Editor's Note: R. Christopher Horak is director of the Elm Tree Alliance, a non-profit research group in San Francisco.

Call result codes show an amazing lack of consistency across different research organizations. To an extent, this is predictable, given that these codes have been independently developed to meet varying needs, and subsequently copied and recopied by other firms with still other needs. Most of these codes were initially designed by researchers and theoreticians. Rarely if ever have the interviewers who use these codes been asked for their input.

For instance, an interviewer needs to know if the respondent has been briefed on the purpose of the call. If the respondent has been briefed, then the interviewer and the operations management end up sounding like idiots when the interviewer starts with "Hi! I'm Christopher Horak with Ajax Research. We're conducting a brief study..." instead of "Hi Mr. Jones! Do you have time to do that survey with me now?" Yet in 20 years of research, I've seen no set of call result codes that made this difference clear.

It would be wise to implement a universal coding system. Many interviewers work for several different firms, frequently nationwide, even worldwide (one of the reasons that those who like to travel gravitate to the trade is that it reliably offers a usable, if not lucrative, trade wherever you go). Having so many systems decreases productivity and increases the potential for miscommunication (especially because many are similar, but different in small ways) since the interviewer has to learn a new system with each firm and avoid confusing that system with any others previously used.

Given the common use of CATI (computer-assisted telephone interviewing) systems, and the fact that many companies keypunch final call results summaries and tallies from paper-form surveys into the data system, it would be best if this universal system of call result codes were numerical. Numbers are also universal across most languages. With this as a start, having spoken to literally hundreds of interviewers over the years about these problems, and with the great creative assistance of Daryl Bunting, we have developed the proposed system below.

The rest of this article will be an abbreviated form of that which we give to the interviewers.

Call result code categories

The first 20 codes (0 through 19) are listed individually, and pertain to call results that are for the most part universal to any telephone study. Code categories in the 20s, 30s and 40s are "study specific," and pertain to respondent qualifications unique to that particular study. Codes in the 50s and 60s generally are used with CATI and data entry systems, and while they are not usually the interviewer's concern, it is necessary to be at least familiar with how they are used.

The first 20 codes are divided into "call backs" and "non-callbacks." The first 10 "call back" codes refer to call results that permit an interviewer to call that number or individual again to complete an interview. These constitute "live" sample after the initial call. The second 10 codes are "non-call back" categories, and consist of situations in which no further contacts are warranted ("dead" sample). The 20s, 30s, and 40s are usually "dead" but may be called back under certain circumstances or on another study.

Call back codes

0 PARTIAL CALL BACK: This code refers to a situation in which a qualified respondent answers a portion of the main questionnaire, but for some reason cannot continue at that time. Importantly, the respondent in this case agrees to be called back so that the interview can be completed. In such a case, please transfer all pertinent information to the written survey form, including page and/or sample number, so that the sample can be tracked down after completion, and so that final call results can be brought current. If you are working on a CATI system, this will be done automatically.

1 BUSY (Specify type*): Both short (major trunk line) and long (individual telephone line) busy signals are coded a 1. In the first case, call back after lunch in that time zone or when whatever-is-keeping-the-call-volume-high-in-that-area is over. In the second case call back in 20 to 40 minutes.

2 NO ANSWER: If you are calling a residential number and no one answers the phone after four to six rings, it is coded a 2. You should not allow more rings unless you are calling a business or contacting a respondent at a prearranged time. For businesses, up to 10 rings are allowed to give time for the harried receptionist to answer the phone.

3 ANSWERING MACHINE/SERVICE/VOICEMAIL: Most answering machines kick in within six rings. Whether an answering machine or an answering service, do not leave a message unless the study specifications authorize it. However, on the first call, you should listen to the message or talk to the operator to obtain any information which may help reach the respondent at another time and hopefully avoid the answering machine/operator again. This code also includes getting lost in impossibly complex voice mail structures or being put on eternal hold.

4 CALL BACK (Specify when,* if possible. If it is a CATI system, be sure to know what the parameters are for an unspecified call back -- usually four hours. A good system will allow you to specify the parameters [i.e., call back frequently between 4 p.m. and 5 p.m.] in case you have been able to determine them from talking to a secretary, assistant, family member, friend, or colleague.): In all cases, someone has answered the phone, but a qualified respondent has not yet been briefed on the purpose of the interview, and no telephone number is left. Always indicate the call back in local time. If it is a CATI system you may have a choice of specifying whether you are recording the callback in either the respondent's time zone or in your time zone. It's best though to get in the habit of converting the respondent's time into your own.

