Analysis Of Complex Survey Data Research Articles

Methods for analysis of complex survey data: an application using the Tanzanian 2015 Demographic and Health Survey and Service Provision Assessment.

BackgroundLow-income and middle-income countries (LMICs) seek to better utilize household and health facility survey data for monitoring and evaluation, as well as for health program planning. However, analysis of this complex survey data are complicated. In Tanzania, the National Evaluation Platform project sought to analyze Demographic and Health Survey (DHS) data and Service Provision Assessment (SPA) data as part of an evaluation of the national One Plan for Maternal and Child Health. To support this evaluation, we used this survey data to answer two key methodological questions: 1) what are the benefits and costs of using sampling weights in rate estimation; and 2) what is the best method for calculating standard errors in these two surveys?MethodsWe conducted a simulation study for each methodologic question. The first simulation study assessed the benefits and costs of using sampling weights in rate estimation. This simulation used weighted and unweighted estimates and examined bias, variance, and the mean squared error (MSE). The second simulation study assessed the best method for calculating standard errors comparing cluster bootstrapped variance estimation, design based asymptotic variance with one level (svy1), and design based asymptotic variance with three levels (svy3). We compared coverage probability and confidence interval length.ResultsOur results showed that although weighted estimates were less biased, unweighted estimates were less variable. The weighted estimates had a lower MSE, indicating that the effect of the bias trade-off was greater than the effect of the variance trade-off for most indicators assessed. The best performer for variance estimation was the cluster bootstrap method, followed by the svy3 method. The svy1 method was the worst performer for most indicators assessed.ConclusionsAs complex survey data become more widely used for policymaking in LMICs, there is a need for guidance on the best methods for analyzing this data. The standard of practice has been a design-based analysis using survey weights and the single-level svy method for calculating standard errors. This study puts forth an alternative approach to analysis. In addition, this study offers practical guidance on determining the best method for analysis of complex survey data.

How aggression-related mindsets explain SES-differences in bullying behavior

ABSTRACTPrevious research suggests that lower socioeconomic status (SES) adolescents bully more than their higher-SES peers. This paper tests whether aggression-related mindsets, defined as mindsets that theoretically influence aggressive behavior, explain the relationship between SES and bullying engagement among adolescents. Using a large and diverse dataset of survey responses from secondary students in the U.S. (N = 146,044 students, 30% White, 70% students of color from 5th-12th grade), this study applies structural regression modeling with complex survey data analysis. Results suggest that differences in aggression-related mindsets, including feelings of academic efficacy, feelings of purpose, global self-esteem, academic-status insecurity, school-related anger, and school-related happiness account for almost half of the relationship between SES and bullying. Students’ school-related anger is the strongest direct predictor of bullying (0.88 standard deviation), which suggests that strategies to reduce adolescent bullying are more effective if they teach anger-reducing skills or eliminate the root causes of students’ school-related anger.

Anaemia among adolescent girls in three districts in Ethiopia

BackgroundAdolescence is characterized by rapid growth and development with a significantly increased need for macro and micronutrients. However, there is little empirical evidence on the burden of anaemia among adolescent girls in developing countries such as Ethiopia. This study aims to address this gap by evaluating the magnitude of anaemia with an aim to guide design of intervention modalities to address anaemia among adolescent girls.MethodsThe study employed a community based cross sectional design. The study was conducted on weekends to capture both in school and out of school adolescent girls. Data was collected from a total 1323 adolescent girls. From each district, we randomly selected villages and ensured that the sampled households had a range geographical spread (lowlands, highlands) within the larger category of rural and urban. We performed anaemia testing using HemoCue B-Haemoglobin analyser. We applied a complex survey data analysis method to estimate the level of anaemia. The hemoglobin level was adjusted for altitude and smoking status. We ran a logistic regression model to evaluate predictors of anaemia.ResultsThe overall anaemia prevalence ranged from 24 to 38%, with an average rate of 29%. Less than half of the girls heard the term anaemia, and about one third knew the relationship between anaemia and the intake of iron rich foods. The risk of anaemia is higher among adolescent girls in their early adolescence period (10–14 years) (Adjusted Odds Ratio (AOR); 1.98; 95% CI; 1.03, 3.82] and among adolescent girls who lived in moderately food insecure households (AOR 1.48; 95% CI; 1.05–2.09). However, knowing the term “anaemia” was found to be protective against the risk of anaemia.ConclusionsThe risk of anaemia was particularly high among adolescent girls in their early age and among those living in food insecure households. The prevalence of anaemia among adolescent girls is a moderate public health problem. According to the WHO set criteria, the districts could be candidates for intermittent iron and Folic acid supplementation program.

