Sample Management

Sample Management is the management of the status and lifetime of a respondent’s sample record through the life of a CATI campaign. 

When the sample records are uploaded, as well as storing the respondent’s own data, the CATI manager adds several sample management fields to the sample record. Some of these may update every time the sample is called regardless of whether the call connects of fails. These fields track the status of the sample and determine if or when the sample is to be called again. 

When the sample starts it is marked as “Fresh”. It has no specific time to call. Once calling starts the first few calls may fail. The sample management keeps track of this. Sample management keeps track of what types of failures there are and how many of each type. It keeps a count of these. Each time a call fails, a re-call time will be set. If a call fails to many time (see recycle rules) it will not be called again. It will be considered as “Exhausted”.

If the interviewer is working in “Dialler mode” (PDS/predictive), then the dialler makes the call internally and the interviewer will not hear any ringing tones, busy tones or call failure messages. In this case, if a call fails the dialler will code it automatically. A call is only passed to an interviewer when it is connected

If the interviewer is working in Preview mode (Sample/Extension), then the interviewer determines when the call is made, and they hear the call progress. If the call fails, the interviewer will code the call with the appropriate code.  

When a call connects (and is not automatically detected as an answer machine/voice mail) the failure counts will get reset to 0. At this point, the respondent should be talking to an interviewer. If the interview is “Refused” or continues to the end, then the interviewer codes the call accordingly and the code should indicate that the sample will not be called again. If the Interview does not progress to the end, then the call can be coded as an “Appointment” and a re-call time set by the interviewer.  

Sample Management Fields

• Completion Code – the code number of last completion code applied to this sample
• Code Description – the code Description of last completion code applied to this sample
• Appointment – This value (Yes/No) shows if there is an appointment set for this sample
• Recall – This value (Yes/No) shows if there is a re-call schedule set for this sample. This would be set after a call fail.
• NextCallDateTime – this is the time an appointment call or re-call is scheduled. It is set if wither Appointment=Yes or Recall=Yes.
• InProgress – This value (Yes/No) shows if the sample is in the diallers predictive pool of sample records about to be called
• Fresh – This value (Yes/No) shows if a sample has not been called yet
• Exhausted - This value (Yes/No) shows if a sample has exceeded its call failure limits (See Recycling Rules). The sample will not be called again by the dialler in predictive mode, but may be retrieved as preview and called that way
• Terminated - This value (Yes/No) shows that the sample will not be called again. The last code applied did not indicate a re-call or an appointment. It is likely to relate to a completed or refused (or similar) survey. The sample will not be called again by the dialler in predictive mode, but maybe retrieved as preview and called that way
• Tries_NOANSWER – this is the count of how many NO ANSWER outcomes have been coded. Once a respondent is contacted this is reset to 0 (See recycling rules).
• Tries_BUSY – this is the count of how many BUSY outcomes have been coded. Once a respondent is contacted this is reset to 0 (See recycling rules).
• Tries_BADNUMBER – this is the count of how many BAD NUMBER outcomes have been coded. Once a respondent is contacted this is reset to 0 (See recycling rules).
• Tries_CONGESTION – this is the count of how many CONGESTION outcomes have been coded. Once a respondent is contacted this is reset to 0 (See recycling rules).
• Tries_AMD – this is the count of how many AMD (AnswerMachine/VoiceMail) outcomes have been coded. Once a respondent is contacted this is reset to 0 (See recycling rules).
• Tries – this is the count of how many of any failed outcomes have been coded. Once a respondent is contacted this is reset to 0 (See recycling rules).

Completion Codes

Completion Codes are codes to be applied to all calls at the end of a call, or to an interview if the call was connected. Codes are associates with campaigns, and each campaign has its own set of codes that can be edited, added to, or deleted from.  The range of codes a campaign has should cover outcomes where a call does not connect (e.g. busy, no-answer, etc) and outcomes where a call does connect, and a respondent interaction occurs. On completion of an interview or after a preview call attempt, the interviewer should be presented with a list of outcome codes and must enter one before continuing. 

CATI management supports code descriptions. These code descriptions are human-friendly tags that need to be presented to interviewers for selection

For dialler (predictive) calls that do not connect, the dialler will apply a code to the call. The call does not need to be passed to an interviewer. The dialler will use a code from a set it is limited to use, and that are restricted to the reasons a call failed to connect. For coding interviewer calls (preview) an interviewer may be allowed to use any of these “failure” codes, including the ones the dialler is limited to use.

