Eating and Health Module (ATUS): Documentation and Methodology

Eating and Health Module (ATUS) Current Findings History of USDA Time Use Studies Documentation • User's Guide • Module Questions Related Resources Quick Facts

Survey Objectives

The objective of the Eating and Health (EH) Module of the American Time Use Survey (ATUS) is to collect data to analyze the relationships among time use patterns and eating patterns, nutrition, and obesity; food and nutrition assistance programs; and grocery shopping and meal preparation.

One of the missions of USDA's Economic Research Service (ERS) is to enhance the understanding of economic issues related to the nutrition and health of the U.S. population. Data collection and research on eating patterns, Body Mass Index (BMI), food and nutrition assistance program participation, program income eligibility, grocery shopping, and meal preparation all contribute to this goal. Specifically, the economic analysis of decisions made under constraints—in this case, time—provides insight for both policies and programs because the decisions individuals make on how to use their 24 hours in a day have short- and long-run implications for income and earnings, health, and other aspects of well-being.

The Division of Cancer Control and Population Sciences of the National Cancer Institute (NCI), ERS's funding partner, has a long history of research and surveillance activities concerning health behaviors, such as diet, weight, and physical activity, that are linked to multiple health outcomes, including cardiovascular disease, diabetes, hypertension, and stroke. Recent evidence indicates that obesity and sedentary behavior are risk factors for cancer.

Much of NCI's research has been based on data obtained from standardized health surveys, such as the National Health Interview Survey (NHIS) and the National Health and Nutrition Examination Survey (NHANES). It is difficult or impossible to use such data to explore the social and environmental context of various health behaviors or to explore tradeoffs associated with time limitations and adopting healthful diets or adding physical activity. Time use survey data, however, enable researchers to investigate these and other areas. NCI's objective in providing support for the Eating and Health Module of the American Time Use Survey is to obtain further information on the temporal, economic, social, and environmental correlates of selected health behaviors. This information will help inform efforts to design effective interventions to improve health behaviors at the individual and population levels.

Module questions

The EH Module asks ATUS respondents about secondary eating and drinking; grocery shopping and meal preparation; Food Stamp Program participation; breakfast and lunch obtained at school; general health, height, and weight; and income. A short version of the Module questions is available. A text version of the EH Module survey instrument is also available from the BLS website.

Module microdata

Each year of the Module microdata has four files that can be downloaded from the BLS website.

EH Respondent file

The EH Respondent file contains information on EH respondents, including variables about grocery shopping, meal preparation, food stamp participation, general health, height and weight, and household income.

EH Activity file

The EH Activity file contains information on respondents' secondary eating and secondary drinking beverages.

EH Child file

The EH Child file contains information about which household children (< age 19) ate a breakfast or lunch in the previous week that was prepared and served at a school, day care center, Head Start center, or summer day program.

EH Replicate Weights file

The EH Replicate Weights file contains the replicate weights that can be used to calculate standard errors and variances for EH Module estimates.

Module data dictionary

The data dictionary lists all the variables available on the four microdata files, and their valid values.

Methodology

Estimation of Total Time Spent Eating and Drinking

The American Time Use Survey (ATUS) asks respondents to report only their primary (main) activities in their diary, with the exception of simultaneous child care. The Eating & Health Module of the ATUS asks respondents about secondary eating and secondary drinking beverages, that is, eating and/or drinking beverages while engaged in another activity the respondent considers primary, such as watching television or driving.

Survey Questions

A common response to the secondary drinking question is something like, “I had a cup of coffee on my desk all day at work.” The guidance given to interviewers is to ask, “How much time were you actively drinking coffee?”

As these data are collected, primary eating is exclusive of secondary eating and secondary drinking. There is no overlap in time between the primary and secondary activities.

Before discussing estimation techniques for total time spent on secondary eating and drinking, it is instructive to examine characteristics of the data and components of eating and drinking.

Characteristics of Primary Eating and Drinking Data

Primary eating and drinking per person on an average day in 2006 was as follows:

Ninety-six percent of Americans age 15 or older engaged in primary eating and drinking at some time on an average day. Eating/drinking occurrences constitute about 10 percent of all activities. (This is by occurrence, and not weighted by time spent in the activity.) This averages to 2.03 primary eating/drinking occurrences per respondent, and an estimated 2.06 occurrences on an average day across the U.S. population age 15 and older.

Characteristics of Secondary Eating and Secondary Drinking Data

Secondary eating and drinking per person on an average day in 2006 was as follows:

Note that the data used in the above table are summary variables over the entire day. So, the 33 percent who reported both secondary eating and secondary drinking may or may not have engaged in secondary eating and secondary drinking during the same primary activity. Time diaries reveal information about secondary eating and secondary drinking occurrences.

Occurrences during all primary activities on an average day were as follows:

Secondary eating and/or secondary drinking occurred during about 6 percent of all activities. There were 1.2 secondary eating/drinking occurrences per person on an average day.

Only primary activities with secondary eating and/or secondary drinking were as follows:

So, only 20 percent of secondary eating and secondary drinking occurrences had both secondary eating and secondary drinking.

Duration of secondary eating and secondary drinking was as follows:

If a respondent reports both secondary eating and secondary drinking, it is not known whether both activities happened at the same time. Consequently, estimating total time spent on secondary eating and drinking is not a simple sum. However, the fact that only 20 percent of primary activities with secondary eating or secondary drinking had both, and the fact that the means and modes of these two variables were relatively small allowed for considering a method of estimating total time spent on secondary eating and secondary drinking.

