By Ylva B Almquist
Now that we have discussed what confidence intervals are (and what they are not), we thought it would be good time to show how to calculate them for descriptive statistics.
For this purpose, we can use the commands ci, centile, and proportion.
5.5.1 Confidence intervals for means
Can be used for continuous variables with a normal distribution.
Function
| Basic command |
ci means varlist |
| Useful options |
ci means varlist, level(#) |
| Explanations |
varlistInsert the name(s) of the variable(s) that you want to use level(#)Specify the confidence level. Default is 95. |
| More information | |
| Command | help centile |
Practical example
| Dataset | ||
| StataData1.dta | ||
| Variable name | Variable label | Values and labels |
| gpa | Grade point average (Age 15, Year 1985) | – |
ci means gpa
In this example, we can see that the mean value for gpa is 3.18. The 95% confidence interval is 3.16-3.19.
5.5.2 Confidence intervals for median
Can be used for continuous variables (with a normal or skewed distribution).
Function
| Basic command |
centile varlist |
| Useful options |
centile varlist, level(#) |
| Explanations |
varlistInsert the name(s) of the variable(s) that you want to use level(#)Specify the confidence level. Default is 95. |
| More information | |
| Command | help centile |
Practical example
| Dataset | ||
| StataData1.dta | ||
| Variable name | Variable label | Values and labels |
| cognitive | Cognitive test score (Age 15, 1985) | – |
centile cognitive
In this example, we can see that the median cognitive test score is 312, and the 95% confidence interval is 312-316.
5.5.3 Confidence intervals for variances and standard deviations
Can be used for variables that are continuous.
Function
| Basic command |
ci variances varlist |
| Useful options |
ci variances varlist, level(#)ci variances varlist, sd level(#) |
| Explanations |
varlistInsert the name(s) of the variable(s) that you want to use sdOption to display confidence interval for standard deviation. level(#)Specify the confidence level. Default is 95. |
| More information | |
| Command | help ci |
Practical example
| Dataset | ||
| StataData1.dta | ||
| Variable name | Variable label | Values and labels |
| cognitive | Cognitive test score (Age 15, 1985) | – |
ci variances cognitive
Here, the variance (5210) and its confidence interval (5061-5367) is shown.
ci variances cognitive, sd
This shows the standard deviation (72.18) and its 95% confidence interval (71.14-73.26).
5.5.4 Confidence intervals for counts
Can be used for continuous variables that are counts.
Function
| Basic command |
ci means varlist, poisson |
| Useful options |
ci means varlist, poisson level(#) |
| Explanations |
varlistInsert the name(s) of the variable(s) that you want to use level(#)Specify the confidence level. Default is 95. |
| More information | |
| Command | help ci |
Practical example
| Dataset | ||
| StataData1.dta | ||
| Variable name | Variable label | Values and labels |
| unemp_42 | Days in unemployment (Age 42, Year 2012) | – |
ci means unemp_42, poisson
In this example, the mean is 17.53 days in unemployment. The 95% confidence interval is 17.45-17.61 days.
5.5.5 Confidence intervals for proportions
Can be used for categorical variables.
Function
| Basic command |
proportions varlist |
| Useful options |
proportions varlist, level(#) |
| Explanations |
varlistInsert the name(s) of the variable(s) that you want to use level(#)Specify the confidence level. Default is 95. |
| More information | |
| Command | help proportions |
Practical example
| Dataset | ||
| StataData1.dta | ||
| Variable name | Variable label | Values and labels |
| educ | Educational level (Age 40, Year 2010) | – |
proportion educ
Here, we get the proportions (which can be translated into percentages) and its confidence interval for the three categories of educ.
In this example, 19.2% have compulsory education (95% CI: 18.4-20.0), 44.2% have upper secondary education (95% CI: 43.2-45.3%), and 36.6% have university education (95% CI: 35.6-37.6%).