QMHMAR1 in years 86-88

CPS
1

Hi friends,

I was checking the data quality flag variable QMHMAR1. QMHMAR1 is a data quality flag for MHMAR1M and MHMAR1Y (i.e., year and month of first marriage). This variable is part of the CPS Fertility Supplement.

  1. Why are the number of cases with “allocated” values (QMHMAR1=4) so high in years 86-88?
    I checked the data. Such allocation is not simply about allocating a NIU value of 9999. For example:
bys year: fre mhmar1y if qmhmar1==4
mhmar1y -- Year of first marriage
---------------------------------------------------------------
                  |      Freq.    Percent      Valid       Cum.
------------------+--------------------------------------------
Valid   1956 1956 |          3       0.00       0.00       0.00
        1957 1957 |         17       0.01       0.01       0.02
        1958 1958 |         56       0.05       0.05       0.07
        1959 1959 |        132       0.11       0.11       0.18
        1960 1960 |        252       0.22       0.22       0.40
        1961 1961 |        308       0.27       0.27       0.66
        1962 1962 |        395       0.34       0.34       1.00
        1963 1963 |        489       0.42       0.42       1.42
        1964 1964 |        537       0.46       0.46       1.89
        1965 1965 |        642       0.55       0.55       2.44
        1966 1966 |        718       0.62       0.62       3.06
        1967 1967 |        730       0.63       0.63       3.69
        1968 1968 |        752       0.65       0.65       4.34
        1969 1969 |        916       0.79       0.79       5.13
        1970 1970 |        989       0.85       0.85       5.98
        1971 1971 |        914       0.79       0.79       6.77
        1972 1972 |        956       0.82       0.82       7.59
        1973 1973 |        885       0.76       0.76       8.35
        1974 1974 |        895       0.77       0.77       9.13
        1975 1975 |        877       0.76       0.76       9.88
        1976 1976 |        874       0.75       0.75      10.64
        1977 1977 |        809       0.70       0.70      11.33
        1978 1978 |        862       0.74       0.74      12.08
        1979 1979 |        996       0.86       0.86      12.93
        1980 1980 |       1014       0.87       0.87      13.81
        1981 1981 |        934       0.81       0.81      14.61
        1982 1982 |        938       0.81       0.81      15.42
        1983 1983 |        906       0.78       0.78      16.20
        1984 1984 |        938       0.81       0.81      17.01
        1985 1985 |        899       0.78       0.78      17.79
        1986 1986 |        310       0.27       0.27      18.05
        9999 NIU  |      95054      81.95      81.95     100.00
        Total     |     115997     100.00     100.00           
---------------------------------------------------------------

As you can see, many people were “allocated” seemingly reasonable values for their years of marriage, so it looks like they are not just artificially created values. Are such values in fact real values, not allocated values?

  1. What do the other values mean? We have
    1 - month changed
    2 - month and year changed
    3 - year changed
    According to what rules are they changed? Are such changes based on cell-mean assignment (e.g., mean values of other people from similar demographic groups) or based on other relevant information reported by the respondent such as age at first marriage?
    Thank you!
2

The IPUMS CPS team has looked into the 1986 - 1988 frequencies and agrees that they are inconsistent with frequencies from surrounding samples. There is insufficient evidence, however, for them to determine whether or not there is a coding error in the original CPS data, so these data are being left as they are. The documentation is being updated to note this issue in QMHMAR1, as well as a few other allocation flags from the fertility supplement that have this same problem in the 1986 - 1988 samples. Thank you for bringing this issue to our attention.