• Missing value imputation methods for parameter estimates and psychometric properties of Likert measures

      Zhou, Qiuping (Pearl); Soeken, Karen (2001)
      Problem. Missing items are a common problem in Likert-type measures consisting of multiple questions. Despite frequent use of imputation methods for missing values, data about the performance of different methods on outcome measures are lacking. Purpose. To assess the performance of four imputation methods on item and scale level statistics and psychometric properties under different data conditions and the mechanism of missing completely at random (MCAR). Methods. This is a secondary data analysis using a dataset consisting of responses to the SF-36. The imputation methods under study include item mean substitution (IMS), person mean substitution (PMS), expectation-maximization algorithm (EM), and stochastic regression imputation (SRI). Missing data conditions include percentage of subjects with missing values (10%, 25% and 40%), percentage of missing items (10%, 20-33%, and 33-50%), sample size (200 and 500), and length of scales (2-, 4- and 10-items). After creation of each missing data condition under MCAR, imputation methods were applied and statistics from the imputed datasets were computed. Accuracy and bias of estimates for item level statistics (mean, SD and correlation), scale level statistics (mean, SD and correlation), and psychometric properties (coefficient alpha, goodness-of-fit statistics of confirmatory factor analysis, factor loadings and factor correlations) were compared and contrasted. One-way ANOVA and GLM were used to analyze data. The significant level was set at alpha ≤ .05. Results. There were differences in performance of imputation methods regardless of data conditions. IMS was consistently the worst in terms of accuracy and bias for all but two parameters. PMS was the second worst in parameter estimates. In contrast, EM and SRI produced more accurate estimates for most parameters considered. EM ranked the best for estimation of item mean and intercorrelations among items and scales while SRI ranked the best for item SD, alpha and goodness-of-fit statistics. The two methods were equivalent for other parameters. Further, their performance was less influenced by missing data conditions. In terms of reducing bias, SRI was better for most of the parameters than EM. Conclusion. Model-based approaches, SRI and EM, are preferred over IMS and PMS for imputing missing items in Likert type measures.
    • The relationship between recurrent placental pathology and recurrent poor pregnancy outcome

      Krulewitch, Cara Joy; Kohler, Helen (1992)
      Poor pregnancy outcome (PPO) is a top agenda item in health and political arenas. Most prominent PPOs are low birth weight (LBW) and preterm delivery (PTD). Effective interventions to maximize pregnancy outcome will diminish their negative impact. The placenta is a diary of the infant's prenatal experiences. Changes in the placenta provide clues to the uterine environment during the pregnancy. Each placental pathology represents a different insult, including severity, duration and type. The Collaborative Perinatal Project of the National Institute of Neurological and Communicative Disorders and Stroke was a longitudinal study of obstetric outcomes in women. The study was conducted from 1959 to 1966 and contains information for 53,518 pregnancies from 10,699 women, including examination of 31,494 placentas (59%). A nested case control study was conducted on this dataset. A sample of 7653 women of black and white races who had two pregnancies during the study period were examined to determine if a relationship existed between the recurrence of PPO and a recurrence of placental pathology. Analysis used several statistical techniques including linear logistic regression and an experimental causal modelling program, TETRAD II. TETRAD II analysis indicated a relationship between marital status and educational incongruity and PPO along with a relationship between recurrent chronic hypoxia and LBW and recurrent infection and PTD. This was the first time TETRAD II was used on real data. Linear logistic regression confirmed these relationships. The risk of repeating LBW or PTD was 1.6 times greater if the woman was unmarried and was 10% greater for each year of greater educational incongruity between the mother and father of the baby. Cigarette smoking in both pregnancies represented 2.1 (1.5-2.9) times greater risk of repeating LBW. There was a 4.2 (1.4-12) times greater risk of repeating PTD if placental chorionic villitis repeated, remaining significant after adjustment for cigarette smoking and social class. There was a 6.5 (1.4-31) times greater risk of repeating LBW if decidual necrosis repeated, remaining significant after adjustment for social class, but becoming marginal when cigarette smoking was included. The main implication of these findings involves the area of preconceptual care for women. This includes preconceptual treatment of infection and nursing support to women at high risk for repeating a PPO.
    • The relationship of prenatal care utilization and tobacco and alcohol use to perinatal and neonatal outcomes: A secondary analysis

      Stuhlmuller, Patricia Lynn; Allen, Karen, Ph.D., R.N. (1998)
      1991 birth and death certificate information for Maryland state residents was used to examine the relationship of entry into prenatal care and number of follow-up visits to the birth outcomes of gestational age, birthweight and neonatal and postneonatal mortality. The impact of maternal race, age, education, marital status, pregnancy history and behaviors were also investigated. The live birth file contained 42,019 cases; 64% were white and 36% were black. Women in the sample ranged in age from 12 to 48 years with an average age of 26.5 years. The linked infant birth/death file contained 227 cases; 60% of the infants died in the neonatal period and 40% died postneonatally. Black infants comprised 66% of the infant deaths and white infants comprised 33%. Correlational analyses, logistic regression analyses and hierarchical multiple regression analyses were used to address the research questions. Analyses were computed for the total sample and sub samples of white and black women. Alcohol use was significantly correlated with age of infant death for the total sample and for whites but not for blacks. Tobacco use was not significantly correlated with infant death. Women in the total sample and black women who used tobacco and alcohol tended to have low birthweight and low gestational age infants. For whites, tobacco use was related to low gestational age and low birthweight. Logistic regression analyses showed that women who had more prenatal visits were more likely to have infants who died postneonatally. Time of entry into prenatal care had no effect on timing of infant death. Hierarchical multiple regression models were better able to explain variance in birthweight verses gestational age. Having prenatal care and more prenatal follow-up visits were found to be significant predictors of high gestational age and birthweight for the total sample, blacks and whites. The impact of the maternal demographics, pregnancy history and maternal behavior variables were found to vary by race. Implications for health care providers rendering prenatal care are discussed.