Healthy Food, Healthy Lives

Over the course of 2020, millions of New Yorkers lost employment or were living on a reduced household income, straining their ability to afford food. @nys_health report on food scarcity during COVID-19:
In December 2020, 19% of adults with children in their household reported that the children were often or sometimes not eating enough in the prior week. @nys_health brief on food scarcity during the pandemic in NY:

Introduction

Food insecurity is a secondary crisis of the coronavirus pandemic. Over the course of 2020, millions of New Yorkers lost employment or were living on a reduced household income, straining their ability to afford food.[1] For many New Yorkers, savings are depleted, stimulus checks have long since been spent, and enhanced unemployment benefits are not sufficient to meet food needs. Furthermore, social distancing measures have disrupted previous pathways to food access, such as meals provided in community settings (e.g., houses of worship) or schools.

This report reviews self-reported survey data related to food scarcity in New York State during the coronavirus pandemic. It presents food scarcity rates by race and ethnicity, age, and household income, and compares New York State with neighboring states. Rates are also shown for child food scarcity. It assesses how different groups of New Yorkers are accessing free meals and groceries and from which access points (e.g., school programs, food pantries). Finally, it measures which methods food-scarce New Yorkers are using to meet their household spending needs. Unless otherwise specified, results in this report are among adult New Yorkers.

Data from this report come the COVID-19 Household Pulse Survey, an experimental data product designed by the U.S. Census Bureau in collaboration with multiple federal agencies. The survey, which is ongoing, is providing near real-time data on household experiences during the coronavirus pandemic. This analysis uses data from phase one (April 23, 2020–July 21, 2020), phase two (August 19, 2020–October 26, 2020), and the first four periods of phase three (October 28, 2020–December 21, 2020). This report extends trends described in a prior NYSHealth brief that focused on the phase-one results. See Methods for more details on the survey and how rates are developed for this report.

How Do Current Trends Compare with Food Scarcity Before COVID-19?

The Household Pulse Survey began collecting data in April 2020, after the onset of the pandemic. To measure the change in food scarcity resulting from the pandemic, other data sources must be used as a benchmark for pre-pandemic periods. A widely used measure comes from the United States Department of Agriculture (USDA), which reports data on food insecurity, a common approach to assessing food scarcity issues. USDA reported that 10.5% of households in New York State surveyed between 2016 and 2018 had experienced food insecurity in the last year.[2] USDA’s research uses different questions to measure food scarcity than the Pulse Survey, and asks about food access over the past year instead of the past week. Therefore, any comparisons with the Pulse data are imperfect. For example, it is likely that the food scarcity rates from the Pulse survey would be even higher if food scarcity were measured over the past year. For more details, see Methods.


[1] New York State Department of Labor. “NYS Unemployment Rate Falls to 8.2% in December 2020.” Accessed February 2021. https://dol.ny.gov/system/files/documents/2021/02/nys-unemployment-rate-falls-to-8.2-in-december-2020_0.pdf.
[2] Alisha Coleman-Jensen, Matthew P. Rabbitt, Christian A. Gregory, and Anita Singh, “Household Food Security in the United States in 2018.” United States Department of Agriculture: Economic Research Report Number 270. Accessed February 2021. https://www.ers.usda.gov/webdocs/publications/94849/err-270.pdf.

