Table 3. Association between eating away from home (EAFH) and BMI (I).
| Dependent variable: BMI | Total | Men | Women | |||
|---|---|---|---|---|---|---|
| Coefficient | S.E. | Coefficient | S.E. | Coefficient | S.E. | |
| EAFH | 0.040 | (0.015)** | 0.070 | (0.021)** | 0.003 | (0.020) |
| Heavy activity | -0.243 | (0.093)** | -0.556 | (0.138)** | 0.021 | (0.127) |
| Middle activity | -0.021 | (0.095) | -0.191 | (0.137) | 0.152 | (0.126) |
| Married | 0.794 | (0.139)** | 0.896 | (0.153)** | 0.672 | (0.227)** |
| Male | 0.402 | (0.104)** | ||||
| Old adult | 0.696 | (0.072)** | 0.255 | (0.110)* | 1.052 | (0.098)** |
| Secondary education | 0.069 | (0.082) | 0.616 | (0.121)** | -0.278 | (0.117)* |
| Tertiary education | -0.466 | (0.153)** | 0.385 | (0.219) | -1.239 | (0.209)** |
| ln(Income) | 0.084 | (0.039)* | 0.141 | (0.056)* | 0.024 | (0.052) |
| Smoking | -0.445 | (0.108)** | -0.395 | (0.115)** | -0.356 | (0.342) |
| Drinking | 0.161 | (0.080)* | 0.197 | (0.095)* | 0.115 | (0.148) |
| Rural | -0.059 | (0.093) | 0.030 | (0.127) | -0.221 | (0.129) |
| South | -1.325 | (0.085)** | -1.328 | (0.118)** | -1.334 | (0.113)** |
| Time trend | 0.238 | (0.030)** | 0.243 | (0.036)** | 0.239 | (0.042)** |
| Constant | 21.730 | (0.417)** | 21.274 | (0.622)** | 22.518 | (0.554)** |
| Observations | 10633 | 5084 | 5549 | |||
| BP-LM test | 1827.62** | 620.33** | 1154.30** | |||
Notes: BMI is the ratio of weight divided by square of height (kg/m2). S.E. refers to the standard error, ** p<0.01; * p<0.05. Independent variables included: EAFH (frequency of EAFH during the survey recall period), light activity (very light and light physical activity, such as office worker, sales person, teacher), middle activity (moderate physical activity such as student or driver), heavy activity (heavy and very heavy physical activity, such as farmer, dancer, loader, miner).educational level (primary education-reference group, secondary education or tertiary education), ln(income) (logarithm of income deflated to 2004), time trend (1 = 2004, 2 = 2006, 3 = 2009, 4 = 2011), smoking (1 = currently smoking), drinking (1 = drank last year), old adult (age>45), wife employed, married, male, rural, and south are dichotomous variables where 1 refers to the control group and 0 to the reference group. Rejection of BP-LM test indicates that random effect multivariable regression model is preferred.