The American Time Use Survey Data Extract Builder (ATUS-X) is a project dedicated to making it easy for researchers to use data from the American Time Use Survey (ATUS). The ATUS is an ongoing time diary study, started in 2003, that is funded by the U.S. Bureau of Labor Statistics (BLS) and fielded by the U.S. Census Bureau. It provides detailed information about the activities in which respondents engage together with extensive information about the characteristics of those respondents and other members of their households. For more information about the ATUS, please click here.
The goals of the ATUS-X project are:
Core funding for the ATUS-X is provided by the National Institute of Child Health and Human Development, National Institutes of Health. The Economic Research Service, U.S. Department of Agriculture has provided additional support.
Advantages of Using the ATUS-X
Time diary data allow researchers to look at and analyze the time people spend doing different things. The ATUS collects information on time spent in each of more than 400 detailed activities. Each ATUS respondent reports on all of his or her activities during a designated diary day. Information on personal and household characteristics also is collected during the ATUS interview. The sample for the ATUS is drawn from households that have completed their participation in the Current Population Survey (CPS) and additional information about ATUS households is available from their CPS interviews. The BLS provides a set of data files that contain information collected during the ATUS interviews along with information collected during sample households' final CPS interviews. These data files are complex. Considerable programming effort may be required to create even relatively simple time use variables and to produce data files suitable for analysis.
The ATUS-X data extract builder is designed to make it easy for users to create data files that contain the time use, personal characteristic and household characteristic variables they want, thus making the data more accessible to a broader audience. The system aggregates time spent in user-specified combinations of activities, applies any requested filters behind the scenes and delivers this information to users in the form of time use variables. Time use variables contain the number of minutes ATUS respondents spent in the user-specified activities and filters during the interview day. Individual and household characteristic variables that a user desires also may be selected. The output from the ATUS-X is a data file consisting of person-level records that contain the variables a user has requested. A statistical software package such as SAS, Stata or SPSS must be used to prepare tabulations or carry out other analyses using the data file.
Registering as a System User
If you are a new ATUS-X user, you must register prior to entering the Data Extract Builder. This is the first step in creating a data extract. To register, click on Register as a New User. You will be asked to provide a bit of personal information, research interests, and to select a password. Registration is easy and you will be asked to register only once. The email address that you enter when you register will be used to login into the system and to let you know when an extract that you have requested is ready to be downloaded.
There is no charge to use the ATUS Data Extract Builder, but we do ask that users of the system notify us by email of any publications, research reports, presentations, or educational material making use of ATUS-X data or documentation. Users should also appropriately cite data obtained from the ATUS-X. Continued funding for the ATUS-X depends on our ability to show our sponsor agencies that the data are being used for productive purposes.
Once you have registered, return to the home page and click on Build an Extract. From there, you will have the opportunity to specify the sample for your extract; define and select time use variables; select personal and household characteristic variables; and request your extract.
Selecting a Sample
ATUS data are released annually; at the present time, data for 2003 through 2006 are available. Approximately 20,000 ATUS interviews were completed in 2003; approximately 14,000 interviews were completed in each year from 2004 through 2006. Interviews were conducted throughout each of the years. Users may specify that their sample should include data from a single year or from any combination of years.
Potential respondents to the ATUS were drawn from the outgoing rotation group of the Current Population Survey (CPS). The original ATUS public use files contain information on all sampled individuals and their households whether or not the sampled individual participated in the ATUS. To serve the broadest possible audience, we have preserved much of the original detail in the public use files, including the information provided for other members of ATUS respondents' households and for members of CPS households whose designated respondent did not participate in the ATUS.
Users are given a choice of four possible groups of people to be included in their data extract:
Time use information is available only for ATUS respondents. For this reason, we expect that most users will select the "ATUS respondents" samples and we have made this the default option.
A user who wishes to use information about other household members to construct household characteristic variables to add to the data records for the ATUS respondents should select the "ATUS respondents and members of their households" sample. Before making this selection, however, we advise users to look at the list of household characteristic variables available for selection from the list of demographic variables, as a suitable variable may be available there.
The Census Bureau has constructed weights intended for use in producing estimates from the ATUS data. These estimation weights take into account the differing probabilities of selection for sample members with different characteristics, the over-representation of Saturdays and Sundays in the designation of respondents' diary days, and the differing response rates for sample members with different characteristics. There may be some users who wish to adjust the Census Bureau weights or produce alternative weights based on more detailed information about the characteristics of survey respondents and nonrespondents. For this reason, we offer the option of selecting samples that include records for ATUS nonrespondents or for ATUS nonrespondents together with other members of their households. Users who select samples that include ATUS nonrespondents will likely want to select the ATUS base weights (BWT) rather than the ATUS final weights (WT06). We expect few users to be interested in samples that include ATUS nonrespondents.