Key Findings

  • In December 2020, 14% of adult New Yorkers reported that there was either sometimes or often not enough to eat in their household in the last 7 days (defined in this report as household food scarcity). This is an increase from May 2020, when 10% of adult New Yorkers reported household food scarcity.
  • New York State’s food scarcity rate was higher than those reported in neighboring states for nearly every month.
  • In December 2020, 19% of adults with children in their household reported that the children were often or sometimes not eating enough in the prior week, because the household could not afford enough food.
  • There are stark disparities in food scarcity by race and ethnicity. In December 2020, nearly 1 in 3 Hispanic New Yorkers (32%) and more than 1 in 5 Black New Yorkers (21%) reported household food scarcity in the last 7 days. These percentages were 2 to 3.5 times higher than among white New Yorkers.
  • More New Yorkers are slipping from food security into food scarcity. In May 2020, nearly one-quarter of adults in households with food scarcity reported being food sufficient prior to the pandemic, but by December 2020, that figure had risen to more than one-third of respondents.
  • Mass losses in employment have likely contributed to increases in food scarcity. Throughout the survey period, more than half of New Yorkers reported that they or someone in their household had lost employment income since the start of the pandemic. The rate of food scarcity was nearly four times higher for those who reported lost household employment income during the pandemic, compared with those who did not (in December, 23% compared with 6%).
  • Throughout the survey period, nearly 1 in 10 New Yorkers reported that their households were accessing free meals or groceries. In most months, school programs were the most-used access points.
  • When asked which sources they were using to meet household-spending needs in the last week, food-scarce New Yorkers most frequently responded that they borrowed from family and friends, used credit cards or loans, or spent their savings or sold assets.

Food Scarcity

Overall

  • 14% of adult New Yorkers reported household food scarcity in the prior week in December 2020 (see Figure 1). The food scarcity rate is defined as the percentage of the adult population that either sometimes or often did not have enough to eat in their household in the last seven days.
  • 83% of New Yorkers who reported household food scarcity in December said they did not have enough to eat because they could not afford to buy more food (data not shown).
  • There are stark disparities in food scarcity by race and ethnicity. In December 2020, nearly 1 in 3 (32%) Hispanic adults and more than 1 in 5 (21%) Black adults reported household food scarcity in the prior week. These percentages were 2 to 3.5 times higher than among white New Yorkers (see “Race/Ethnicity” tab within Figure 1).
  • Rates of food scarcity were higher among younger (18 to 34 years old) and middle-aged adults (35 to 64 years old), ranging from 10% to 18% over the survey period, in comparison with adults 65 years and older, whose food scarcity rates ranged from 3% to 7% (see “Age” tab within Figure 1). One explanation for these findings may be that there are fewer older adults in the workforce, so their income was less likely than other age cohorts to be disrupted by job loss.
  • Higher food scarcity was associated with lower incomes. Self-reported food scarcity rates consistently surpassed 25% for households reporting less than $25,000 per year in income; households reporting annual income of $100,000 or more were consistently associated with food scarcity levels below 5% (see “Income” tab within Figure 1).[1]
  • New York State’s food scarcity rate was higher than those reported in neighboring states for nearly every month (see “State” tab within Figure 1). Although the New York State and national rates were similar in May 2020, New York’s rate surpassed the national average the following month and remained slightly higher than the national average for the rest of the survey period.
[1] Household income is defined as total 2019 household income before taxes.

Food Scarcity

Child Food Scarcity

  • In December 2020, 19% of adults with children in their household reported that the children were not eating enough in the last 7 days because the household could not afford enough food (defined as child food scarcity) (see Figure 2).[1]
  • Hispanic and Black adults reported the highest rates of child food scarcity in their households. In December 2020, 37% of Hispanic adults and 23% of Black adults living with children in the household reported child food scarcity in the prior week. These percentages were multifold higher than among white New Yorkers (see “Race/Ethnicity” tab within Figure 2).
  • There were stark disparities in child food scarcity across income brackets. In December 2020, New Yorkers earning less than $50,000 in household income reported child food scarcity rates at least 6 times higher than New Yorkers earning more than $100,000 (see “Income” tab within Figure 2).
[1] Households with children are defined as households where at least one person under the age of 18 lives in the household.

Food Scarcity

Newly Food Scarce

  • Many households that were food sufficient prior to the pandemic have newly slipped into food scarcity during the pandemic. In May 2020, nearly one-quarter of adults in households with food scarcity in the prior week had been food sufficient before the pandemic. By December 2020, more than one-third of those reporting food scarcity in the prior week had been food sufficient before the pandemic (see Figure 3). Food sufficient is defined as a household having enough of the kinds of food wanted or having enough food, but not always the kind wanted.
  • Food-scarce adults in the higher household income groups were generally more likely to be newly food scarce (see “Income” tab within Figure 3).

