stata - How to overwrite a duplicate observation -

i conducted phone survey , here prototype of dataset:

var1    var2  6666    1 6666    2 7676    2 7676    1 8876    1 8876    2 89898   1 89898   2 9999    1 9999    2 5656    1 5656    2 2323    1 2323    2 9876    1 7654    1 

var1 unique identifier each case in survey (in case, phone numbers).

var2 outcome of survey: 1 (successful), 2 (not successful).

i want keep observations each var1 var2 == 1, yet retaining observations each var1 whosevar2 == 2 if there no case var2 == 1.

i have tried

duplicates drop var1 if var2 == 2, force 

but not getting desired output

the question wrongly titled: don't want overwrite anything.

your syntax doesn't work wish because not want. asking whether there duplicates of var1 if var2 == 2 , command pays no attention whatsoever observations var2 == 1.

your example includes no observations var2 == 2 there no corresponding observation var2 == 1. have added 1 such.

here's 1 way of meeting goal. show in passing duplicates command have nothing example; nor expected anything.

. clear   . input  var1    var2            var1       var2   1.     6666        1   2.     6666        2   3.     7676        2   4.     7676        1   5.     8876        1   6.     8876        2   7.     89898       1   8.     89898       2   9.     9999        1  10.     9999        2  11.     5656        1  12.     5656        2  13.     2323        1  14.     2323        2  15.     9876        1  16.     7654        1  17.     42      2   18. end   . duplicates list var1 if var2 == 2   duplicates in terms of var1  (0 observations duplicates)  . bysort var1 (var2) : assert _n == 1 | _n == 2   . var1 : drop if _n == 2 & var2[2] == 2  (7 observations deleted)  . list, sepby(var1)        +--------------+      |  var1   var2 |      |--------------|   1. |    42      2 |      |--------------|   2. |  2323      1 |      |--------------|   3. |  5656      1 |      |--------------|   4. |  6666      1 |      |--------------|   5. |  7654      1 |      |--------------|   6. |  7676      1 |      |--------------|   7. |  8876      1 |      |--------------|   8. |  9876      1 |      |--------------|   9. |  9999      1 |      |--------------|  10. | 89898      1 |      +--------------+ 

another way

. bysort var1 (var2) : keep if _n == 1 & var2[2] == 2  

in fact

. bysort var1 (var2): keep if _n == 1  

keeps observations var2 == 1 if there , otherwise keep singletons var2 == 2.

the hidden assumptions seem include @ 2 observations each distinct var1. note use of assert checking assumptions dataset.