summarize information not accurate for age variable

[Krystle Wittevrongel]

hi all,

I am new to Stata and using it (15.1) for my graduate thesis work with administrative health data.

I have > 200,000 records, 1 for each patient visit, with numerous variables. I have 18 values for age (from 0 to 17). When I summarize to get the descriptive stats it's tell me min is 1 and max is 18, which are not actually accurate which leads me to believe that the mean is also incorrect. Do I need to somehow recode the ages to be read correctly? Or is the fact there is zeroes present (appropriately so) throwing it off?

Thanks so much.


Krystle

[William Lisowski]

Welcome to Statalist

I believe your data was read into Stata with age treated as a string variable, and you used encode rather than destring to create a numeric variable.

. input str2 age_s

         age_s
  1. 0
  2. 1
  3. 2
  4. 3
  5. 4
  6. 5
  7. 6
  8. 7
  9. end

. destring age_s, generate(age1)
age_s: all characters numeric; age1 generated as byte

. encode age_s, generate(age2)

. list, clean

       age_s   age1   age2  
  1.       0      0      0  
  2.       1      1      1  
  3.       2      2      2  
  4.       3      3      3  
  5.       4      4      4  
  6.       5      5      5  
  7.       6      6      6  
  8.       7      7      7  

. list, clean nolabel

       age_s   age1   age2  
  1.       0      0      1  
  2.       1      1      2  
  3.       2      2      3  
  4.       3      3      4  
  5.       4      4      5  
  6.       5      5      6  
  7.       6      6      7  
  8.       7      7      8  

. codebook

------------------------------------------------------------------------------------------------
age_s                                                                                (unlabeled)
------------------------------------------------------------------------------------------------

                  type:  string (str2), but longest is str1

         unique values:  8                        missing "":  0/8

            tabulation:  Freq.  Value
                             1  "0"
                             1  "1"
                             1  "2"
                             1  "3"
                             1  "4"
                             1  "5"
                             1  "6"
                             1  "7"

------------------------------------------------------------------------------------------------
age1                                                                                 (unlabeled)
------------------------------------------------------------------------------------------------

                  type:  numeric (byte)

                 range:  [0,7]                        units:  1
         unique values:  8                        missing .:  0/8

            tabulation:  Freq.  Value
                             1  0
                             1  1
                             1  2
                             1  3
                             1  4
                             1  5
                             1  6
                             1  7

------------------------------------------------------------------------------------------------
age2                                                                                 (unlabeled)
------------------------------------------------------------------------------------------------

                  type:  numeric (long)
                 label:  age2

                 range:  [1,8]                        units:  1
         unique values:  8                        missing .:  0/8

            tabulation:  Freq.   Numeric  Label
                             1         1  0
                             1         2  1
                             1         3  2
                             1         4  3
                             1         5  4
                             1         6  5
                             1         7  6
                             1         8  7

.

You need to return to your original data and use destring, rather than encode, to convert numbers stored as strings to numeric variables - age, and any other variables for which you did that.

Better still, you should start by trying to understand why your age values were treated by Stata as strings. If it was because some values contained text like "UNK" for unknown, then you need to handle that differently, by replacing the "UNK" values with a string representing a Stata numeric missing value (. or .a through .z) and then applying destring.

. input str3 age_s

         age_s
  1. 0
  2. 1
  3. 2
  4. 3
  5. 4
  6. UNK
  7. 5
  8. 6
  9. 7
 10. end

. replace age_s = "." if age_s == "UNK"
(1 real change made)

. destring age_s, generate(age1)
age_s: all characters numeric; age1 generated as byte
(1 missing value generated)

. list, clean

       age_s   age1  
  1.       0      0  
  2.       1      1  
  3.       2      2  
  4.       3      3  
  5.       4      4  
  6.       .      .  
  7.       5      5  
  8.       6      6  
  9.       7      7

[Krystle Wittevrongel]

thank you so much for your response. It does appear there was a single nonnumeric entry which is why encode was used over destring. I have used gen age1=real(age) and that appears to have solved the problem as well. Does this equate to what you proposed?

[William Lisowski]

It does equate to what I proposed, as would

destring age, generate(age1) force

The reason they equate is because you have created a new variable, preserving the original, and have taken the time to precisely what the problem was - a single non-numeric entry. That was the important part of my proposal, not the command itself. As a result, you've (a) preserved the original data (not so much an issue since there was just a single anomaly) and (b) ensured that you understand your data. Too many users just apply destring, replace force and lose important information.

From this you've learned that, as the output of help encode tells us,

Do not use encode if varname contains numbers that merely happen to be stored as strings; instead, use generate newvar = real(varname) or destring

[Krystle Wittevrongel]

thank you so much!

[Nick Cox]

The name force for the option really is intended to convey that its use is not preferred over caution and consideration.