Quantile Regression Analysis of Survey Data Under Informative Sampling.

For complex survey data, the parameters in a quantile regression can be estimated by minimizing an objective function with units weighted by the original design weights. However, when the complex survey sampling design is informative (i.e., when the design weights are correlated with the study variable even after conditioning on other covariates), the efficiency of this design-weighted estimator may be improved. In this article, we propose several weight-smoothing estimators for quantile regression analysis of complex survey data collected with an informative sampling design. Our new estimators incorporate nonparametric methods for modeling the weight functions and pseudo-population bootstrap methods for variance estimation. A simulation study compares, our proposed methods with the original design-based method in terms of bias, standard error, mean squared error, and confidence coverage. Our proposed estimators have smaller bias and mean squared error than does the design-based estimator. We further illustrate and compare estimators for the 1988 US National Maternal and Infant Health Survey.

Methods for analysis of complex survey data: An application using the Tanzanian 2015 Demographic and Health Survey and Service Provision Assessment

Methods for analysis of complex survey data: An application using the Tanzanian 2015 Demographic and Health Survey and Service Provision Assessment

HIV prevalence among men who have sex with men in Brazil: results of the 2nd national survey using respondent-driven sampling.

This paper reports human immuno-deficiency virus (HIV) prevalence in the 2nd National Biological and Behavioral Surveillance Survey (BBSS) among men who have sex with men (MSM) in 12 cities in Brazil using respondent-driven sampling (RDS).Following formative research, RDS was applied in 12 cities in the 5 macroregions of Brazil between June and December 2016 to recruit MSM for BBSS. The target sample size was 350 per city. Five to 6 seeds were initially selected to initiate recruitment and coupons and interviews were managed online. On-site rapid testing was used for HIV screening, and confirmed by a 2nd test. Participants were weighted using Gile estimator. Data from all 12 cities were merged and analyzed with Stata 14.0 complex survey data analysis tools in which each city was treated as its own strata. Missing data for those who did not test were imputed HIV+ if they reported testing positive before and were taking antiretroviral therapy.A total of 4176 men were recruited in the 12 cities. The average time to completion was 10.2 weeks. The longest chain length varied from 8 to 21 waves. The sample size was achieved in all but 2 cities.A total of 3958 of the 4176 respondents agreed to test for HIV (90.2%). For results without imputation, 17.5% (95%CI: 14.7–20.7) of our sample was HIV positive. With imputation, 18.4% (95%CI: 15.4–21.7) were seropositive.HIV prevalence increased beyond expectations from the results of the 2009 survey (12.1%; 95%CI: 10.0–14.5) to 18.4%; CI95%: 15.4 to 21.7 in 2016. This increase accompanies Brazil's focus on the treatment to prevention strategy, and a decrease in support for community-based organizations and community prevention programs.

Household income, income inequality, and health-related quality of life measured by the EQ-5D in Shaanxi, China: a cross-sectional study

BackgroundIn advanced economies, economic factors have been found to be associated with many health outcomes, including health-related quality of life (HRQL), and people’s health is affected more by income inequality than by absolute income. However, few studies have examined the association of income inequality and absolute income with HRQL in transitional economies using individual data. This paper focuses on the effects of county or district income inequality and absolute income on the HRQL measured by EQ-5D and the differences between rural and urban regions in Shaanxi province, China.MethodsData were collected from the 2008 National Health Service Survey conducted in Shaanxi, China. The EQ-5D index based on Japanese weights was employed as a health indicator. The income inequality was calculated on the basis of self-reported income. The special requirements for complex survey data analysis were considered in the bivariate analysis and linear regression models.ResultsThe mean of the EQ-5D index was 94.6. The EQ-5D index of people with low income was lower than that in the high-income group (for people in the rural region: 93.2 v 96.1, P < 0.01; for people in the urban region: 95.5 v 96.8, P < 0.01). Compared with people with moderate inequality, the EQ-5D index of those with high inequality was relatively lower (for people living in the rural region: 91.1 v 95.8, P < 0.01; for people living in the urban region: 95.6 v 97.3, P < 0.01). Adjusted by age, gender, education, marital status, employment, medical insurance, and chronic disease, all the coefficients of the low-income group and high income inequality were significantly negative. After stratifying by income group, all the effects of high income inequality remained negative in both income groups. However, the coefficients of the models in the high income group were not statistically significant.ConclusionIncome inequality has damaging effects on HRQL in Shaanxi, China, especially for people with low income. In addition, people living in rural regions were more vulnerable to economic factors.

