Sponsored by the National Institute on Aging                                                                                                           November 24, 2009
Main page for MIDUS website
Information on MIDUS I
Information on MIDUS II
Products using the MIDUS Data
MIDUS Data
MIDUS Pilot Program
Findings From MIDUS Study
Other links of interest


MIDUS Samples

MIDUS I Samples:  Five distinct, but related samples comprised the original study, these included members of national (random-digit-dialing) sample, along with subjects obtained via over sampling in five metropolitan areas.  In addition, interviews were conducted with siblings of the national sample, plus twins identified in the Twin Screening Project.  Across all samples, respondents included noninstitutionalized, English-speaking adults, 25 to 74 years of age, selected from the working telephone banks in the coterminous United States. The technical report on the methodology of the MIDUS Survey (field procedures, response rates weighting of data) is available from the MIDMAC website (http://midmac.med.harvard.edu/tech.html).  Each of the information is briefly described below.

National Sample.  The National Survey of Mid-Life Development in the U.S. was conducted with a multi-stage sampling design.  At Stage 1, an equal probability sample of telephone numbers was selected.  The sample was drawn by Survey Sampling, Inc.  Predesignated households were then selected in random replicates.  Contact persons were informed that participation would entail completing a telephone interview and two mail questionnaires.  Information was collected from the contact person regarding household composition to determine the availability of a potential respondent meeting the eligibility criteria stated above.  After explaining the study to the informant, a household listing was generated of English-speaking people in the age range 25-74, and a random respondent was selected.  Oversampling of older people and of men was achieved by varying the probability of carrying out the interview at this stage as a joint function of the age and sex of the randomly selected respondent.

Oversamples.  Oversampling was conducted in five metropolitan areas (Boston, Atlanta, Chicago, Phoenix, San Francisco) for the purpose of recruiting samples with whom more intensive interviews could be conducted.  The geographic locations corresponded to the locations of associated investigators.  For example, a large oversample was recruited for the Boston in-depth study of life management.  This investigation provided important pilot research for the cognitive battery that is proposed for MIDUS II.

Siblings.  Siblings were identified by members of the national sample.  Among those who reported that they had one or more siblings, 529 members of the national sample were randomly selected.  Using only siblings within a family that had the same biological mother and father, a group of 951 siblings was identified, thus including more than one sibling in some families.  Siblings of the national sample respondents were recruited with the cooperation of the respondents, who were asked to provide interviewers with their contact information, and to communicate with their siblings about participation prior to the time a recruiter made the contact attempt.  The combined groups of 951 and 529 yielded 1,614 sibling pairs.

Twins.  The 998 twin pairs participating in the MIDUS Twin Screening Project represent the first national sample of twin pairs ascertained randomly via the telephone.  Twins were recruited in a two-stage sampling design.  First, screening of a representative national sample of approximately 50,000 households was conducted to identify the presence of a twin in the family.  This was done as part of ongoing national omnibus surveys.  In the second stage, student recruiters from the University of Michigan contacted the twin households and attempted to recruit twins to participate in the survey.  Cooperating twins were asked to provide contact information for their co-twins.  On occasion, there was more than one twin pair per family.

Response Rates.  The response rate for the MIDUS I telephone interview (RDD sample) was 70%, which is generally considered quite well for a population survey. Among the telephone respondents 86.3% completed the lengthy self-administered questionnaires (SAQ), yielding an overall response rate of 60.8%. Comparison of the MIDUS I sample with the Current Population Survey (CPS, 1995) revealed that the sample under-represented those with a high school education or less and African Americans. Alternatively, it over-represented older males (by intention, to facilitate gender comparisons by age). The representation by gender and marital status was close to the CPS.