Skip to main content

Table 4 Variance estimates and variance partition coefficients [in parenthesis] for multilevel linear (standard error) and logit regressions (95% CI) at occupational group-, district-, region- and state-levels from null and adjusted models (National Sample Survey, India 2011–12)

From: Why market orientation matters for agriculture and fishery workers? Unravelling the association between households’ occupational background and caloric deprivation in India

Level

Rural India

Urban India

Null model

Fully adjusted modela

Null model

Fully adjusted modela

 

Calorie Consumption per Capita per Household

State

28,904 [7.4%]

36,781 [10.8%]

21,361 [4.4%]

20,100 [5.3%]

 

(10340)

(8762)

(8448)

(5963)

Region

9777 [2.5%]

8056 [2.4%]

8173 [1.7%]

7577 [2.0%]

 

(2546)

(2060)

(3105)

(2772)

District

20,352 [5.2%]

20,560 [6.0%]

17,541 [3.6%]

18,703 [4.9%]

 

(3665)

(3630)

(2800)

(3122)

NCO Occupational group

39,954 [10.2%]

26,386 [7.8%]

88,193 [18.1%]

56,244 [14.8%]

 

(3467)

(2971)

(23976)

(19608)

Household

293,923 [74.8%]

248,559 [73.0%]

352,001 [72.2%]

276,570 [72.9%]

 

(31788)

(28095)

(38399)

(37475)

 

Households with insufficient caloric intake

State

0.216 [5.0%]

0.409 [9.2%]

0.064 [1.5%]

0.149 [3.4%]

 

(0.048; 0.384)

(0.202; 0.615)

(−0.015; 0.142)

(0.024; 0.274)

Region

0.145 [3.4%]

0.143 [3.2%]

0.216 [4.9%]

0.214 [4.9%]

 

(0.056; 0.233)

(0.04; 0.245)

(0.081; 0.352)

(0.063; 0.364)

District

0.253 [5.9%]

0.281 [6.3%]

0.329 [7.5%]

0.376 [8.5%]

 

(0.189; 0.316)

(0.216; 0.346)

(0.249; 0.409)

(0.269; 0.483)

NCO Occupational group

0.380 [8.9%]

0.315 [7.1%]

0.483 [11.0%]

0.381 [8.6%]

 

(0.291; 0.469)

(0.233; 0.398)

(0.294; 0.672)

(0.185; 0.577)

  1. Note: To compute the total variance for the multilevel logistic regression model, we follow the latent variable methods approach and assume the between-household variation to have a the variance of a standard logistic distribution of 3.29 (Browne et al. 2005, Goldstein et al. 2002)
  2. aThe models are adjusted for age and sex of household head, household size, education of household head, religion, social group, wealth quintile, place of residence, sampling weights and standard errors clustered at the NCO occupational group, district, region and state level