Employment Income Loss

Overall

  • Mass losses in employment are likely a contributor to the increase in food scarcity among previously food-sufficient New Yorkers.
  • In December 2020, more than 50% of adult New Yorkers reported a loss in household employment income since the start of the pandemic (see Figure 4).

Employment Income Loss

Loss of Employment Income and Food Scarcity

  • In December 2020, 23% of New Yorkers who reported a loss in household employment income since the start of the pandemic were food scarce in the prior week (see Figure 5). This rate was nearly four times higher than among New Yorkers who had not lost household employment income.
  • Rates were highest for New Yorkers of color, younger and middle-aged adults, and those with lower household incomes (see tabs within Figure 5).
  • Benefit programs, such as federal Economic Impact Payments (stimulus checks) and Enhanced Unemployment/Pandemic Unemployment Assistance, are intended to assist those who lost income as a result of COVID-19, including meeting their food needs. The fact that rates of food scarcity were higher for New Yorkers who reported a loss of household employment income since the start of the pandemic indicate that these benefits may not be adequate to keep some New Yorkers food sufficient.
  • Losing household employment income appeared to have the largest effect on food scarcity among Black New Yorkers, relative to any other race or ethnicity. In December, Black New Yorkers who lost household employment income during the pandemic had food scarcity rates approximately nine times higher than their counterparts who had not lost household employment income (37% as compared with 4%). This differential was substantially higher than among other races and ethnicities (see “Race/Ethnicity” tab within Figure 6). Other months of data (not pictured) point to challenges faced by Hispanic and Black New Yorkers, who have relatively high food scarcity rates, even without employment income loss.

Free Meal and Grocery Programs

Overall

  • Many New Yorkers have turned to free meal and grocery programs to help meet their food needs during the pandemic.
  • Over the survey period, 8%–11% of adult New Yorkers lived in a household that had accessed free meals or groceries in the prior week (see Figure 7).
  • Hispanic and Black adults were consistently more likely to report that their households accessed free meals or groceries in the prior week over the survey period. In December 2020, approximately 18% of Hispanic adults and 16% of Black adults reported that their households accessed free meals or groceries, compared with 9% of Asian adults and 5% of white adults (see “Race/Ethnicity” tab within Figure 7).
  • Over the survey period, younger and middle-aged adults (ages 18–64) were more likely than older adults (ages 65 and older) to have accessed free meals or groceries (see “Age” tab within Figure 7).
  • In general, the lower an adult’s household income, the higher their household’s likelihood of having accessed free meals or groceries in the prior week. In December 2020, more than 20% of adults in the lowest-income group reported accessing free meals or groceries in the prior week (see “Income” tab within Figure 7).
  • If free meal and grocery programs were not available to New Yorkers, overall rates of food scarcity would likely be much higher, especially among New Yorkers of color and low-income households.

Free Meal and Grocery Programs

By Access Point

  • In May 2020, nearly half of New Yorkers in households accessing free meals or groceries received food from school programs. That share decreased during the summer months, when the number of New York City’s school-based meal hubs providing free grab-and-go meals at the start of the pandemic declined (see Figure 8).[1] By December 2020, the proportion of New Yorkers accessing free meals or groceries from school programs rose again to more than 40%. These changes may have been influenced by the reopening of New York City public schools in October and some classrooms in December 2020; children attending in-person schooling receive grab-and-go meals in the classroom.
  • As the use of school programs as a source of free meals and groceries decreased over the course of summer 2020, the use of food pantries increased, surpassing the level of school programs by September 2020 (see Figure 8). This shift may have been influenced by the launch of the Nourish New York Program in May 2020, which provided $25 million from the State’s special public health emergency fund for food banks and emergency food providers.[2] The proportion of New Yorkers accessing free meals or groceries from pantries decreased over the fall, dropping to 22% in December 2020.
[1]  “NYC’s Grab-and-Go Meal Program Headed for an After-School Schedule,” The City. Accessed February 2021. https://www.thecity.nyc/2020/9/8/21428441/new-york-grab-and-go-meal-public-school-pandemic-food-insecurity.
[2] New York State Department of Agriculture and Markets. “New York State Provides Additional Details on $25 Million in Funding for Food Banks Through Nourish New York Initiative.” Accessed February 2021. https://agriculture.ny.gov/news/new-york-state-provides-additional-details-25-million-funding-food-banks-through-nourish-new.

