The telephone household survey is conducted during six discrete dialing periods throughout the year. This approach may require a large pool of part-time interviewers because of the non-continuous nature of the work. Interviewers must have strong communications skills and be able to interact with people in a friendly and professional manner. Experience in telephone interviewing is desirable to effectively deal with a variety of situations while conducting the interview.
At least one interviewer per shift must be bilingual (English-Spanish). All contacts requiring a Spanish-speaking interviewer must be referred to this interviewer, who would administer the questionnaire using a Spanish translation.
An extensive training session must be held for all personnel who have not worked on the survey in previous waves. This training must cover general telephone interviewing procedures as well as procedures specific to the MRFSS. Subjects addressed in the telephone survey training sessions must include, but not necessarily be limited to:
o Recording call attempts and completions,
o Screening respondents for eligibility,
o Setting appointments and making callbacks,
o Overcoming respondent resistance and discouraging refusals,
o Recording answers,
o Proper probing and clarifying imprecise or confusing responses,
o Reading the questionnaire verbatim,
o Obtaining complete answers,
o Random digit telephone number generation,
o Assuring confidentiality of responses,
o Awareness of county quota systems, and
o Callbacks to verify high numbers of reported trips.
Training sessions must include a general overview of the background, purpose and design of the MRFSS. Questions and discussion should be encouraged to ensure that all interviewers understand the importance and overall purpose of the study. This overview must be followed by an item-by-item explanation of the data collection instruments and a review of all materials used in conduct of the interview.
Each training session must contain periods of role playing to result in good interviewing technique. All trainees must conduct practice interviews with supervisors to allow first-hand criticism of their interviewing technique.
All interviewers, including those employed during previous waves, must receive a final briefing before the start of each wave's dialing period. This briefing will refresh techniques established in previous waves, review the basic details of the study, explain the exclusions in questionnaires that are applicable to that wave and point out any changes in forms. Questions would be strongly encouraged during these final briefings.
Project supervisors must oversee the operation of the telephone household survey. These individuals must be experienced in telephone interviewing and be knowledgeable about the MRFSS. They must also have effective skills in managing and motivating personnel.
Supervisors must validate interviews by recontacts and/or monitoring of interviews in progress (see Section 3.5.1). Additional training or other remedial action must be taken when appropriate. Interview monitoring allows supervisors to offer suggestions to interviewers to help improve their interviewing technique.
Supervisors must review all completed questionnaires on a daily basis during the dialing period. They must check to see that coding forms are completed fully and accurately. Any apparent mistakes or inconsistencies must be checked with interviewers and, if necessary, the respondent would be recontacted to clarify an answer. Respondents who report an abnormally high number of trips (greater than the 95th percentile for each wave, state and mode calculated from the five most recent years of data, exclusive of the current year) should be recontacted for verification.
Counties with any part of their boundary within 25 miles of the coast or major bays or estuaries are considered coastal counties and are included in the telephone household survey. There are several exceptions to this definition:
1. The boundary is extended to 50 miles in the South Atlantic and Gulf subregions from May through October.
2. Some counties on the Pacific Coast inland of the 25 mile zone are included since they represent metropolitan areas that contained anglers known to go saltwater sportfishing.
3. North Carolina coastal counties are within 50 miles of the coast from November through April and within 100 miles of the coast from May through October.
For the 1995 MRFSS, NMFS allocated 208,000 telephone households among Atlantic and Gulf states in proportion to fishing effort. Allocations are expected to be similar in 1996. On the Pacific coast (California and Oregon), the allocation is expected to be about 55,000 telephone household interviews. The effort index used for 1993-1995 was a three year (1990-92) average of coastal county resident trips by state. Within each subregion/state, a base level of 200 telephone households was allocated for each wave to assure that sufficient data were available to produce estimates. Allocations among waves are also proportional to effort. County allocations are proportional to the square root of the number of full-time, occupied households with telephones in a coastal county divided by the sum for the state of the square root of the number of full-time, occupied households with telephones in each coastal county.
A national sample frame of blocks of coastal county household telephone working numbers has been developed under past MRFSS contracts. Each block must have had at least one residential number assigned to be included in the sample frame. Blocks are the first five digits of every telephone number within each area code. The first three digits (prefix or exchange) are assigned to specific geographical areas by the telephone companies. Prefixes must be used in meeting allocations of calls by county. The fourth and fifth digits are often designated as business or residential use, so blocks including business numbers can be screened out of the sample frame for more efficient dialing. Unassigned blocks have also been eliminated from the sample frame. Since not all blocks are assigned by the telephone companies, random digit dialing including these blocks would result in wasted effort.
The sample frame must be maintained on a continuing basis through the use of current telephone directories, reverse directories (sorted by blocks instead of names) and information from telephone companies. This information is released in a revolving manner rather than once a year, so the updating process must be done on a continual basis. Using this approach, the most current telephone sample would be available for generation of telephone numbers for each dialing period.
The number of households to be dialed in each county must be sub-allocated by screened telephone blocks. The number of calls to each block must be proportional to its occurrence in the population. For example, if 3.7 percent of all telephones in the county begin with 274-45, then 3.7 percent of all interviews must be conducted with households in this block. The last two digits are generated randomly. This approach ensures that all telephone households, even those with unlisted numbers, are eligible to be reached in the survey.
If the number dialed has been disconnected and a new number assigned, the new number should not be dialed as a replacement. The primary interest of the survey is not the household, but the randomly generated telephone number and the results obtained when dialing that number.
Duplicate telephone numbers must be screened out of the generated set of random numbers for each wave; however, with replacement sampling across waves, it is possible that in sparsely populated counties households will be contacted that have been interviewed during previous waves. Special tracking procedures must be used to ensure that no more than one percent of the households contacted are included in the sample more than once during the year.
All household dialings must be completed within a two week period during the last week of each wave and the first week of the next wave. This approach results in a uniform recall period of 60 days for all respondents that most closely matches the two month wave. If all calls were made during a two week period after completion of the wave and anglers were asked about their trips during the specific wave time period, then the recall period could vary from 60 days to 75 days depending on when the interview took place.
Telephone calls must be made during that time of day that maximizes the potential to contact individuals. At least five attempts must be made on each number to screen for households and their eligibility for an interview. These attempts must be stratified by weekend/weekday and day/evening as listed below. At least five additional attempts must be made to reach eligible anglers in each household. That is five attempts per household, not per eligible angler in the household. When each number is dialed, the telephone should be allowed to ring five times before the interviewer classifies it as a "no answer." The percentage of "no answer" or "busy" results must not exceed 10 percent of the total calls attempted in any subregion.
The pattern of dialing will be such that each number will have at least one weekday attempt and three night or weekend attempts, at least one of which must be a weekend attempt. The time delineating day and night is 5 p.m. Once a household is contacted, call-backs must be made on an appointment basis. Household members should be questioned as to the best time to call back in order to interview the eligible respondent(s).