5 CALL BACK (Specify when:* respondent briefed): When a qualified respondent has been identified, briefed on the study purpose and agrees to participate at another time, determine a specific date and time using code 5 for the call result. The key difference between codes 4 and 5 is that with a 4, the prospective respondent has not been briefed on the purpose of the call, while with a 5, the prospective respondent has been briefed on the purpose of the call.

6 LEFT MESSAGE (May call back): This code is seldom used for consumer surveys. In focus group recruiting and business surveys it is occasionally desirable to leave a phone number with a secretary or on voice mail. State your name, the purpose of your call, and ask the potential respondent to return the call. Leave the page or record number (call it an extension number if you must), so that you will be able to find the appropriate sample without difficulty (a good CATI system should make it possible to call up a particular record at any time,* but failing that, you can still do the survey on paper, and find the appropriate sample point later).

7 WILL CALL (Specify when to call back if no call*): Here, the potential respondent has been reached, cannot be interviewed at the time of the call, and has volunteered to (insisted on) call(ing) back at another time. It is suggested that the interviewer establish an agreed upon time to recontact the respondent, should they be unable to call back by the time scheduled. Again, like code 6, this is rarely a permissible code for consumer surveys. Here, too, remember to leave a record or sample page number.

8 REFERRAL NUMBER (Call back): There are two basic types of code 8. The first is a telephone number change announced by an automated operator. The second is a receptionist or any person who gives a number where a qualified respondent can be reached. Be sure to record the new number and draw a line through the old.* When you attempt the new number, the call result will receive its own appropriate result code. If the new number is out of the area acceptable to study specifications, the call is coded an 18 (non-call back).

9 OTHER CALL BACK (Specify*): This code is used to cover special situations particular to study requirements. For example, if a respondent is supposed to have received or used some item prior to the interview and has not yet done so, the call is coded a 9 and the respondent is not called back for an appropriate amount of time to allow receipt or use of the item. In all cases, record all circumstances on both the sample and the daily tally sheet.*

Non-call back codes

10 COMMUNICATIONS BARRIER: In general, this category includes all instances where the interviewer can't adequately communicate with a respondent to complete an interview. Most commonly this code is used when the respondent does not speak the language in which the interview is to be conducted. If an interviewer is bilingual, it is permissible to communicate with a respondent in a different language about the purpose of the call, but the interview itself must be conducted in the original language. Examples of other occasions when a code 10 is used are when a respondent is sufficiently hard of hearing, has Parkinson's or Alzheimer's disease, is mentally unstable, under the influence of drugs, or not intelligent enough to negotiate the questions. Please note that under no circumstance are you to give evidence of disapproval or judgment of any kind, whatever your personal opinion might be.

11 NOT AVAILABLE DURING STUDY/INTERVIEWING HOURS (Specify CB time*): This code is used when a respondent cannot be reached for an interview, either during the hours of the interviewing shift or during the time period scheduled for the study. It is important to specify the time/date/hours of availability in case interviewing hours are changed or extended to incorporate these respondents.* This code also applies when a qualified focus group respondent is unable to attend a focus group discussion because of a schedule conflict.

12 TERMINATE: When a respondent fulfills all qualifying (screener) questions, but quits because of boredom or irritation before the interview is completed, it is coded 12, terminated. Here, the respondent does not agree to be called back for interview completion. On rare occasions, an interviewer may choose to terminate an interview if the respondent is excessively abusive on the phone, falsifies answers, or makes the interviewer's job of completing the interview too difficult.

13 REFUSE: A 13 is used when a respondent hangs up or refuses to participate prior to answering all qualifying (screener) questions on the main questionnaire (if a respondent has answered all qualifiers before refusing to answer any more questions, that is classified a "terminate," not a refusal [see above]).

14 DISCONNECT/ NOT WORKING: This code refers to operator announcements that the sample telephone number has been disconnected or is no longer in working order. Occasionally a recording will say that the number is temporarily out of order. Code this as a 14 also. Experience has shown that less than ten percent of these phones will be dialable before the end of the study. Do not confuse this with an 18, particularly a referral to an unpublished number.

15 WRONG NUMBER: Occasionally, the dialing sample will give a listed number that is incorrect for a specified individual, business or household. The key to using 15 is that the business or person reached is not what the sample said it would be (obviously, you have to be using a type of sample that specifies this information). Do not confuse this with a 50, a number which the interviewer has dialed incorrectly.

16 PHONE LOCATION NOT QUALIFIED: In consumer surveys, sample numbers connecting to businesses, government offices, telephone booths, etc., -- in short, locations other than residences - are coded 16. In business surveys, an unqualified phone location might be a residence, phone booth, or an office where corporate decisions are not handled and no referral number is available.

