About this Q and A
Some basic rules

Answering Machines and Voice Mail

Cellular and Mobile Phones

Fax and Computer Lines

Multiple Phones in a Household

Unusual Household Situations

Vacation Homes

Group Residences and Dorms

Bad phone numbers

Interviewer errors

These questions arose during various telephone surveys conducted by SRC-Berkeley. The answers were provided by Tom Piazza, SRC Manager of Statistical Services. We welcome any discussion of the advice given here. Contact Tom at piazza@csm.berkeley.edu.

Many of the questions concern the appropriate outcome or disposition code to assign to a case. Some outcomes mean that a case is excluded from the sample for purposes of calculating response rates -- unlike refusals or respondents who can never be found at home. See the discussion of basic rules for some general guidance on this matter.

Most of the questions in this Sampling Q & A deal with the issue of whether or not a selected telephone number, in a random-digit sample, should be considered a valid residential number.

If the phone number is a residential one, it must be used in calculating the response rate, computed as the number of completed interviews divided by the total number of residential phone numbers in the sample.

If the phone number is NOT a residential one, it can be excluded from the denominator in the calculation of response rates. The issue is often to decide which outcome code or disposition to assign to a specific case.

Outcome codes that DO count against the response rate are the following:

Direct Refusal

Informant has refused to enumerate the household, or the selected respondent has refused to complete the interview.

Indirect Refusal

The interviewers were unable to get past a residential answering machine to talk to anyone, or a respondent has been selected but was never available to complete the interview,

Out of Town for Duration

The respondent (or everyone in the household, if this can be ascertained) is out of town for the duration of the study.

Unable to Participate

The respondent is eligible to be interviewed but is unable to participate because of some illness or disability.

Not in Service

Not a working phone number

Not a Residence

Business, computer line, etc.

Assumed not a Residence

This is a key category, since the status of many telephone numbers cannot be determined with certainty, and different organizations will devote different degrees of effort to making this determination. See below the rules followed at SRC-Berkeley.

Ineligible

No one in the household falls within the definition of the survey population. (For instance, the project may need to screen for certain age groups.)

Practical difficulties often arise because various calls to the same telephone number result in outcomes that are inconclusive or apparently inconsistent. For example, a phone number that usually rings (suggesting that it is a working number) can sometimes produce funny beeps (suggesting that it is Not-in-service). The problem is to develop rules to deal with these situations.

The procedures currently followed at SRC-Berkeley to resolve the status of a phone number with inconclusive calling outcomes -- that is, a phone number that no one has ever answered AND at which no answering machine with a message has ever been encountered -- can be summarized as follows:

• Funny signals, beeps, fast-busy signals, and messages indicating circuit problems and temporary disconnects MAY only be temporary problems, and the phone number could be a working number. But after 3 such outcomes, (each of which involves an initial call plus a second call to confirm that the correct number was dialed) the number can be coded Not-in-service.

• Modem signals MAY be consistent with a residential number, since many people now have computers and fax machines at home. But after 3 such outcomes, (each of which involves an initial call plus a second call to confirm that the correct number was dialed) the number can be coded Not-a-residence.

• A phone number that rings but is never answered, or is always busy, MAY be a residential number, and it needs to be called many times before excluding it from the sample. But the number can be coded Assumed-not-a-residence under the following conditions:
  • A total of 18 calls have been made to the number.
  • The calls include at least:
    • 1 Mon-Thurs morning (9-12 am)
    • 1 Mon-Thurs afternoon (12-6 pm)
    • 4 Mon-Thurs evening (6-9:30 pm)
    • 2 Saturday
    • 2 Sunday (4-9:30 pm)
  Although a small proportion of these numbers may be residential, they are most likely to be disconnected commercial numbers or simply not in service.

• A phone number with a combination of the above outcomes (funny signals, modem signals, no-answer, busy) is coded according to the particular limit reached first. For instance, after 10 calls, of which 7 are no-answers and 3 are funny signals, the phone number will be finalized as Not-in-service because the limit of 3 funny signals has been reached.

(TP 2/23/96)