Creating and Selecting Time Use Variables
The next step in the data extract process is to create and/or select any time use variables the user desires. A time use variable returns the number of minutes meeting specified criteria that a user reports during their designated day. Each designated day includes a 24-hour period that begins at 4 a.m. and ends at 4 a.m. the following day. Time use variables may be defined with reference to the specific activity(ies) involved, the time of day that the activity(ies) occurred, the location of the activity(ies), and/or whether the respondent did or did not have a child under the age of 13 in their care during the time interval. In the next version of the Data Extract Builder, users also will be able to define time use variables with reference to other people who may have been in the company of the respondent during the activity(ies).
Several basic time use variable options are available: selecting a time use variable that already exists; modifying an existing time use variable; or creating a new time use variable from scratch.
Selecting an Existing Variable. The first time that a data user enters the Data Extract Builder, the existing time use variables available for selection will include time use variables defined within the ATUS activity coding structure and time use variables used by the BLS in its survey publications. The full ATUS coding structure is a 3-tier, 6-digit system. The following seventeen 2-digit activities constitute the coding structure's major groupings:
Each of the major activity groupings is further disaggregated with additional 2-digit intermediate codes (more than 100 total) and finally with a 2-digit detailed activity groupings (more than 400 total). Time use variables defined at any of the three levels of aggregation in the activity coding structure may be selected for inclusion in a user's data extract. For further details about the activity coding structure, please click here.
The BLS uses a slightly different classification structure for its publications that include ATUS data. One major difference between the publication classification structure and the activity coding classification structure is that, for publication purposes, the BLS includes time spent traveling in connection with another activity together with the time spent actually engaged in the activity. For example, the publication category "Eating and drinking" includes time spent in travel related to eating and drinking, such as driving to a restaurant. Another difference is that several of the publication categories combine categories from the activity coding structure. For example, the publication category "Leisure and sports" combines the activity coding structure categories of "Socializing, relaxing and leisure" and "Sports, exercise and recreation." Further details can be found in this bridge.
In addition to selecting time use variables defined under either the activity coding structure or the BLS publication structure, data users also have the option of selecting any time use variable created on a previous visit to the Data Extract Builder. These previously-created variables will be listed on the same page as the pre-loaded time use variables available for selection.
Modifying an Existing Time Use Variable. Another option for data users is to modify an existing time use variable. Variables may be modified by adding or deleting activities from a pre-loaded variable or by applying filters to the variable specification.
Currently available filters restrict the activities selected to those occurring in certain locations or at certain times of day, or to activities that involve or do not involve having a child under the age of 13 in the respondent's care (secondary child care). For example, if you were interested in studying shift work, you might create two time use variables by combining Work and Work Related activities with a time filter for two paid work shifts: first shift (7 a.m. to 3 p.m.) and second shift (3 p.m. to 11 p.m.). You could then compare the personal and job characteristics of individuals who work during the first shift with those who work during the second shift. In the next version of the Data Extract Builder, scheduled for release in June 2009, we will include two additional filters. The first will allow users to apply filters that consider the amount of time spent with other household members or with particular groups of non-household members, and the other will differentiate time engaged in food and eating-related secondary activities.
Once any desired modifications have been applied to the existing variable, the user is asked to name the new variable and provide a short description that is retained for later reference. The system automatically stores the activity code and filter selections used to create the modified time use variable. These specifications may be referenced at a later date if there are questions about exactly how the modified variable was defined.
Creating a New Time Use Variable. In some cases, none of the existing time use variables may be similar to the variable a user wishes to create. In this case, users may create new time use variables from scratch. If this option is chosen, the user will be asked to choose the activities to be included from the activity coding structure and to choose any filters for time of day, location or secondary child care to be applied. In the next version of the Data Extract Builder, scheduled for release in June 2009, users will have the option of applying filters related to with whom an activity occurred or eating and drinking as a secondary activity. Once a variable has been specified, the user is asked to name the variable and provide a short description that is retained for later reference. The system automatically stores the activity codes and filter selections used in creating a new time use variable, in case there are questions later about exactly how the variable was defined.
As already noted, any time use variables that a user creates are added to the list of time use variables that may be selected for future data extracts. At the present time, users do not have the option of deleting previously-created time use variables from their list. That option will be added in the future. At present, the system also does not support the sharing of time use variables created by one user with other system users.
Selecting Individual and Household Characteristic Variables
In addition to any time use variables they may create or select, users typically also will include personal and household characteristic variables on their data extract. Personal characteristics include such things as the respondent's age, sex, marital status, labor force status or earnings. Household characteristics include such things as household size, whether there are children resident in the household, and whether the household owns or rents its dwelling. For a complete list of the personal and household characteristic variables available to system users, please click here.
A description of each characteristic variable may be obtained by clicking on the variable's name on the selection list. For categorical variables, codes and frequency counts for each year they exist in the data are also provided. In some cases, the coding of the variables changed across survey years. In the Data Extract Builder, uniform codes are assigned across all of the data collection years wherever possible.