Household Spending Sources

  • When asked which methods their households had used in the past week to meet spending needs, more than 50% of food-scarce households in December reported they had borrowed money from friends or family, 42% reported they used credit cards or loans, and 37% reported they used money from savings or selling assets (see Figure 9). These were the three most commonly reported spending methods.
  • These spending methods are likely to set food scarce households up for further financial hardship in the future. These methods were also more commonly reported than public benefit programs designed to support low-income households, like unemployment benefit payments and Supplemental Nutrition Assistance Benefits (SNAP) (see Figure 9). The findings suggest that these programs are not meeting the food assistance needs of many New Yorkers. Furthermore, many New Yorkers in need of food assistance may not qualify for these programs or may face barriers to enrollment.

Methods

Data

The data used for the analysis are part of the COVID-19 Household Pulse Survey, an experimental data product designed by the U.S. Census Bureau in collaboration with multiple federal agencies. The data are available from:

U.S. Census Bureau. “Household Pulse Survey Public Use File.” Accessed February 2021. https://www.census.gov/programs-surveys/household-pulse-survey/datasets.html.

The survey is designed to provide near real-time data on household experiences during the coronavirus pandemic across all states to inform federal and state recovery planning. Phase one of data collection began on April 23, 2020, and was generally conducted on a weekly basis until July 21, 2020. Phase two of data collection began on August 19, 2020, and was generally conducted every two weeks until October 26, 2020. Phase three of data collection began on October 28, 2020 and is currently collecting data generally every two weeks until March 1, 2021. Households were contacted via e-mail and/or a mobile phone number to complete an internet questionnaire.

Only adults were surveyed. The Census Bureau drew the sampling frame from the Census Bureau Master Address File, supplemented by the Census Bureau Contact Frame. The Census Bureau weighted the survey responses to account for nonresponse. This weighting also adjusted the survey responses to be more representative of demographic distributions—including by educational attainment, sex, age, and race and ethnicity—in each state. Weighted data were used in the analysis based on the weights provided by the Census Bureau.

The unweighted counts of weekly responses in New York State are displayed in Table 1 below. More information on the survey design, including the survey instruments, is available from:

U.S. Census Bureau. “Household Pulse Survey Technical Documentation.” Accessed February 2021. https://www.census.gov/programs-surveys/household-pulse-survey/technical-documentation.html.

Methods

Calculation of Rates

Throughout the report, rates are calculated as the percentage of the applicable (weighted) population that self-reported a certain attribute (e.g., being food scarce or having experienced a loss of employment income). Rates exclude individuals who did not respond to the question, as recommended by the Census Bureau. Although the survey was conducted on a weekly (during phase one) or biweekly (during phases two and three) basis, and results were reported as such by the Census Bureau, we reported results that combined multiple time periods of the survey. That is, we created monthly estimates by averaging the weekly or biweekly data approximately corresponding to each month (see Table 2 for the survey weeks assigned to each month). This was done to increase the sample size associated with each estimate and smooth out weekly or biweekly variation in estimates that may be a result of small sample sizes, particularly when reporting results on subgroups of the population.

Note: Week 1 and week 13 are not included in the analysis because they do not correspond to a full-month period. Data collection for week 1 took place April 23–May 5, 2020, and data collection for week 13 took place August 19–August 31, 2020.

