388 0.46 [0.29,0.62] 6,874 0.67 [0.52,0.82] 10,90 confidence interval in brackets [lower bound, upper bound]. Source: Authors’ estimates using 2006?8 American Time Use Survey and Eating Health Module data. * Week 4 includes calendar month days 29, 30, and 31 for those months that contain those days. doi:10.1371/journal.pone.0158422.tweek 1, a relatively long duration in week 2, then short durations in weeks 3 and 4. However, again due to large standard errors we cannot make a strong statement about eating time duration over the benefit cycle for either SNAP participants or for the low-income non-SNAP group using descriptive statistics. The estimated total time spent eating for SNAP participants in week 2 has a particularly large standard error. This is because this group has disproportionately more respondents who reported very long eating times, that is, more than 600 minutes, which created a dispersed distribution. Such long eating times are typically reported when a respondent attends an event such as a reception, or reports engaging in secondary eating “all day.” This resulted in a large standard error relative to the estimates for the other weeks. Because of large standard errors in the descriptive statistics, another method of analysis is needed to better determine the story told by the time use data as to eating patterns over the benefit cycle.Logistic Regression ModelWe estimated a model of the likelihood of an individual going an entire day without eating. The model developed is the probability of Pr ?1jx??F ?where: yi = 1 no primary eating/drinking or secondary eating occurrences yi = 0 at least one primary eating/drinking occurrence and/or secondary eating occurrence F = logistic cumulative distribution function x = matrix of individual-level variables pi ?Pr i ?1jxi ?if yi ?1 pi ?1 ?Pr i ?1jxi ?if yi ?0 and where the matrix x contains variables pertaining to SNAP, calendar characteristics,PLOS ONE | DOI:10.1371/journal.pone.0158422 July 13,8 /SNAP Benefit CycleFig 1. Time spent in eating (primary eating/drinking and secondary eating), in minutes, on an average day 2006?8, age 15 and over. Week 4 includes calendar month days 29, 30, and 31 for those months that contain those days. Black lines are 90 confidence intervals for each estimate. Source: Authors’ estimates using 2006?8 American Time Use Survey and Eating Health Module data. doi:10.1371/journal.pone.0158422.ghousehold characteristics, individual characteristics, and geographical information. Individua.