Using iMCFA to Perform the CFA, Multilevel CFA, and Maximum Model for Analyzing Complex Survey Data.

To construct CFA, MCFA, and maximum MCFA with LISREL v.8 and below, we provide iMCFA (integrated Multilevel Confirmatory Analysis) to examine the potential multilevel factorial structure in the complex survey data. Modeling multilevel structure for complex survey data is complicated because building a multilevel model is not an infallible statistical strategy unless the hypothesized model is close to the real data structure. Methodologists have suggested using different modeling techniques to investigate potential multilevel structure of survey data. Using iMCFA, researchers can visually set the between- and within-level factorial structure to fit MCFA, CFA and/or MAX MCFA models for complex survey data. iMCFA can then yield between- and within-level variance-covariance matrices, calculate intraclass correlations, perform the analyses and generate the outputs for respective models. The summary of the analytical outputs from LISREL is gathered and tabulated for further model comparison and interpretation. iMCFA also provides LISREL syntax of different models for researchers' future use. An empirical and a simulated multilevel dataset with complex and simple structures in the within or between level was used to illustrate the usability and the effectiveness of the iMCFA procedure on analyzing complex survey data. The analytic results of iMCFA using Muthen's limited information estimator were compared with those of Mplus using Full Information Maximum Likelihood regarding the effectiveness of different estimation methods.

Complex Survey Data Analysis with SAS

Complex Survey Data Analysis with <i>SAS</i>

Introduction to the Design and Analysis of Complex Survey Data

We give a brief overview of common sampling designs used in a survey setting, and introduce the principal inferential paradigms under which data from complex surveys may be analyzed. In particular, we distinguish between design-based, model-based and model-assisted approaches. Simple examples highlight the key differences between the approaches. We discuss the interplay between inferential approaches and targets of inference and the important issue of variance estimation.

Health-related Quality of Life among Korean Older Women with Osteoarthritis: Comparison between Young-Old and Old-Old

Purpose : The purpose of this study was to investigate the health-related quality of life by age among older Korean women with osteoarthritis. Methods : This study design was secondary data analysis from the fifth KNHANES. The subjects were 790 Korean women with osteoarthritis over the aged 65. Data were analyzed with SPSS 21.0 using complex survey data analysis methods. Results : The health-related characteristics such as restriction of activity, sleeping time, BMI, waist circumference and walking were shown significant differences between youngold and old-old. Also there were significant differences in the EQ-5D index, mobility, self-care and daily living between young-old and old-old. Perceived health status, restriction of activity, and physical activity influenced on health-related quality of life both young-old and old-old. Conclusion : Based on the results, it is suggested that the health care providers should consider perceived health status, restriction of activity, and physical activity when they plan to design program for the improvement of health-related quality of life for both young-old and old-old women with osteoarthritis.

