Using Stata for Survey Data Analysis

regress depvar varlist [if exp] [in range] [options] 
where 
depvar
is the dependent variable 
varlist
is the list of independent variables 
The regress command has many options for specifying the type and format of the output. Type “help 
regress” for more information. Some examples of the command:
regress y x1 x2 x3 x4 x5
regress y with x’s as independent variable 
regress y x1 x2 x3 x4 x5 if region==1 
same regression but only in one region 
by region: regress y x1 x2 region*
region* means all variables starting with region 
 
Example 23 presents the results of a regression analysis of the determinants of food expenditure using 
the merged data from the household.dta file and the food expenditure.dta file. The dependent variable 
is the log of food expenditure per capita. The explanatory variables include the log of per capita 
expenditure, stratum (whether or not the household is in an urban area), the household size, the sex of 
the head of household, and two dummies to represent the west and central regions. For each 
explanatory variable, the output includes six columns: 
Coefficient: 
The effect of a one unit change in the explanatory variable on the 
dependent variable.
Standard error 
A measure of the accuracy of the estimate of the coefficient 
t statistics 
The coefficient divided by the standard error. Generally, any t 
statistics greater than 2 is considered statistically significant. 
P > |t| 
The probability that the coefficient is actually zero, given the data. 
Any value of P less than .05 is considered statistically significant
95% confidence interval 
Two columns showing the lower and upper limit of 95% confidence 
interval. This means the data indicate that we can be 95% sure that 
the true value lies between the upper and lower limit.

---