1.the first part I gave you from my instructor’s requirement
Your assignment for this research essay is to analyze either food insecurity in Alabama and
provide a recommendation for action. These topics are quite broad; for this essay project you
will choose a topic and narrow it sufficiently to create a compelling, researchable, and
For each of these topics, literally millions of pages have already been written. What I want
from you is an exploration of the topic that gets several layers beneath the surface. What is the
biggest debate in your topic area? Explore it in your dialogue. Then decide for yourself what
matters, what people need to know, and what needs to happen. Whatever it is, drill down,
get the meaty information, and see what seems to matter.
Your sources for this paper must be very fresh. You can use only two sources (at most) that
are more than 24 months old. For the rest of your research, you can quote ONLY from
sources no more than 24 months old. If you cannot verify the age of your source (it’s an
undated blog rather than a dated article, for instance), you cannot use it unless it is an “extra”
source (not one of the 2). You’ll need at least 2 sources for this essay in addition to any you
might use from the list I've provided.
Some initial resources for our research project:
1.Food Insecurity in Alabama
2.Feeding America (hunger and poverty facts)
3.Zekeri, Andrew A. "Educational Attainment and Self-Rated Health Status among Single Mothers in
Rural Alabama." Psychological Reports, vol. 113, no. 1, Number 1/August 2013, pp. 175-179.
double spaced, Times New Roman
2 new sources from library databases in addition to baseline research I've provided for you
2.this part is what I want to speak
I hope the theme of research project is about food insecurity in Alabama State
You must use sources at least2 sources from source 3.4.5.
University of Kentucky
Discussion Paper Series
Beyond Income: What Else Predicts Very Low Food
Security among Children?
Patricia M. Anderson
Kristin F. Butcher
Hilary W. Hoynes
University of California, Berkeley
Diane Whitmore Schanzenbach
Anderson, P., Butcher, K., Hoynes, H., & Schanzenbach, D., Beyond Income: What Else Predicts Very Low
Food Security among Children? University of Kentucky Center for Poverty Research Discussion Paper Series,
DP2014-06. Retrieved [Date] from http://www.ukcpr.org/Publications/DP2014-06.pdf.
Patricia Anderson, 6106 Rockefeller Hall, Department of Economics, Dartmouth College, Hanover, NH
03755; Email Patty.Anderson@dartmouth.edu; Phone: (603)646-2532
University of Kentucky Center for Poverty Research, 302D Mathews Building, Lexington, KY, 40506-0047
Phone: 859-257-7641; Fax: 859-257-6959; E-mail: firstname.lastname@example.org
Beyond Income: What Else Predicts Very Low Food Security among Children?
Patricia M. Anderson (Dartmouth College)
Kristin F. Butcher (Wellesley College)
Hilary W. Hoynes (University of California-Berkeley)
Diane Whitmore Schanzenbach (Northwestern University)
We examine characteristics and correlates of households in the United States that are most likely
to have children at risk of inadequate nutrition – those that report very low food security (VLFS)
among their children. Using 11 years of the Current Population Survey, plus data from the
National Health and Nutrition Examination Survey and American Time Use Survey, we describe
these households in great detail with the goal of trying to understand how these households differ
from households without such severe food insecurity. While household income certainly plays an
important role in determining VLFS among children, we find that even after flexibly controlling
for income-to-poverty rates some household characteristics and patterns of program participation
have important additional explanatory power. Finally, our examination of the NHANES and
ATUS data suggests an important role for both mental and physical health in determining the
food security status of children.
Note: We thank Mary Zaki for excellent research assistance. We also thank Joshua Gliken, Patrick Gould, Grace
Ma, Nicholas Paine, Jamie Song and Linh Vu for excellent research assistance under the auspices of the James O.
Freedman Presidential Scholar Program at Dartmouth College. This project was supported with a grant from the
University of Kentucky Center for Poverty Research through funding by the U.S. Department of Agriculture, Food
and Nutrition Service, contract number AG-3198-B-10-0028. The opinions and conclusions expressed herein are
solely those of the authors and should not be construed as representing the opinions or policies of the UKCPR or any
agency of the Federal Government.
Access to healthful food during critical periods of fetal and child development is an
important determinant of long-term health and economic well-being.1 In this study, we examine
households in the United States that are most likely to have children at risk of inadequate
nutrition – those that report very low food security among their children. Although food
insecurity in the United States is quite common (about 20 percent of households with children in
2012), very low food security among children is relatively uncommon (about 1.2 percent of
households in 2012).2 Even though households with very low food security among children make
up a small percentage of households, the percent of households with this status has roughly
doubled over the last decade. Further, these households account for a disproportionate share of
children, as poor households tend to have more children, and the children in these households are
those for whom the risks of inadequate nutrition during critical periods of development are a real
possibility. In this study, we examine the characteristics and correlates of households with very
low food security among children. Among most low-income households, even those that report
that they are food insecure, children appear to be insulated from food insecurity themselves.
