Data The present study uses Bangladesh Demographic and Health Survey (BDHS) 2011 data. The National Institute of Population Research and Training (NIPORT) of the Ministry of Health and Family Welfare is the main authority of BDHS data. The BDHS is a nationally representative survey designed to provide information on basic national indicators of social improvement including nutritional status of mother and children, childhood mortality, fertility, maternal and child health, awareness of AIDS and domestic violence. To ensure greater precision the two-stage stratified random sampling techniques were used during the survey where each of the seven administrative divisions of Bangladesh (Barisal, Chittagong, Dhaka, Khulna, Rajshahi, Rangpur and Sylhet) is divided into a number of clusters. In the first stage, the mauza in rural strata was considered as primary sampling units (PSUs) and in the second stage, households from each PSU were selected using a systematic sampling scheme. A primary sampling unit (PSU) is usually a natural cluster of households. The BDHS data are hierarchical due to its formation where individuals are being nested into PSUs, and PSUs into divisions. In the 2011 BDHS, a total of 10,996 ever-married women of age 15 – 49 from the selected households were interviewed to collect data on fertility, family planning, child and maternal health. In our study, we use 5273 under-five children generated from the women sample in rural Bangladesh to assess the nutritional status as well as its associated factors. Measures of child nutritional status The Z-score method is the most popular and common method of measuring child nutritional status (WHO, 1986; Cogill, 2003). The Z-scores are calculated on the basis of different anthropometric measures such as age, height and weight along with WHO Child Growth Standards (WHO, 2006). Determination of child nutritional status The main goal of this study is to examine the factors related to the response variables (e.g., stunting, wasting and underweight). In the survey data set the dependent variable (child nutritional status) is categorized into three groups as follows: (i) severely malnourished if Z-score is less than -3.0, (ii) moderately malnourished if Z-score is between -3.0 to -2.01, and (iii) healthy if Z-score is greater than or equal to -2.0. However, for the convenience in the regression modeling the dependent variables are re-coded into two groups (binary) because some of the higher categories have fewer observations. The categories are as follows: (i) 1 = child is malnourished if Z-score is less than -2.00 and (ii) 0 = child is not malnourished if Z-score is greater than or equal to -2.00 (reference category). The primary preference of explanatory variables for this study was based on previous other studies on the factors influencing children’s nutritional status. The independent variables used in the study are division, child age in month, sex of child, twin child, birth order, birth interval, religion, mother’s education, father’s education, mother’s age at first child birth, age of household head at first child birth, toilet facility, sources of drinking water, wealth index, child received vitamin A, place of delivery, total children ever born, household size, child size at birth, child suffered from fever, cough and diarrhoea, had television, had radio, and refrigerator and access to electricity. 2.3.1 Two-level random intercept binary logistic regression model. It is very likely that the cluster or community (PSU) effect on the response variable will be present when there is a hierarchical data structure in the survey, for example, BDHS 2011. The traditional logistic regression ignoring such cluster effect is inappropriate as the standard errors of regression coefficients are underestimated leading to the significance of a regression coefficient that could be ascribed to likelihood. This may instigate wrong policy formulation (for example, see Khatun et al., 2012). In this context, to overcome this problem a multilevel logistic regression model containing both fixed effects and random effects that attempts to capture the unobserved heterogeneity between clusters is commonly used (Pinheiro and Bates, 2000; Goldstein, 2003; Demidenko, 2004). The use of appropriate multilevel model provides efficient estimates and the policies devised on the basis of these are reliable. In addition, the significance and extent of the community effect help to find if there is any community that is performing poorly