National Survey of Family Growth (NSFG)

Which of the groups on Card 2 describe your racial background? Please select one or more groups.
[Interviewers are instructed to enter all that apply. They are to enter all groups that are part of the mixture if the respondent reports a mixture of several races (biracial, mixed, mulatto, etc.).]

• American Indian or Alaska Native (AI/AN)
• Asian
• Native Hawaiian or other Pacific Islander (NH/PI)
• Black or African American
• White

[If respondent selected multiple race groups, interviewer asks this question.]
Which of these groups, that is (Race groups selected) would you say best describes your racial background?

Analysis can be done for the four major census regions (Northeast, Midwest, South, West) and for metropolitan and nonmetropolitan areas. Estimates cannot be made for individual states or for smaller areas.

Cycle 3 (1982): Question F-47:
“Which of the groups on this card best describes your racial background?”
Alaskan Native or American Indian has 83 cases (out of 7,969 women).
Only one race was coded in Cycle 3.

Cycle 4 (1988): Question F-9:
“Which of the groups on card 30 best describe your racial background?” (Code all that apply)
Alaskan Native or American Indian has 238 cases out of 8,450, allowing multiple mentions. The standard RACE recode is shown only as black, white, and other, but the original variable contains more detail.

Cycle 5 (1995): Question IC-3 and 4:
“Which of the groups on Card I-1 best describes your racial background?”
First mention= 344 “Alaskan Native or American Indian.”
2nd mention = 9 “Alaskan Native or American Indian.”
3rd mention = 3 “Alaskan Native or American Indian.”

CDC does not recommend using the AI/AN/NA race identifiers on the Cycle 1 (1973) and Cycle 2 (1976) data sets, as this information was collected differently and should not be combined with later cycles.

While there are no CDC-specific guidelines for pooling data, the agency provides the following suggestions for combining the data:

Researchers can append the data from each cycle of interest and use the year of the survey as an independent variable. For obtaining unbiased standard errors for each cycle:

• For the 2002 and 1995 surveys (Cycles 6 and 5) researchers can use SUDAAN or STATA software.
• For 1988 (Cycle 4) there are Balanced Repeated Replicate (BRR) weights available. STATA version 9 does BRR variance estimation. The command and the column locations of the weights can be obtained from CDC.
• For 1982 (Cycle 3): Replicate weights to allow calculation of valid standard errors for complex sampling were not included on public use files. Instead, generalized variance estimates were developed and published for estimated numbers and percentages.

Additionally, researchers should examine the sampling methodology and question text and response options across the different cycles to decide whether pooling data across the different cycles is feasible.

In the full sample data, the number of NH/PI is very small.

There are 3 ways to access data through the RDC once a project has been approved:

1. “remote access,” in which the user submits a SAS program electronically and the output is screened and returned electronically;
2. by physically going to the RDC at NCHS, and doing the research there, using any software the researcher wants to use; and
3. staff-assisted access, in which the researcher submits a program to an RDC staff member who submits and screens it and returns the output.

There are fees for each type of access.

Research Data Center: (301) 458-4277 or e-mail: rdca@cdc.gov