Frames2: A Package for Estimation in Dual Frame Surveys

Data from complex survey designs require special consideration with regard to estimation of finite population parameters and corresponding variance estimation procedures, as a consequence of significant departures from the simple random sampling assumption. In the past decade a number of statistical software packages have been developed to facilitate the analysis of complex survey data. All these statistical software packages are able to treat samples selected from one sampling frame containing all population units. Dual frame surveys are very useful when it is not possible to guarantee a complete coverage of the target population and may result in considerable cost savings over a single frame design with comparable precision. There are several estimators available in the statistical literature but no existing software covers dual frame estimation procedures. This gap is now filled by package Frames2. In this paper we highlight the main features of the package. The package includes the main estimators in dual frame surveys and also provides interval confidence estimation. In practice, the assumption that the sampling frame contains all population units is rarely met. Often, one finds that sampling from a frame which is known to cover approximately all units in the population is quite expensive while other frames (e.g. special lists of units) are available for cheaper sampling methods. However, the latter usually only cover an unknown or only approximately known fraction of the population. A common example of frame undercoverage is provided by telephone surveys. Estimation could be affected by serious bias due to the lack of a telephone in some households and the generalised use of mobile phones, which are sometimes replacing fixed (land) lines entirely. The potential for coverage error as a result of the exponential growth of the cell-phone only population has led to the development of dual-frame surveys. In these designs, a traditional sample from the landline frame is supplemented with an independent sample from a register of cell-phone numbers. The dual frame sampling approach assumes that two frames are available for sampling and that, overall, they cover the entire target population. The most common situation is the one represented in Figure 1 where the two frames, say frame A and frame B, show a certain degree of overlapping, so it is possible to distinguish three disjoint non-empty domains: domain a, containing units belonging to frame A but not to frame B; domain b, containing units belonging to frame B but not to frame A and domain ab, containing units belonging to both frames. As an example, consider a telephone survey where both landline and cell phone lists are available; let A be the landline frame and B the cell phone frame. Then, it is possible to distinguish three types of individuals: landline only units, cell-only units and units with both landline and cell phone, which will compose domain a, b and ab, respectively. Nevertheless, one can face some other situations depending on the relative positions of the frames. For example, Figure 2 shows the case in which frame B is totally included in frame A, that is, frame B is a subset of frame A. Here domain b is empty. We also may find scenarios where the two sampling frames exactly match, as depicted in Figure 3, where ab is the only non-empty domain. Finally, the scenario where domain ab is empty has no interest from a dual frame perspective, since it can be

한국 전·후기 노인의 삶의 질 관련요인과 성별 차이

= Abstract = Objectives: As a sequel to the former analysis of the quality of life (QoL) among young-old and old-old in Korea, this research was aimed to identify factors related to the quality of life and the gender difference after controlling for the related factors among Korean elderly. Methods: Selected elderly data of 1,339 subjects from the 5th Korea National Health and Nutrition Examination Survey conducted in 2010 was analyzed. In this survey, QoL was measured using Euro Quality of Life (EQ-5D) instrument. Data were analyzed using complex survey data analysis on IBM-SPSS 20.0. The related factors were identified using general linear models with backward elimination. The gender difference was tested also using general linear models. Results: The distributions of educational level, family income level, and presence of cohabitant were different between male and female elderly in both young-old and old-old age group. So were the health behaviors and perceived health, and experience of stress, depression, and suicidal thoughts. QoL and its subscales- mobility, self care, daily living, pain and discomfort, and anxiety and depression- were consistently better among male elderly regardless of age group. Among the variables considered, education, family income level, presence of cohabitant, perceived health, age group and BMI were found to be related to the QoL at p=.05, and presence of chronic diseases at p=.10. The difference in QoL between male and female elderly after controlling for the variables was statistically significant. Conclusion: Improving QoL is particularly important for the elderly. In order to improve QoL of the elderly, age- and gender- differences need to be considered when developing services and programs for the elderly.

