APPENDIX 1: METHODOLOGY

Except as otherwise noted, the statistics presented in this report were developed from data obtained from, "The Survey of Inmates in State and Federal Correctional Facilities, 1997," conducted by the U.S. Census Bureau for the Bureau of Justice Statistics (BJS) of the U.S. Department of Justice. The survey consisted of personal interviews of a nationally representative sample of state and federal prison inmates, as well as a statistically representative sample of state prisoners in California, New York, and Texas. The BJS used the 1997 survey as the basis for its report, "Incarcerated Parents and Their Children," published in August 2000. Working with the data obtained from New York prisoners in the 1997 survey, we weighted each respondent by the respondent's probability of being selected for the sample. The weighting variables were included in the prisoner survey data. Our analysis also used the estimates of generalized variance parameters for male and female inmates provided in the documentation accompanying the 1997 survey.

How accurate are our estimates? The population of New York prisoners in the 1997 survey was selected to be representative of the state's male and female inmate populations, inmates in various facility types (small vs. large), as well as other characteristics. Representative samples of inmate mothers and fathers were not purposefully selected. This introduces the possibility of serious sampling error, especially among incarcerated mothers-as only 104 out of an estimated total number of 2,369 were sampled for the 1997 survey.

From the estimated generalized variance parameters we derived margins of errors for various percentages in the report. The size of the margins of error will depend on a variety of factors including, most significantly, the estimated population size on which the sample is based. The table below indicates a few key margins of error that will provide a general gauge of the quality of the estimates. As is immediately evident from Table A, the margins of error for estimates derived solely from female samples are relatively large because of the small number of women in the survey sample

Table A: Sample Margins of Error for Various Percentage Estimates

Characteristic	Fathers' estimates (Base)	Mothers' estimates (Base)	Margin of error (Male)	Margin of error (Female)
Have minor children	57.5% (65839)	64.1% (3693)	+/- 4.32	+/- 8.94
Child living with grandparents	14.1% (37859)	51.2% (2368)	+/- 4.01	+/- 11.64
Child living other parent	88.8% (37859)	20.4% (2368)	+/- 3.64	+/- 9.38
Inmates with two children	16.4% (65015)	20.8% (3658)	+/- 3.26	+/- 7.57
Living 500 miles or more from home	17.3% (37859)	20.1% (2368)	+/- 4.36	+/- 9.33
Never visited	49.5% (37859)	52.5% (2368)	+/- 5.77	+/- 11.63