Survey Design CPS

I have a question about the CPS survey design. The documentation states that the individuals are interviewed for 4 months then left out for 8 months and then interviewed for another 4 months. During the 4 months of interviews how often are the individuals interviewed, one time or fours times?

Households are interviewed four times during the first four months, once per month (and again four times in the second round, for a total of eight). Note that CPS samples households, so if a given household member leaves or enters the household in the middle of the CPS rotation, that individual will have fewer than eight survey responses. You can find a lot more info about the CPS rotation pattern and the panel aspect of CPS at this page.

Is the individual asked different supplement questions on top of the basic questions each time he or she is interviewed? So If I use basic monthly samples I will have multiple observations per individual but if I use a supplement sample (e.g. ASEC) I will have one observation per individual? How can I check if I have multiple observations or a single observation per individual?

There are two types of samples available on IPUMS CPS:

  1. Basic Monthly samples. These include the standard CPS questions (including labor force) available under the Person/Core and Household/Core variable groups. For all months EXCEPT March, these samples also include data from the corresponding supplement for that month, if there was a supplement. For each of these samples, there is a single record per person.

  2. ASEC samples. ASEC is also known as the March Supplement. This supplement comes in a different file from the March Basic Monthly data. However the ASEC file includes all the information in the March Basic sample (but not vice versa). If you download an extract with both ASEC and March Basic samples, you will have duplicate observations for the individuals in the March Basic sample. You can identify which sample a given record came from using the variable ASECFLAG. The ASEC sample includes additional people, part of the ASEC Oversample.

Regarding duplicate observations: A given individual can be identified using the variable CPSIDP. This is constant across samples. However, a given individual will have multiple observations if they are present in multiple samples. You can identify unique samples using a combination of YEAR, MONTH, and ASECFLAG. A given survey response for a specific individual will only be duplicated if you include both ASEC and March Basic samples in your extract. Note that individuals in the ASEC oversample all have CPSIDP=0, and they are not present in the March Basic sample. These people cannot be linked across months.

Generally I would recommend not creating extracts including both ASEC and Basic Monthly data unless you have a specific reason to do so.

1 Like