Effects of Enamel Fluorosis and Dental Caries on Quality of Life

The objectives of this study were to determine the impact of enamel fluorosis and dental caries on oral health–related quality of life (OHRQoL) in North Carolina schoolchildren and their families. Students (n = 7,686) enrolled in 398 classrooms in grades K-12 were recruited for a onetime survey. Parents of students in grades K-3 and 4-12 completed the Early Childhood Oral Health Impact Scale (ECOHIS) and Family Impact Scale (FIS), respectively. Students in grades 4-12 completed the Child Perceptions Questionnaire (CPQ8-10 in grades 4-5; CPQ11-14 in grades 6-12). All students were examined for fluorosis (Dean’s index) and caries experience (d2-3fs or D2-3MFS indices). OHRQoL scores (sum response codes) were analyzed for their association with fluorosis categories and sum of d2-3fs and D2-3MFS according to ordinary least squares regression with SAS procedures for multiple imputation and analysis of complex survey data. Differences in OHRQoL scores were evaluated against statistical and minimal important difference (MID) thresholds. Of 5,484 examined students, 71.8% had no fluorosis; 24.4%, questionable to very mild fluorosis; and 3.7%, mild, moderate, or severe fluorosis. Caries categories were as follows: none (43.1%), low (28.6%), and moderate to high (28.2%). No associations between fluorosis and any OHRQoL scales met statistical or MID thresholds. The difference (5.8 points) in unadjusted mean ECOHIS scores for the no-caries and moderate-to-high caries groups exceeded the MID estimate (2.7 points) for that scale. The difference in mean FIS scores (1.5 points) for the no-caries and moderate-to-high groups exceeded the MID value (1.2 points). The sum of d2-3fs and D2-3MFS scores was positively associated with CPQ11-14 (B = 0.240, p < .001), ECOHIS (B = 0.252, p ≤ .001), and FIS (B = 0.096, p ≤ .01) scores in ordinary least squares regression models. A child’s caries experience negatively affects OHRQoL, while fluorosis has little impact.

A diagnostic tool for regression analysis of complex survey data

Although a nonparametric regression model allows us to obtain a graphical display of the relationship between the response variable and the independent variables, the exact form of the regression function is not evident. This weakness can be overcome by a parametric model since the relationship between the dependent and the independent variables is specified mathematically. The object of this paper is to introduce a new graphical approach, called the shift function plot, with which a hypothesis test is constructed to evaluate the goodness-of-fit of a parametric regression model. Under multi-stage stratified sampling schemes, we apply the shift function plots to survey data. Asymptotic properties of the survey estimator of the shift function are established. An empirical example from the 1990 Ontario Health Survey is used to illustrate the application of the shift function.

40-Year Trends in Meal and Snack Eating Behaviors of American Adults

BackgroundUnderstanding changes in profiles of eating behaviors over time may provide insights into contributors to upward trajectories of obesity in the US population. Yet little is known about whether or not characteristics of meal and snack eating behaviors reported by adult Americans have changed over time. ObjectiveTo examine time trends in the distribution of day’s intake into individual meal and snack behaviors and related attributes in the US adult population. DesignThe study was observational with cross-sectional data from national surveys fielded over 40 years. Participants/settingNationally representative dietary data from nine National Health and Nutrition Examination Surveys conducted from 1971-1974 to 2009-2010 (N=62,298 participants aged 20-74 years) were used to describe eating behaviors. Outcomes examinedThe respondent-labeled eating behaviors examined included main meals (breakfast, lunch, and dinner), and snacks (before breakfast, between breakfast and lunch, between lunch and dinner, after dinner, or other). For each eating behavior, percent of reporters, relative contribution to 24-hour energy intake, the clock time of report, and intermeal/snack intervals were examined. Statistical analysisMultivariable logistic and linear regression methods for analysis of complex survey data adjusted for characteristics of respondents in each survey. ResultsOver the 40-year span examined reports of each individual named main meal (or all three main meals) declined, but reports of only two out of three meals or the same meal more than once increased; the percentage of 24-hour energy from snacks reported between lunch and dinner or snacks that displaced meals increased; clock times of breakfast and lunch were later, and intervals between dinner and after-dinner snack were shorter. Changes in several snack reporting behaviors (eg, report of any snack or ≥2 snacks), were significant in women only. ConclusionsSeveral meal and snack eating behaviors of American adults changed over time, with a greater change in snack behaviors of women relative to men.

Alcohol consumption and risky drinking patterns in Malaysia: findings from NHMS 2011.

