How to create categorical variables from continuous variables

Sometimes, we wish to create categorical variables from continuous variables. For example, we may wish to create a variable for high versus low systolic blood pressure from a continuous variable for systolic blood pressure.

Let's begin by opening and describing an example dataset from the Stata website.

. use https://www.stata.com/users/youtube/rawdata.dta, clear
(Fictitious data based on the National Health and Nutrition Examination Survey)

. describe

Contains data from https://www.stata.com/users/youtube/rawdata.dta
 Observations:         1,268                  Fictitious data based on the
                                                National Health and Nutrition
                                                Examination Survey
    Variables:            10                  6 Jul 2016 11:17
                                              (_dta has notes)

 
Variable      Storage   Display    Value
    name         type    format    label      Variable label
 
id              str6    %9s                   Identification Number
age             byte    %9.0g                 
sex             byte    %9.0g                 Sex
race            str5    %9s                   Race
height          float   %9.0g                 height (cm)
weight          float   %9.0g                 weight (kg)
sbp             int     %9.0g                 Systolic blood pressure (mm/Hg)
dbp             int     %9.0g                 Diastolic blood pressure (mm/Hg)
chol            str3    %9s                   serum cholesterol (mg/dL)
dob             str18   %18s
 
Sorted by: id

The description tells us the variable sbp contains measurements of systolic blood pressure measured in millimeters of mercury (mmHg). Let's summarize sbp.

. summarize sbp


    Variable 
 
        Obs        Mean    Std. dev.       Min        Max
 
 
 
         sbp 
 
      1,268    131.1554    29.43287         65        720

The miniumum value of sbp is 65 and the maximum is 720. sbp is a continuous variable, but sometimes researchers dichotomize systolic blood pressure into the categories "less than or equal to 120 mmHg" and "greater than 120 mmHG". We can create this dichotomous variable using Stata's recode command.

. recode sbp (min/120 = 0) (120/max = 1), gen(hisbp)
(1,268 differences between sbp and hisbp)

The recode command creates a new variable named hisbp from sbp. It maps values of sbp from the minimum to 120 to category "0" in the variable hisbp. And it maps values of sbp greater than 120 to the maximum value to category "1" in the variable hisbp. Let's use Stata's summarize command with bysort to check our work. You can type help bysort if you are not familiar with the bysort command.

. bysort hisbp: summarize sbp



-> hisbp = 0
 
 

 
    Variable 
 
        Obs        Mean    Std. dev.       Min        Max
 
 
 
         sbp 
 
        509    109.0216    9.226669         65        120



-> hisbp = 1
             
    Variable 
 
        Obs        Mean    Std. dev.       Min        Max
 
 
 
         sbp 
 
        759    145.9987    29.00644        122        720

The minimum and maximum values of sbp are 65 and 120, respectively, for category "0" of hisbp. And the minimum and maximum values of sbp are 122 and 720, respectively, for category "1" of hisbp.

We could have labeled the categories of hisbp in our recode command. Let's use recode again to create a new variable named hisbp2. This time type labels with double quotes inside the parentheses after each category number. Then type label list to view the label definition.

. recode sbp (min/120 = 0 "<=120") (120/max = 1 ">120"), gen(hisbp2)
(1,268 differences between sbp and hisbp2)

. label list
hisbp2:
           0 <=120
           1 >120

Category "0" is labeled "<=120" and category "1" is labeled ">120". Note that recode created the label definition hisbp2 and attached it to the variable hisbp2. It has done the work of label define and label values for us (type help label if you are not familiar with the label commands).

Let's list observations 507 through 511 to see the labels in our dataset.

. list sbp hisbp hisbp2 in 507/511


     
 sbp   hisbp   hisbp2 
507. 
 106       0    <=120 
508. 
  98       0    <=120 
509. 
 100       0    <=120 
510. 
 146       1     >120 
511. 
 128       1     >120 

We used recode to create two categories of sbp, but we could have created three or more. Here's a quick example using triple slashes to type the command on multiple lines.

. recode sbp (min/100 = 1 "<=100")   ///
             (100/129 = 2 "100-120") ///
             (120/max = 3 ">120")    ///
             , gen(hisbp3)

You can watch a demonstration of these commands by clicking on the link to the YouTube video below. You can read more about these commands by clicking on the links to the Stata manual entries below.

See it in action

Watch Data management: How to create a categorical variable from a continuous variable.