Question of replace

[Kelsi Cheng]

My dataset contains a varible of sex, the formate of which is string and the value of which contain male and female. I want to replace male with 1, and female with 0, so I write command as follows, but just get such results:
1. replace sex=1 if sex=="male"
type mismatch
2. replace sex=1 if sex="male"
invalid syntax

so what is the problem and how to correct it?
Thanks so much for answering

[Justin Niakamal]

You're missing an = in #2. In Stata, as in other programming, = is an assertion. You are setting something equal to something else. In contrast, == is conditional.
Read more about = vs == here.

 replace sex=1 if sex =="male"

[Kelsi Cheng]

You're missing an = in #2. In Stata, as in other programming, = is an assertion. You are setting something equal to something else. In contrast, == is conditional.
Read more about = vs == here.

replace sex=1 if sex =="male"

but i write as your code in my first line, and it says type mismatch

[Justin Niakamal]

Your variable is a string (I overlooked that). Try

 replace sex= "1" if sex == "male"

[Sven-Kristjan Bormann]

If you don't insist on using a string variable, then you could convert the variable to a numeric variable with -encode-

encode sex, generate(sex_num)

[Kelsi Cheng]

I also tried this, but still does'nt work

[Kelsi Cheng]

Your variable is a string (I overlooked that). Try

replace sex= "1" if sex == "male"

I found the format of sex is int ! As follows, isn't it weird? it has value ranging from 1-9, but when listed, it indicate"male" and "female"

[Joseph Coveney]

I found the format of sex is int ! As follows, isn't it weird? it has value ranging from 1-9, but when listed, it indicate"male" and "female"

The variable has values labels, so that although it is integer, when you list or tabulate it will display the text of the value labels and not the underlying numerals. Try

list sex in 1/5, nolabel

to see the underlying numeric values.

Try

label list V104

to see all of the (presumably nine) value labels for sex type that are assigned to the numerals.

[Kelsi Cheng]

The variable has values labels, so that although it is integer, when you list or tabulate it will display the text of the value labels and not the underlying numerals. Try

list sex in 1/5, nolabel

to see the underlying numeric values.

Try

label list V104

to see all of the (presumably nine) value labels for sex type that are assigned to the numerals.

wow! It is really label!

As follows:
. list sex in 1/5, nolabel

+-----+
| sex |
|-----|
1. | 1 |
2. | 2 |
3. | 1 |
4. | 2 |
5. | 1 |
+-----+

. label list V104
V104:
1 Male
2 Female
9 Missing
Then how should I handle this, remove the label?
THX soooo much!

[Joseph Coveney]

I want to replace male with 1, and female with 0

Then how should I handle this, remove the label?

There is a command (_strip_labels) that will remove the labels, but removing the label alone won't recode your variable in the way that you want. So, instead, I recommend creating a new variable that has the coding that you want.

assert !missing(sex)
generate byte sex_n = sex == 1
quietly replace sex_n = .m if sex == 9

[Kelsi Cheng]

There is a command (_strip_labels) that will remove the labels, but removing the label alone won't recode your variable in the way that you want. So, instead, I recommend creating a new variable that has the coding that you want.

assert !missing(sex)
generate byte sex_n = sex == 1
quietly replace sex_n = .m if sex == 9

Thanks sooo much! It does work. I just have a little doubt: this method generate a new varible, just defining situation of male , why automatially make female 0?
And another similar situation, but not only have 2 value(0 and 1), but a continous value from 1-9, for example one's occupation group(also have label values), how to modify the code?
sorry for disconnected these days, just see your answer, sincere thanks again!

[Joseph Coveney]

I just have a little doubt: this method generate a new varible, just defining situation of male , why automatially make female 0?

I'm not quite sure whether I understand your question. Are you asking me why did I choose to make female 0? If so, then it's because you asked to:

My dataset contains a varible of sex, the formate of which is string and the value of which contain male and female. I want to replace male with 1, and female with 0

Or are you asking how does this method of generating the new variable automatically make female 0? If so, then it's the result of Stata's binary (two-value) logic rule.

generate byte X = Y == Z

makes all values of X to be zero (i.e., false) where Y is not Z (i.e., for those observations where the proposition Y == Z is false), and makes all values of X to be 1 (i.e., true) where Y is Z (i.e.,for those observations where Y == Z is true) .