Estimating Total Time on Secondary Eating and Drinking

ERS’s methodology for estimating total time spent on secondary eating and drinking is illustrated in the following example. A respondent spent 5 minutes engaged in secondary eating (EUEDUR24) and 15 minutes engaged in secondary drinking (EUDDUR24) while doing paid work between 1 p.m. and 5 p.m. (4 hours). It is not known if there was any overlap of the 5 minutes of eating with the 15 minutes of drinking, only that both happened during paid work time from 1 p.m. to 5 p.m. Later in the day, the respondent spent 2 hours watching television from 8 p.m. to 10 p.m. and reported engaging in secondary eating and secondary drinking the entire 2 hours (120 minutes).

EUEDUR24 (amount of time spent in secondary eating) and EUDDUR24 (amount of time spent in secondary drinking) cannot be totaled into a simple sum because first, it is not known if there is overlap, and second, adding the two time durations may result in more time than the total length of the activity. In the example, adding 120 minutes of secondary eating with 120 minutes of secondary drinking during the time the respondent watched television would result in more than 2 hours and could result in a day more than 24 hours long.

To estimate the total time spent on secondary eating and secondary drinking, ERS first estimated a minimum and maximum value where the minimum value assumes that secondary eating and secondary drinking occur at the same time, that is, with full overlap. The maximum value assumes no overlap of the two activities, or as little overlap as possible, but the duration is not longer than the primary activity.

Returning to the example in which the respondent engaged in secondary eating and secondary drinking during 4 hours of paid work, the minimum time (all overlap) is 15 minutes and the maximum is 20 minutes (the sum of secondary eating and secondary drinking). This calculation is straightforward as both 15 minutes and 20 minutes are less than the duration of the primary activity (4 hours). For the activity of watching television, the duration of the primary activity constrains both the minimum and maximum to 120 minutes (2 hours).

The following table provides estimates for the minimum and maximum of secondary eating and secondary drinking times:

Note that estimates are both for the entire population age 15 and older and for participants only (those who engaged in secondary eating or secondary drinking). Population estimates include those who do not engage in secondary eating and drinking, so there are zeros added into the calculation, whereas the participants-only mean is calculated without zeros.

The minimum and maximum estimates are close—a 1.7-minute difference for the population and a 3.2-minute difference for participants. ERS looked at standard errors and calculated the 90-percent confidence interval for each estimate as a consideration in using the midpoint of the minimum and maximum as the estimate of the total time:

The minimum and maximum estimates of total time are not statistically different at the 90-percent confidence level. This is true for both population estimates and participant estimates. For the present analysis, the midpoint was used to estimate total time spent on secondary eating/drinking.

The ERS estimate of total time spent on secondary eating and secondary drinking is then the midpoint of the minimum and maximum values presented above.

Programming Notes—Estimating Total Time in Secondary Eating and Drinking Activities

Based on the methodology, estimates were programmed in SAS 9.1, using the ATUS Activity file and the EH Activity file data. The minimum, maximum, and midpoint values for secondary eating/drinking were determined in the following manner.

Data files needed:

2006 ATUS Activity file and 2006 EH Activity file. These files can be linked by the variables TUCASEID and TUACTIVITY_N.

Variables needed:

Steps

Total Time Spent in Eating and Drinking Activities

In order to estimate the total time spent in eating and drinking activities on an average day, ERS added the mean time spent on primary eating and drinking, the mean time spent on associated activities (waiting associated with eating/drinking and travel related to eating/drinking), and the midpoint of the minimum and maximum of the total time spent on secondary eating and secondary drinking. See the EH Module Users Guide for a discussion on the detailed activities included in primary eating and drinking and in associated activities.

Consideration was given to whether it would be conceptually appropriate to sum primary eating and drinking time with secondary eating and secondary drinking time. One may argue that primary and secondary activities are fundamentally different in intensity and so cannot be combined. In making this decision, the following points were considered:

• The primary eating and drinking times and the secondary eating and secondary drinking times are all time durations only. Nothing is known about calories consumed nor about intensity of the activities. Research using this data will determine whether or not there are relationships between primary eating/drinking time, secondary eating time, secondary drinking time, and BMI.



• The focus is on eating and drinking activities and not analysis of all activities over the course of a day.

To analyze the constraints of a 24-hour day, however, only primary activities should be used, as summing primary and secondary activities would result in a day greater than 24 hours. So, the appropriateness of using a total time in eating/drinking activities depends on the application.

Looking at the data, it can be seen that respondents appear to be conscientious in their reporting of secondary eating and drinking times:

Secondary eating occurrences:

Secondary drinking occurrences:

Only 13.5 percent of secondary eating occurrences were reported as being for the entire time of the primary activity. Slightly over half, 52 percent, of secondary drinking occurrences were reported as being for the entire time of the primary activity. In both cases, the mean times of the primary activities are considerably larger than the mean times of secondary eating and secondary drinking.

Certainly, the quickest way out of the question Were you eating the entire time you were [fill: ACTIVITY]? would be to respond “yes.” By just looking at the data it appears that respondents are trying to be accurate and are putting thought into their reported actual time spent on secondary eating and in secondary drinking. And, a response of “yes” for the entire time of primary activity is a valid response. Because the EH Module asks respondents the duration of the secondary activity, no adjustments are made to the secondary eating and secondary drinking time durations.

No survey or data collection effort is perfect, however, the ATUS and the EH Module are obtaining the only nationally representative time use survey containing estimates on primary and secondary eating. For applications analyzing eating patterns, ERS concludes that estimating a total time of secondary eating and drinking and a total time in all eating and drinking is appropriate. These methods have been subjected to outside review, and ERS will continue to review and refine the methodology.