To identify the characteristics of current drinker and risky alcohol-drinking pattern by profiles in Malaysia. We analyzed data from the National Health and Morbidity Survey 2011. It was a cross-sectional population-based with two stages stratified random sampling design. A validated Alcohol Use Disorder Identification Test Malay questionnaire was used to assess the alcohol consumption and its alcohol related harms. Analysis of complex survey data using Stata Version 12 was done for descriptive analysis on alcohol use and risky drinking by socio-demography profiles. Logistic regression analysis was used to measure the association of risky drinking status with the socio-demography characteristics. The prevalence of current alcohol use was 11.6% [95% confidence interval (CI): 10.5, 12.7], among them 23.6% (95% CI: 21.0, 26.4) practiced risky drinking. The onset for alcohol drinking was 21 years old (standard deviation 7.44) and majority preferred Beer. Males significantly consumed more alcohol and practiced risky drinking. Current alcohol use was more prevalent among urbanites, Chinese, those with high household income, and high education. Conversely, risky drinking was more prevalent among rural drinkers, Bumiputera Sabah and Sarawak, low education and low household income. The estimated odds of risky drinking increased by a factor of 3.5 among Males while a factor of 2.7 among Bumiputera Sabah and Sarawak. Education status and household income was not a significant predictor to risky drinking. There was an inverse drinking pattern between current drinker and risky drinking by the socio-demography profiles. Initiating early screening and focused intervention might avert further alcohol related harms and dependence among the risky drinkers.

Accounting for Selection Bias in Association Studies with Complex Survey Data

Obtaining representative information from hidden and hard-to-reach populations is fundamental to describe the epidemiology of many sexually transmitted diseases, including HIV. Unfortunately, simple random sampling is impractical in these settings, as no registry of names exists from which to sample the population at random. However, complex sampling designs can be used, as members of these populations tend to congregate at known locations, which can be enumerated and sampled at random. For example, female sex workers may be found at brothels and street corners, whereas injection drug users often come together at shooting galleries. Despite the logistical appeal, complex sampling schemes lead to unequal probabilities of selection, and failure to account for this differential selection can result in biased estimates of population averages and relative risks. However, standard techniques to account for selection can lead to substantial losses in efficiency. Consequently, researchers implement a variety of strategies in an effort to balance validity and efficiency. Some researchers fully or partially account for the survey design, whereas others do nothing and treat the sample as a realization of the population of interest. We use directed acyclic graphs to show how certain survey sampling designs, combined with subject-matter considerations unique to individual exposure-outcome associations, can induce selection bias. Finally, we present a novel yet simple maximum likelihood approach for analyzing complex survey data; this approach optimizes statistical efficiency at no cost to validity. We use simulated data to illustrate this method and compare it with other analytic techniques.

Evaluation of Generalized Variance Functions in the Analysis of Complex Survey Data

Abstract Two sets of diagnostics are presented to evaluate the properties of generalized variance functions (GVFs) for a given sample survey. The first set uses test statistics for the coefficients of multiple regression forms of GVF models. The second set uses smoothed estimators of the mean squared error (MSE) of GVF-based variance estimators. The smooth version of the MSE estimator can provide a useful measure of the performance of a GVF estimator, relative to the variance of a standard design-based variance estimator. Some of the proposed methods are applied to sample data from the Current Employment Statistics survey.

A Complex Survey Data Analysis of Tb Mortality in South Africa

Many countries in the world record annual summary statistics such as economic indicators like Gross Domestic Product (GDP) and vital statistics for example the number of births and deaths. In this paper we focus on mortality data from various causes including Tuberculosis (TB). TB is an infectious disease caused by bacteria called Mycobacterium tuberculosis. It is the main cause of death in the world among all infectious diseases (Herchline and Amorosa, 2010). An additional complexity is that HIV/AIDS acts as a catalyst to the occurrence of TB. People infected with mycobacterium tuberculosis alone have an approximately 10% life time risk of developing active TB, compared to 60% or more in persons co-infected with HIV and mycobacterium tuberculosis (Vaidynathan and Singh, 2003). In 2006, South Africa was ranked seventh highest by the World Health Organization (WHO, 2009) among the 22 TB high burden countries in the world and fourth highest in Africa. The research work in this presentation uses the 2007 Statistics South Africa (STATSSA) data on TB as the primary cause of death to build statistical models that can be used to investigate factors associated with death due to TB. Logistic regression and generalized linear models (GLM) will be used to assess the effect of some risk factors or predictors to the probability of deaths associated with TB. This study will be guided by a theoretical approach to understanding factors associated with TB death. Of the 615312 deceased, (89%) died from natural death, (2%) were stillborn and (9%) from non-natural death possibly accidents, murder, suicide. Among those who died from natural death and disease, (12%) died of TB.Keywords: TB mortality, prevalence, HIV incidence, TB/HIV co-infection, survey data analysis, logistic regression model, and STATSSA South Africa.