|
|
| |
|
MethodologyOverview OverviewThe survey data collected by BBM come from a stratified sample of Canadians aged two years and over. Their viewing during a specified seven-day period is collected by means of a diary-type questionnaire and is recorded in quarter-hour intervals. Households included in the sample are sent one diary for each working television in the home. The diaries are designed to record the tuning of each individual household member two years of age or older. Demographic characteristics (including age group, sex, education, occupation and linguistic characteristics) are also collected on this document. The results of four seven-day periods during the fall are combined to produce the Fall Sweep Survey which forms the basis for this publication. Sample DesignThe ten provinces of Canada are geographically divided into approximately 400 mutually exclusive units called cells. For the most part, these cells are counties, census divisions, census agglomerations or other areas defined by Statistics Canada. For each cell BBM determines the number of respondents necessary to provide statistically significant data. Both the estimated size of the cell population and the level of detail (e.g. demographic groups) required are considered in calculating the required sample size. Records of response rates and statistics on the composition of families in the cell are used to calculate the number of households to which diary packages must be mailed to achieve the required sample. EnumerationTelephone directories and lists are used to draw a systematic sample independently within in each cell. Where there is more than one Post Office serving a given cell, samples are selected so that each area served by a separate Post Office is considered independently. Approximately five to twelve weeks prior to the survey start date, specially trained and supervised telephone interviewers contact the selected households and ask to speak with a responsible adult. The interviewer verifies the mailing address and records the home language, and the number of television sets in the house. Households that cannot be contacted the first time are re-called up to as many as five additional times. These calls are spread through different parts of the day and different days of the week. The Diary MailingThe list of successfully “enumerated” households becomes the sample for a specific survey. Although the Fall Sweep Survey reporting period is four weeks, each household is asked to provide just one wee’s information. To achieve the four-week coverage, the enumerated households are randomly apportioned over all four weeks of the survey. The DiaryEach “enumerated” household is mailed a package with a separate diary booklet for each television set. The diary has separate pages for each day of the week and is designed so that the viewing of each household member (and visitor) can be recorded separately. Respondents are requested to register their television viewing by recording the station call letters, the program name, and the channel number for each quarter-hour period of viewing during the day (6:00 a.m. until 2:00 a.m. the following morning). A separate section of the diary requests the respondent to provide background information (e.g. age, sex, language, education, occupation, ownership of VCR, etc.) for each of the household members. A small premium is enclosed with each diary together with a letter explaining the purpose of the survey. To encourage participation in the survey, a system of reminder phone calls is also used. Diary Editing and ProcessingInformation from returned diaries is first manually checked for complete information, then transferred to computer tape. These data are then subjected to a careful and extensive series of manual and computer checks. About 50% of diaries require further manual editing. If necessary, respondents are called to confirm information. Estimates of television viewing are then made for each of the defined areas of the country. In cases of simultaneous substitution on cable television systems, recorded tuning is attributed to the actual station signal transmitted over the cable regardless of which station’s call letters were indicated by the respondent in the diary. Simultaneous substitution may occur when two stations are broadcasting the same program at the same time. Under certain conditions, the cable company may substitute one signal for the other. This most commonly occurs when one station is Canadian and the other is American. In such a case, at the request of the Canadian station, the signal of the Canadian station would be transmitted on channel normally carrying the American station as well as on the channel carrying the Canadian station. To compensate for possible differences in response rates, the data are weighted by age/sex groups. In the Montréal and Ottawa-Hull areas the data are further weighted by “home language” group (i.e. English, French, other). Minimum Reporting StandardsBBM has established a number of reporting standards or minimum sample requirements. Neither BBM nor Statistics Canada will release data where the number of valid respondents for a specific area or demographic group falls below the reporting standard. Sampling error tables are not available. Program CharacteristicsThe file of program characteristics (i.e. program type, language, Canadian content, etc.) created by the CRTC is derived from several sources including administrative records and personal contact with television stations and program producers. Where possible, coding is compared with previous years’ files to maintain consistency. As a final check, data are manually verified by the various partners in the project. Station SchedulesThe file containing the schedules of the television stations is prepared for BBM by Mediastats Inc. using information provided by the stations by means of a mail-back document and telephone verification. Respondents Reporting No Television ViewingNon-viewers are included in the survey data as long as they meet all the normal editing standards. The average hous of television viewing presented in Table 1 are per-capita averages including both viewers and non-viewers. In Fall 1993, 95% of the population aged two years and over reported watching television at least once during the survey week. Change in MethodologySome important changes in survey methodology should be noted. Prior to the fall of 1984 each BBM survey diary mailed was “attached” to an individual. The television viewing recorded was specific to that individual without regard to where the viewing took place (e.g., at home or at the neighbors). For the Fall 1984 Sweep, the diary mailed was “attached” to the residence or the television set. Viewing patterns of all the residents were recorded but no provision was made to capture “out-of-home” viewing. For the Fall 1985 and subsequent surveys, adjustments were made to include the viewing of visitors to the home as a surrogate for the “out-of-home” viewing recorded under the pre-fall-1984 methodology. To maintain consistency in the average hours per week shown in Figure 1, the original value for 1984 has been inflated by a factor of 2.6%. This factor was derived by examining the percentage of visitor viewing records in the Fall 1985 survey. Data LimitationsData collected by the BBM survey represent the television viewing of all Canadian aged 2 years and older with the following exceptions:
Availability of Viewing DataThe Statistics Canada Television Viewing Data Bank contains information beginning with the fall of 1982. Data are maintained on magnetic storage media in the Main Computer Centre of Statistics Canada. Special tabulations are available on a cost-recovery basis and can be provided on micro-computer diskette. Some tabular data on viewing hours only are available for previous periods. |
|
|
