Automated open-end coding

Open-end coding converts open-ended responses (like the word "coke") into closed-ended data (like a checkbox question).

This process involves two distinct tasks, building a code frame (checkbox options) and coding the responses (assigning answers to checkboxes). Clients may provide a code frame or base it on prior research. The process can also be iterative (e.g., if "Other" grows too large, new codes may be added.)

You can code open-ends automatically during fielding, or you can request to do this after fielding.

Automated during fielding: Works best for short, well-defined responses (e.g., brand names).
After fielding: Used for longer or more complex responses.

Note: Open-end coding is considered a customized survey request. IntelliSurvey programmers should discuss additionally pricing internally.

For large datasets, particularly when coding occurs post-fielding, outsourcing may be an option.

For additional assistance, contact Support or send an email to help@intellisurvey.com.

Example

If responses are short and well-defined (e.g., company or brand names), IntelliSurvey can program automated coding. Below is an example for coding soda brands.

This methodology works for single words or short phrases only.

1. What is the first soda brand that comes to mind?
type: text
 
1OEX. Soda brands [will be populated automatically]
type: checkbox
translate: n
required: n
invisible: y
1. Coke
2. Sprite
3. Dr. Pepper
4. Mountain Dew
5. RC Cola
6. Sierra Mist
7. Pepsi
99. Don't Know/Not sure

set value
question: Q1OEX
value: <<END
# arrays of strings; each string can contain regular expression syntax
my @opt1 = ("coke", "coco", "coca", "koce", "koka", "koke", "Koco", "Koko", "Coc", "Cok");
my @opt2 = ("Sprite", "Sprit", "spite", "Spriet");
my @opt3 = ("Pepper", "Peper", "Pepa", "Dr.", "Peppa", "Peppar", "Pebbar", "Pebar", "Peppr", "Pebber", "Peber");
my @opt4 = ("Mountain", "Mountian", "Moutaine", "Dew", "Drew", "Montane", "Montain");
my @opt5 = ("RC", "RC Cola", "Cola RC", "Cola (RC)", "RC Kola");
my @opt6 = ("Sierra","Mist","Seirra","Siera");
my @opt7 = ("Pepsi", "Pesi", "Pespi", "Peps", "Pepi", "Peppi");
my @opt99 = ("don't know", "dont know", "don't no", "dont no", "none", "donno", "nothing", "non", "null", "dunno", "nada", "na");
my @f;
push @f, 1 if grep {$Q1 =~ /$_/i} @opt1;
push @f, 2 if grep {$Q1 =~ /$_/i} @opt2;
push @f, 3 if grep {$Q1 =~ /$_/i} @opt3;
push @f, 4 if grep {$Q1 =~ /$_/i} @opt4;
push @f, 5 if grep {$Q1 =~ /$_/i} @opt5;
push @f, 6 if grep {$Q1 =~ /$_/i} @opt6;
push @f, 7 if grep {$Q1 =~ /$_/i} @opt7;
push @f, 99 if grep {$Q1 =~ /$_/i} @opt99;
\@f;
END
condition: $Q1 ne ""

Key considerations:

Case Sensitivity: The /i flag makes matching case-insensitive (e.g., "COKE" = "coke"). To enforce case sensitivity, remove /i.
Trade-offs: Avoid using very short strings (e.g., “na” in “Canada Dry” may trigger an incorrect match for “Don’t Know”).
Variant optimization: If a shorter version covers all cases, longer variations aren't needed (e.g., "Cok" covers "Coke").

Search

Example

Comments

Example

Related articles