17 FAX/MODEM/PAGER: If, after dialing, you hear a piercing tone or series of tones/beeps, and no human voice, you have connected with one of the above pesky devices. However, should someone answer a fax number and identify it as such, it is still coded 17, unless you are able to get a referral to another number, in which case it will be coded an 8 or an 18. If the number serves as both a voice and data line, you may do the survey or list the call result as a call back if you can reasonably expect to speak to the respondent. Otherwise code it as a 17. When a prompt asks you for an access or security code it is most often a preliminary to modem communications. While occasionally these are also voice lines to secure installations, code these a 17 as well.

18 REFERRAL NUMBER (Non-call back): This code is used when you are referred to a number for a qualified respondent and that number is beyond the geography of the study. This is especially true in focus groups when the decision maker may have to drive three hours to get to a meeting, or when a survey is specific to a region and you are referred to a national headquarters outside of the area. Be sure to write down the referral number on the sample since it may be useful at a later date.* If you hear a recording that tells you that the number has been changed to an unpublished number, code this an 18 as well.

19 OTHER NON-CALLBACKS: These tend to be unusual or extreme circumstances, such as a respondent in the process of moving, in the midst of a calamity, or recently deceased. Record all circumstances on both the sample and the daily tally sheet.* If these occur often enough, usually more than three percent of the time, they may be included in the final call result tally given to the client.

Other codes

20s NOT QUALIFIED: These codes are used for specific qualifier (screening) questions asked either prior to or at the beginning of the main questionnaire, designed to determine respondent eligibility. If a respondent doesn't qualify for the interview the code 20 is added to the number of the question which disqualifies the respondent. The following is an example:

Question 5a. How old are you?

Under 21...................................1 (THANK AND TERMINATE)
21-30.......................................2 (CONTINUE)
31-40.......................................3 (CONTINUE)
41 and over...............................4 (THANK AND TERMINATE)

If the respondent is 47, then the sample is coded 25a4. The two (25a4) conveys the respondent was not qualified, the five a (25a4) tells at what question number the interview was terminated, and the four (25a4) indicates the category of the question in which the disqualified respondent fell. For some reason the basic procedure has proved somewhat confusing in practice to the interviewers. Simply put, just take the question number, and place a 2 in front of it, then append the response number if this is necessary to indicate separate categories within the same question.

30s OVER QUOTA: Some studies are structured to obtain a specific number or proportion (quota) of respondents with particular eligibility requirements, e.g., age groups, income level, use frequency, job functions, etc. As these quotas are filled, subsequent respondents with those qualifications are coded 30 plus the question number. Using the above example, if we have too many 21- to 30-year-old completes, and the only respondent in the house is 26, s/he will be coded as 35a2.

40s STUDY SPECIFIC CODES: These codes are reserved for unique situations that occur often enough to warrant analysis but are not covered by the screener, main questionnaire or any other codes. Management will always specify what circumstances will be coded with 40s.

50s: These are for call results that are not intrinsic to the sample or potential respondents, such as a phone company equipment malfunction. 50 itself is always reserved for a (interviewer error) misdialed number. This is generally used only with CATI systems, but sometimes the principle may require an exact number of calls for a study done on paper. A CATI system, after entry of a call result of 50, should return to the same record number.* Management will specify if/when you are to record a call result in the 50s.

60s: The number of callbacks allowed to obtain a complete interview varies from study to study, but is generally three to five. 60s are only used for data entry purposes. Numbers that are still "live" after the total number of permitted calls are now "dead" because the numbers have been called the maximum number of times. For data entry purposes a "busy" that has been called the maximum number of times is converted from 01 to 61, while a "busy" that has been called less than the maximum number of times remains an 01 (a good CATI system will allow call backs [04 and 05] to be conserved should a potential respondent, finally contacted on the fifth call, agree to do the survey at another time*).

XX DIRECTORY ASSISTANCE: On rare occasions, usually in business surveys or focus group recruiting, you may wish to track down a phone number by calling directory assistance. This call will be coded as XX. Keep in mind that directory assistance costs money so try to wait until you can ask for at least two numbers (three on the West Coast).*

YY COMPLETE: A completed interview. For some reason, this is the call result that is most often left unrecorded by interviewers. Do not make this mistake.

Having used this system now for over four years, the results thus far have been most satisfactory.

However, we do not consider the system to be carved in stone. It is my hope that many groups will institute its use, and then inform me of any problems so that the system may be perfected. Please write me in care of Quirk's Marketing Research Review, or leave a message on my voicemail at 415-824-6986. After a suitable period for comments it is my intention to submit this system to appropriate organizations for consideration as a standard.

* It can be hoped that anyone who is developing a new CATI system or is reprogramming an old system will include memoranda boxes and decision trees to allow the inclusion and use of the appropriate information at these points.