Here, we explore what publicly available data can tell us about households in the U.S. where the
children live at the extremes of poverty.3
Using 11 years of the Current Population Survey, plus data from the National Health and
Nutrition Examination Survey and American Time Use Survey, we describe these households in
See Currie (2009) for a review of the literature on the importance of early life incomes, and Hoynes, Schanzenbach
and Almond (2012) for a specific example of the benefits of childhood food stamp receipt on reducing the likelihood
of poor adult outcomes.
These statistics come from http://www.ers.usda.gov/topics/food-nutrition-assistance/food-security-in-the-us/keystatistics-graphics.aspx#children
We are using the term “extremes of poverty” loosely, not the formal definition of “extreme poverty” defined by the
World Bank as households living on $2 or less per person per day. Edin and Schaefer (2013) use this formal
definition and find that 4.3 percent of non-elderly households with children in the U.S. were in this category in 2011.
great detail. Although income is clearly an important part of the story, most households even at
very low-income-to-poverty ratios do not have food insecure children. Our goal in this paper is
to try to understand what is different about those households that do. We are not attempting to
provide a causal analysis of, for example, the impact of program participation or health status on
the incidence of very low food security among children. However, we will present the
correlations between a household reporting very low food security among children, and a large
list of household descriptors. Insights that come out of these detailed descriptions can be used to
guide further research and policy.
We proceed by first describing the data we use, explaining definitions of different types
of food insecurity, and showing the prevalence of very low food insecurity among children.
Focusing on data for households with children where the income to poverty ratio is less than
300% of the poverty threshold, we present summary statistics on participation in various public
programs and household characteristics by different food security levels. By linking
longitudinally across two years of Current Population Surveys, we also show transition rates into
different programs and food security levels.
After establishing the correlates of food insecurity, we turn to regression analysis. Again,
it is important to emphasize that this is not a causal analysis, but rather a “horse-race” style
analysis to see which correlates of very low food security among children are statistically
significant when income-to-poverty ratios and other covariates are held constant. The thought
experiment here is that if income is the only thing that matters for determining children’s food
security, then even if income does a poor job of explaining the variation in children’s very low
food security status, nothing else should be systematically correlated with the outcome. Those
things that remain robustly statistically significant suggest correlates of unmet need and may
provide guidance for public policy aimed at addressing the extremes of poverty.
Our findings suggest that some household characteristics and patterns of program
participation, even controlling flexibly for income-to-poverty, systematically predict very low
food security among children. For example, controlling for household size, having a larger share
of the household in the 13-to-18 age range is positively associated with very low food security
among children, suggesting that rapidly growing teenage children may put greater stress on a
household’s ability to provide food security for them. Participation in programs like free and
reduced priced lunch and SNAP are positively correlated with very low food security among
children, suggesting a selection story where these are struggling households that have already
identified themselves as requiring assistance, but who continue to have unmet needs. Finally, our
examination of the NHANES and ATUS data suggests an important role for both mental and
physical health in determining the food security status of children.
II. Data Sources
A. The Current Population Survey
Food insecurity is officially measured in the U.S. based on a supplement to the Current
Population Survey (CPS). Since 2001, this supplement has been part of the December survey.
Because the questions refer to the past twelve months, we consider the food security measure to
refer to the calendar year of the survey. Food security is defined based on a battery of 18
questions (10 if there are no children in the household), which are shown in Appendix Tables 1a
and 1b. Based on the answers to these questions, households are categorized as food secure or
food insecure. Food insecure households are further broken down into those suffering from very
low food security. In addition to the overall food security status of the household, there are
specific designations for the children in the household, based on the questions about the children.
The children themselves may be food secure or food insecure, and food insecure children may be
suffering from very low food security. Appendix Table 2 shows how each of these six categories
is defined. Very low food security among children (the topic of this paper) is clearly quite severe,
with five or more of the eight questions specifically about children having to be answered in the
affirmative to be so classified.
In order to analyze the determinants of very low food security among children, it is
important to not only have data on the answers to the 18 food security questions, but also to have
good information on the household’s income and program participation. The March supplement
to the CPS collects this information, in reference to the previous calendar year. The CPS
sampling frame allows us to match this March supplement to the December supplement for a
subset of the sample. A CPS household is in the sample for four consecutive months, out of the
sample for eight months, and then back in for four consecutive months. Thus, for households
where December is the first of one of their set of consecutive months, they will also be surveyed
in March and the two surveys can be matched at the household level. Additionally, starting in
2002, the March supplement sample was expanded by asking the questions of the February and
April sample households that were not also in the March sample, as well as some of the prior
November sample. Matching on the household identifier across these months results in a sample
of about 14,000 matched households per year. We limit our sample to households with children
and with income of 300 percent or less of the poverty line. Thus, our main analysis sample has
about 3,000 observations per year.