Responses for each of the month periods were combined and averaged without any additional weighting. That is, aside from the application of the survey weights supplied by the Census Bureau to make the results more generalizable to the full New York State population, no additional weighting was applied in the development of the rates for this analysis. More information on the weighting methodology used by the Census Bureau is available from:

U.S. Census Bureau. “Household Pulse Survey Technical Documentation.” Accessed February 2021. https://www.census.gov/programs-surveys/household-pulse-survey/technical-documentation.html.

Methods

Categorizations

Adults who responded that their household “sometimes” did not have enough to eat or “often” did not have enough to eat in the prior seven days were categorized as food scarce. Adults who responded that their households had “enough of the kinds of food wanted” or “enough, but not always the kinds of food wanted” prior to March 13, 2020, were categorized as food sufficient prior to March 13, 2020. Children were categorized as food scarce if an adult responded that the children in the household were “often” or “sometimes” not eating enough in the prior seven days because the household could not afford enough food.

Note that the food scarcity question on the Household Pulse Survey asks about the availability of food within the last week. A common approach for assessing food scarcity issues—typically referred to as food insecurity—asks about the availability of food over the prior 12 months. For example, the USDA reports such results, which are also used by food advocacy organizations such as Feeding America.[1] The USDA food insecurity research takes into account several facets of food availability, including cost barriers, consistency of food access issues, and the impact on mental and physical wellbeing. Given these differences in methods, it can be expected that the food scarcity rates reported here would differ from more traditional food insecurity estimates.

To analyze adults by race/ethnicity, we categorized adults with an ethnicity of Hispanic identified in the data as Hispanic. We categorized adults with an ethnicity of Non-Hispanic as Black only, Asian only, or white only, according to their race code identified in the data. Adults with a race identified in the data as “Other or two or more races” were excluded from analyses by race/ethnicity as a result of low counts.

Adults were categorized into age groups based on the birth year provided in the data. Because month and date of birth were not collected from survey respondents, ages were treated as the respondent’s age as of December 31, 2020.

Household income is defined in the survey as total 2019 household income before taxes. Loss of household employment income since the start of the pandemic is defined as an adult or someone in their household experiencing a loss of employment income since March 13, 2020. Free meal access points are defined in greater detail in Table 3.

[1] Alisha Coleman-Jensen, Matthew P. Rabbitt, Christian A. Gregory, and Anita Singh, “Household Food Security in the United States in 2018.” United States Department of Agriculture: Economic Research Report Number 270. Accessed February 2021. https://www.ers.usda.gov/webdocs/publications/94849/err-270.pdf.

Limitations

Confidence intervals are not provided with the estimates. Although we attempted to improve the reliability of estimates by developing estimates based on multiple time frames of the survey, readers should interpret the precision of the estimates with caution, particularly those for subgroups of the New York State population. Rather than focusing on specific point estimates, these data are most useful for understanding the persistence of patterns over time, including the identification of changes in the direction of trends (e.g., persistent increases followed by persistent decreases), and the relativeness of estimates of one group in comparison with another (e.g., persistent patterns of differences in estimates by race).

As with most surveys, biases can occur in the survey estimates. The Census Bureau has identified certain biases as a result of measurement error, coverage error, nonresponse error, and processing errors that could have occurred in the administration of the COVID-19 Household Pulse Survey. For more details see the Limitations section at:

National Center for Health Statistics. “Household Pulse Survey.” Accessed February 2021. https://www.cdc.gov/nchs/covid19/pulse/mental-health.htm.

Some of the errors may have been more likely to occur because the COVID-19 Household Pulse Survey was meant to provide near real-time information during the pandemic. This meant there was limited time for testing questions to help ensure that survey questions were consistently clear to respondents. For example, it may be possible that respondents did not have a common understanding of what “before March 13th” meant for certain questions (i.e., immediately before versus anytime in the past). As another example, respondents who received free food through school-based meal programs in New York City via delivery may have categorized such programs as either a school-based program, a home-delivery program, or both. Processing errors (e.g., incorrect coding of data) may have also been more likely because of the rapid timeline.