The names assigned to the ATUS-X variables are different from the names of the variables on the original ATUS data files. An effort was made to assign intuitive names that are as simple as possible (for example, SEX rather than PESEX). Some of the ATUS-X characteristic variables were collected or updated at the time of the ATUS interview; others were collected at the time of the CPS interview that occurred 2-5 months earlier. In most cases, the suffix _CPS8 is attached to the names of variables collected during the CPS interview. For a very few variables that were collected at the time of the CPS interview but contain information that should not have changed, the _CPS8 suffix is omitted. For example, information on whether the respondent is an immigrant and the countries in which the respondent's mother and father were born are collected as part of the CPS; the variables containing this information do not have a _CPS8 suffix, but their origin is noted in the accompanying variable description.
In addition to the characteristic variables themselves, the data file contains data quality flags for many of the variables that indicate whether the value coded for the variable was directly reported or imputed and, if the latter, the source of the imputed information. In a few cases, other data quality flags that contain information about a variable are available. System users are given the option to include or not to include available data quality flags on their data extract. If the flag option is selected , all data quality flags for the variables included in the data extract will be delivered as part of the data file.
Some data users may wish to link the extracts they create with other ATUS data files or with information from CPS interviews completed by the ATUS households. Linking person level ATUS records requires the use of a household level identifier (CASEIDP in our data, which is the same thing as TUCASEID in the original data) and a person level identifier (LINENO in our data, which is the same thing as TULINENO in the original data). These identifiers are selected automatically when a user creates a data extract. To link a data extract with CPS data files, additional identifiers are required; for details, please click here.
Many data users will require estimation weights for analyzing the data they extract. The methodology used to construct estimation weights has been updated in each year of the ATUS. The methodology applied for the 2006 survey year is more refined than that applied in earlier years. Weights using the 2006 methodology were calculated and added to the 2003, 2004 and 2005 data files. In our data, this variable is named WT06. WT06 is the preferred weighting variable to use for most analyses and the only weighting variable that is suitable for use in analyses that combine 2006 data with data for other years. For additional details, please see the variable description for WT06.
Because the sample for the ATUS is not a simple random sample, standard errors for estimated means and model parameters calculated using standard statistical software will not be correct. Most statistical software packages have special procedures for correctly calculating the standard errors for estimates based on data from surveys such as the ATUS that have complex designs. The original ATUS data files contain replicate weights that can be used for this purpose. If a user selects RWT06, the 160 replicate weights that are intended for use together with WT06 will be added to the user's data file. Please note that including RWT06 will result in a data file that is relatively large.
Obtaining a Data Extract
When you have finished defining and selecting the variables to be included on your data extract, the next step is to create the extract. At the present time, all ATUS-X data extracts consist of data files that have the person as the unit of observation. The ATUS collects information on households, people and activities. The extraction system rectangularizes the data: That is, it puts household information on the person records and aggregates time spent in user-defined activities and puts time use variables on the person records. The rectangularized data file does not retain households or activities as separate records. No information is lost by rectangularizing the data and this is the format preferred by most researchers. In the future, the extraction system will include the option of requesting data in different formats.
The ATUS-X data files are produced in a fixed-column ASCII format. In addition to the ASCII data file, the system creates a statistical package syntax file to accompany each extract. The syntax file is designed to read in the ASCII data while applying appropriate variable and value labels. SPSS, SAS, and Stata are supported. You must download the syntax file with the extract or you will be unable to read the data. The syntax file may require minor editing to identify the location of the data file on your local computer. Detailed instructions are available here.
A codebook file also is created with each extract. It records the characteristics of your extract and should be downloaded for your records.
Data extracts are generated on our server. The system sends out an email message to the user when the extract is completed. The user must download the extract and analyze it on their local machine. The time needed to make an extract depends on the number and size of samples requested and the load on our server. Extracts can take from a few minutes to an hour or more to prepare. There is no need to stay active on the ATUS-X site while the extract is being made; you will be notified by email when it is ready for download.
Because of space constraints, extracts are retained on the system server only for a limited period of time. Extract specifications, however, are retained indefinitely. This means that it is always possible for a user to recreate a previously-created extract. To access the specifications for previously-created extracts that you with to recreate, enter the Data Extract Builder and click on "My Extracts."
Future Plans for ATUS-X
ATUS-X is funded through 2011 by a grant from the National Institute of Child Health and Human Development, National Institutes of Health. In the first phase of the project, we have developed a website for that allows data users to aggregate the time spent in combinations of user-defined activities, restricted if desired by time of day, location and/or whether the person had children under the age of 13 in their care during the activity, and to select additional variables that characterize the individual ATUS respondents and their households. We will continue to improve our website and documentation. Over the next three years, we plan an annual data system release each June. In the release scheduled for June 2009, we will add the ability to create time use variables that consider the amount of time spent with other household members or with particular groups of non-household members and time engaged in food and eating related secondary activities. Later planned additions include the addition of the capacity to generate customized weights for use in conjunction with user-defined sample populations (2010) and to link ATUS data with CPS Basic Monthly Survey and Supplement data (2011). If you have a suggestion about other improvements you would like to see, please send us an email.
We have every expectation of continuing the project beyond 2011 provided the American Time Use Survey continues to be funded, but will have to secure further funding as our current grant expires. To be successful, we need to have a large body of users and published works that we can document. Please inform us of any presentations you may give or papers you may write using ATUS data. We also request that you properly cite the use of this data.