While our main analysis uses this matched December-March CPS data set, a subset of
households can be matched from one year to the next. A household that first joins the CPS
survey in December will rotate out the following March, but rejoin the sample for December
through March of the next year. For these households, we can observe the one-year transitions
across program participation and food security status. The result of this matching process gives
us about 4,500 households per year that can be matched to the previous year. Note, however, that
because of a change in the household identifier between 2003 and 2004, we are unable to match
across those years. Again limiting our sample to households with children and income of 300
percent or less of the poverty line leaves us with only about 750 observations per year.
B. National Health and Nutrition Examination Survey
While the official measures of food insecurity come from the CPS supplements, the same
battery of questions is asked in the much smaller National Health and Nutrition Examination
Survey (NHANES), which since 1999 has been fielded over consecutive two-year periods (i.e.
1999-2000, followed by 2001-2002, etc.). The NHANES includes a range of different
questionnaire modules, physical examinations, and a food diary, all used to evaluate the health
and nutrition status of the country. While typically not everyone in the household is a part of the
NHANES (and many children are sampled without any adult household members), the food
security questionnaire is completed at the household level for all sample members. In particular,
the status of children is ascertained whether or not the child is a sample member. Over half of the
actual sample members are the children themselves, but for our purposes we are most interested
in information that is unavailable in the CPS, such as the dietary data, and questionnaires on drug
use and mental and physical health that are characteristics associated with the adults in the
household. Thus, we restrict our sample further to only those observations where the sample
member is over 18. The result is a sample of almost 9,000 observations. However, many of the
questions and their samples change over time in the NHANES, meaning that for many variables
we have much smaller samples.4
C. American Time Use Survey
The American Time Use Survey (ATUS) asks respondents to report on how, where, and
with whom they spend their time. Respondents are a randomly chosen subset drawn from
households that have completed their final CPS monthly survey response. To be useful for our
analysis, a household must have participated in the December Food Security Supplement. Since
the ATUS is asked between 2 and 5 months after a household completes its final CPS survey, the
households that participated in a December CPS were surveyed for the ATUS between the
months of February and August. The respondent is surveyed about his or her activities
sequentially, walking through the 24-hour period that began at 4 a.m. on the designated day and
continued through 3:59 a.m. on the following day. Respondents describe in their own words the
primary activity in which they were engaged at each point in the day, and these activities are
coded into categories. While we primarily show results across the major groupings (e.g. eating
and drinking; working; household services), we also break out some activities such as food
preparation and food shopping in more detail. We limit the sample to households with children.
When the data are pooled across 2002-2010 December CPS data that can be linked to the ATUS,
we have a sample of 17,341 respondents, 2413 of which are food insecure and 100 of which have
very low food security among children.
For example, the depression screener was only given to all adults in the last three waves of the survey. Prior to that,
only a half sample of 20 to 39 year olds was screened for depression.
III. Analysis Using the Current Population Survey
A. Descriptive Analysis
As noted above, a child is classified as suffering from very low food security (VLFS) if
five or more of the questions about the child are answered in the affirmative. Essentially, then, it
is impossible to be so classified unless there are extreme circumstances in the household such as
the size of the child’s meals being cut or the child being hungry, but with no more money for
food. It is perhaps not surprising, then, that even among poorer households with income at or
below 300% of the poverty line, the rate of very low food security among children remains
relatively low, averaging about 0.013 over our CPS sample. That average masks some important
time variation, with rates reaching as high as 0.021 at the start of the Great Recession. The
average also masks geographic variation, as shown in Figure 1. In several states, such as
Colorado and New Hampshire, the rate of very low food security among children over this time
period averages under 0.003, while in states such as Missouri and Maryland it is over 7 times
higher, at 0.020. As will be described in more detail below, state fixed effects are insignificant in
a regression explaining whether a household contains a child with very low food security, while
year fixed effects are significant. However, controlling for year has no real impact on the role of
other explanatory variables. Note that the regression results reported below control flexibly for a
household’s income to poverty ratio, so it may be that the geographic variation we observe in
Figure 1 is at least partially driven by differences in financial well-being across states.
Table 1a begins our descriptive analysis by looking at rates of program participation and
at some various demographics for each of four samples. First, is the full sample of households
with children and income below 300% of the poverty line. Second is a subset of this sample
made up of only households that are coded as being food insecure, followed by the subset with
very low food security. Finally, we look at those households containing very low food secure
children. Columns (1), (3), (5) and (7) present the means for these four samples, with the
following columns giving the standard deviations. Looking across columns the columns of
means gives us in ...
Purchase answer to see full