Coverage error may have occurred as households were invited to participate in the survey via cellphone and e-mail. New Yorkers without cellphones, computers, or internet access therefore may have been underrepresented. For example, this likely influenced the low response for “shelter” as a food access point, as New Yorkers in shelters may be less likely to have access to this technology.

Also, the response rate for the Household Pulse Survey was substantially lower than many other federally sponsored surveys, which would make it more susceptible to nonresponse error. Although the federal government employs quality-control procedures to minimize certain biases, the extent of such biases has not yet been evaluated for the COVID-19 Household Pulse Survey.

Methods

Data

The data used for the analysis are part of the COVID-19 Household Pulse Survey, an experimental data product designed by the U.S. Census Bureau in collaboration with multiple federal agencies. The data are available from:

U.S. Census Bureau. “Household Pulse Survey Public Use File.” Accessed February 2021. https://www.census.gov/programs-surveys/household-pulse-survey/datasets.html.

The survey is designed to provide near real-time data on household experiences during the coronavirus pandemic across all states to inform federal and state recovery planning. Phase one of data collection began on April 23, 2020, and was generally conducted on a weekly basis until July 21, 2020. Phase two of data collection began on August 19, 2020, and was generally conducted every two weeks until October 26, 2020. Phase three of data collection began on October 28, 2020 and is currently collecting data generally every two weeks until March 1, 2021. Households were contacted via e-mail and/or a mobile phone number to complete an internet questionnaire.

Only adults were surveyed. The Census Bureau drew the sampling frame from the Census Bureau Master Address File, supplemented by the Census Bureau Contact Frame. The Census Bureau weighted the survey responses to account for nonresponse. This weighting also adjusted the survey responses to be more representative of demographic distributions—including by educational attainment, sex, age, and race and ethnicity—in each state. Weighted data were used in the analysis based on the weights provided by the Census Bureau.

The unweighted counts of weekly responses in New York State are displayed in Table 1 below. More information on the survey design, including the survey instruments, is available from:

U.S. Census Bureau. “Household Pulse Survey Technical Documentation.” Accessed February 2021. https://www.census.gov/programs-surveys/household-pulse-survey/technical-documentation.html.

Methods

Calculation of Rates

Throughout the report, rates are calculated as the percentage of the applicable (weighted) population that self-reported a certain attribute (e.g., being food scarce or having experienced a loss of employment income). Rates exclude individuals who did not respond to the question, as recommended by the Census Bureau. Although the survey was conducted on a weekly (during phase one) or biweekly (during phases two and three) basis, and results were reported as such by the Census Bureau, we reported results that combined multiple time periods of the survey. That is, we created monthly estimates by averaging the weekly or biweekly data approximately corresponding to each month (see Table 2 for the survey weeks assigned to each month). This was done to increase the sample size associated with each estimate and smooth out weekly or biweekly variation in estimates that may be a result of small sample sizes, particularly when reporting results on subgroups of the population.

Note: Week 1 and week 13 are not included in the analysis because they do not correspond to a full-month period. Data collection for week 1 took place April 23–May 5, 2020, and data collection for week 13 took place August 19–August 31, 2020.

Responses for each of the month periods were combined and averaged without any additional weighting. That is, aside from the application of the survey weights supplied by the Census Bureau to make the results more generalizable to the full New York State population, no additional weighting was applied in the development of the rates for this analysis. More information on the weighting methodology used by the Census Bureau is available from:

U.S. Census Bureau. “Household Pulse Survey Technical Documentation.” Accessed February 2021. https://www.census.gov/programs-surveys/household-pulse-survey/technical-documentation.html.

