Pregnancy Risk Assessment Monitoring System (PRAMS) - Sampling and Analysis Methodology
PRAMS Sampling and Analysis Methodology
New York State utilizes a stratified random sampling approach to select PRAMS samples. Each sample consists of two strata: mothers of low birth weight and normal birth weight infants. This is done to obtain a reasonable sample size for mothers of low birth weight infants and generate stable estimates. Any records with missing or invalid birth weight information were put into the low birth weight group in the sampling frame.
Sampling ratio applied to each birth weight stratum
|Low Birth Weight (<2,500g)||3/22|
|Normal Birth Weight (2,500g+)||1/121|
Although the probability of being selected into each stratum differs, the number of women from each stratum in the resulting sample is approximately equal.
To produce population-based estimates, each mother was assigned a sampling weight. These sampling weights account for varying probabilities of selection of mothers because of stratification by birth weight. In addition, the weights account for non-response and non-coverage.
Since PRAMS utilizes the stratified random sampling method, estimates generated from PRAMS data are subject to sampling error. The standard error of an estimate measures the sampling variability among all possible samples that could have been drawn from the sampling frame. The standard error for each estimate is expressed via confidence interval (CI).
The weighted percent of the response for selected questions and 95% CIs are provided. CIs provided were obtained from SAS using the “surveyfreq” procedure via the specification of first order Taylor Series linearization approximation. Unknown or missing values are not included in the sample size.