The completion codes applied to a call are stored with the Call History (callstats) table under the “Code” and “Code Description” fields. Note, the code is stored with the Call History record, so in preview mode, if the agent does not go on to make a call, no call history record will be created and no code will be associated with the sample record for that event

When the dialler receives a code for a call it will also apply the code to the sample record involved in the call/interview. Each sample record has a set of “Sample Management” fields that allow the dialler to decide if or when a sample is to be called. For instance, if a code is applied that indicates a survey has been completed, then the sample will not be called again; if a code is applied that indicates the call was busy then the sample rescheduled for a later call. If a sample fails to connect after many calls, then the codes indicating this are counted and if they exceed a limit, the sample will not be called again. See the “Recycle Rules” section.

Completion Code Fields

In the CATI manager, the set of completion codes for each campaign can be accessed. Each one contains the following fields.

• Code – a unique number relating to the code. If a new code is added the user is able to specify a code number. This should be unique. By convention codes denoting call connection failure should be < 10. A set of such code is automatically available on the creation of a new campaign.
• Description – a human-readable tag or description of what the code should indicate.
• Outcome – an optional field determining to which failure outcome the completion code is associated. This allows users to customise the completion codes associated with standard outcomes (eg BUSY, NOANSWER, AMD). If a code has an associated outcome, then when it is applied it will increment the “Tries” count in the sample record. Some outcomes (NOANSWER, BUSY, BADNUMBER, CONGESTION, AMD) have their own try counts. The try count associated with one of these outcomes will get incremented if a code with that outcome is applied (See Sample Management for more details
• Schedule – This field determines if the code should come with a date and time at which the sample should be recalled. The values for this field are NOT_REQUIRED, OPTIONAL and COMPULSORY. COMPULSORY or OPTIONAL should be used when an appointment for a call back is required. This field is passed to the interviewer’s console application. The application will make sure a data and time comes with the code. 
• Recycling – If “Yes” then the dialler will try and call the sample again. the sample is scheduled to be called back based on the AutoRecallDelta (see next item). If set to “No” then the sample will not be called back unless an appointment is made (see Schedule). The value is normally set to “Yes” for codes indicating failed call that require a re-call of the sample. The value is normally set to “NO” for codes indicating contact had been made with the respondent and no more calls are required
• AutoRecallDelta – if a sample is to be recycled (called again) but no specific time is required, then this time delta (format Day.Hour:Minute:Second) will be used to determine what the sample will be called back. It adds the time delta to the current time. It is important to understand that as sample re-call will not be set to a time outside of the activity hours of a campaign. If the addition of the delta falls in a period of inactivity it will be shifted to be within the next active period. 
• HideFromAgent – this value (Yes/No) determining if this code should not be available to an Interviewer.
• NextCallPreview – this value (Yes/No) determines if the coded sample should be previewed at next call. This might be used for an appointment re-call
• NextCallAgent – this is the interviewer name for which this sample will be given at the next call. If that interviewer is not available within 15 minutes of any scheduled call time then the sample will be given to another interviewer
• Dialler – this denotes if the code can be used by the dialler (as opposed to the interviewer) for the coding call. This code will relate to call failure codes only

Recycle Rules

Recycle rules are parameters that determine for when a re-call should be scheduled and how many re-calls are permitted. If any of the limits are exceeded, then the sample will be marked as exhausted 

There is a fixed set of recycling parameters associated with failure outcomes. The following are parameters

• NOANSWER – Max_Tries: the maximum number of tries a sample may be coded as outcome NOANSWER
• BUSY – Max_Tries: the maximum number of tries a sample may be coded as outcome BUSY
• CONGESTION – Max_Tries: the maximum number of tries a sample may be coded as outcome CONGESTION
• BADNUMBER – Max_Tries: the maximum number of tries a sample may be coded as outcome BADNUMBER
• AMD – Max_Tries: the maximum number of tries a sample may be coded as outcome AMD (Answer Machine/Voice Mail)
• MAXTRIES - Max_Tries: the maximum number of tries a sample may be coded with a combination of failure outcomes 

The sample record in the dialler keeps track of how many of each outcome the calls have been coded. However, once a connection to a respondent is made, all the counts in the sample are set to 0.