Methods

Categorizations

Adults who responded that their household “sometimes” did not have enough to eat or “often” did not have enough to eat in the prior seven days were categorized as food scarce. Adults who responded that their households had “enough of the kinds of food wanted” or “enough, but not always the kinds of food wanted” prior to March 13, 2020, were categorized as food sufficient prior to March 13, 2020. Children were categorized as food scarce if an adult responded that the children in the household were “often” or “sometimes” not eating enough in the prior seven days because the household could not afford enough food.

Note that the food scarcity question on the Household Pulse Survey asks about the availability of food within the last week. A common approach for assessing food scarcity issues—typically referred to as food insecurity—asks about the availability of food over the prior 12 months. For example, the USDA reports such results, which are also used by food advocacy organizations such as Feeding America.[1] The USDA food insecurity research takes into account several facets of food availability, including cost barriers, consistency of food access issues, and the impact on mental and physical wellbeing. Given these differences in methods, it can be expected that the food scarcity rates reported here would differ from more traditional food insecurity estimates.

To analyze adults by race/ethnicity, we categorized adults with an ethnicity of Hispanic identified in the data as Hispanic. We categorized adults with an ethnicity of Non-Hispanic as Black only, Asian only, or white only, according to their race code identified in the data. Adults with a race identified in the data as “Other or two or more races” were excluded from analyses by race/ethnicity as a result of low counts.

Adults were categorized into age groups based on the birth year provided in the data. Because month and date of birth were not collected from survey respondents, ages were treated as the respondent’s age as of December 31, 2020.

Household income is defined in the survey as total 2019 household income before taxes. Loss of household employment income since the start of the pandemic is defined as an adult or someone in their household experiencing a loss of employment income since March 13, 2020. Free meal access points are defined in greater detail in Table 3.

[1] Alisha Coleman-Jensen, Matthew P. Rabbitt, Christian A. Gregory, and Anita Singh, “Household Food Security in the United States in 2018.” United States Department of Agriculture: Economic Research Report Number 270. Accessed February 2021. https://www.ers.usda.gov/webdocs/publications/94849/err-270.pdf.

Limitations

Confidence intervals are not provided with the estimates. Although we attempted to improve the reliability of estimates by developing estimates based on multiple time frames of the survey, readers should interpret the precision of the estimates with caution, particularly those for subgroups of the New York State population. Rather than focusing on specific point estimates, these data are most useful for understanding the persistence of patterns over time, including the identification of changes in the direction of trends (e.g., persistent increases followed by persistent decreases), and the relativeness of estimates of one group in comparison with another (e.g., persistent patterns of differences in estimates by race).

As with most surveys, biases can occur in the survey estimates. The Census Bureau has identified certain biases as a result of measurement error, coverage error, nonresponse error, and processing errors that could have occurred in the administration of the COVID-19 Household Pulse Survey. For more details see the Limitations section at:

National Center for Health Statistics. “Household Pulse Survey.” Accessed February 2021. https://www.cdc.gov/nchs/covid19/pulse/mental-health.htm.

Some of the errors may have been more likely to occur because the COVID-19 Household Pulse Survey was meant to provide near real-time information during the pandemic. This meant there was limited time for testing questions to help ensure that survey questions were consistently clear to respondents. For example, it may be possible that respondents did not have a common understanding of what “before March 13th” meant for certain questions (i.e., immediately before versus anytime in the past). As another example, respondents who received free food through school-based meal programs in New York City via delivery may have categorized such programs as either a school-based program, a home-delivery program, or both. Processing errors (e.g., incorrect coding of data) may have also been more likely because of the rapid timeline.

Coverage error may have occurred as households were invited to participate in the survey via cellphone and e-mail. New Yorkers without cellphones, computers, or internet access therefore may have been underrepresented. For example, this likely influenced the low response for “shelter” as a food access point, as New Yorkers in shelters may be less likely to have access to this technology.

Also, the response rate for the Household Pulse Survey was substantially lower than many other federally sponsored surveys, which would make it more susceptible to nonresponse error. Although the federal government employs quality-control procedures to minimize certain biases, the extent of such biases has not yet been evaluated for the COVID-19 Household Pulse Survey.