Mediamark Research 
Teenmark 2004 Technical Guide

The MRI TEEN Study

Contents:

SAMPLE DESIGN

SURVEY QUESTIONNAIRE

DATA COLLECTION AND QUESTIONNAIRE SCOPE 

INTERNET AND ON-LINE SERVICES 

RADIO LISTENING

TELEVISION VIEWING

CABLE TELEVISION

PSYCHOGRAPHICS

DEMOGRAPHIC AND RELATED DATA

PRODUCT DATA

DATA PROCESSING

PROJECTION

SAMPLING TOLERANCES

The 2004 MRI TEEN Study comprises data collected from a random sample of respondents, 12-19 years of age, residing in the 48 contiguous United States.

The source of the sample frame from the TEEN Study is MRI's Doublebase 2004 households. While conducting the latter study, which used a strict area probability sample, MRI collected sex and age information about every person residing in each household. From April 15, 2003, to July 15, 2003 (for waves 47-48) and April 19, 2004, to July 19, 2004 (for Waves 49-50) MRI attempted to survey every teenager in those households, excepting those who were 2004 Doublebase respondents.

Because of the time lag between the initial MRI interview for the adult syndicated study (the 2004 Doublebase), MRI considered the amount of time elapsed between the initial collection of household information and the execution of the TEENMARK study to determine respondent eligibility. For example, children who were 11 years of age at the time of the initial screening a year earlier had become eligible respondents, but 19 year olds in the initial screening were no longer eligible for the Teen study. In order to ensure that only teenagers were surveyed, the sex and age of the designated teen respondents were matched to the information originally collected. Any respondents under 12 or over 19 years of age were eliminated from the study.

Back to Table of Contents

A mailed questionnaire was used to collect all data for the study. The survey instrument contained questions pertaining to media exposure for magazines, radio, television and on-line services. Demographic and product usage data were also collected. Each mailed questionnaire "packet" contained a $10.00 incentive for completing and returning the survey. In addition to the incentive, MRI held a random drawing in which three teenage respondents each won a Dell Computer and monitor or a Toshiba portable DVD player.

A second questionnaire was sent to all teenagers who had not responded to the first mailing within 30 days. No additional incentive was included in this mailing. The completion rate for the study is shown below:

Back to Table of Contents

Ideally, the TEENMARK methodology would be identical to the methodology of the adult study. The Teen Study, however, was conducted entirely by mail, rather than by a combination of in-home, personal interviews and self-administered questionnaires; TEENMARK respondents were not, by design, the only persons surveyed in their households, while -- at the time -- the previously-surveyed adults were. Because teenagers represent insignificant proportions of the markets for many products and of the audiences of many media; e.g., alcoholic beverages and television news programs, product category and media questions were appropriately abbreviated. Finally, some product categories and media whose markets and audiences are primarily teenage oriented were included in TEENMARK research but not in the study of adult consumers.

The procedure employed in the teen study was based on two straightforward questions.  Each respondent was asked:

1. About how many issues of an average four do you read or look into?; and

2. Have you read or looked into any copy in the (last publication interval)? The publication interval      obviously conformed to the publication frequency of the listed magazines.

The list of magazines was rotated both alphabetically and by publication cycle so that four versions of the magazine questioning procedure were created.

The TEENMARK magazine readership methodology was different from that employed in the adult study.  The divergent approaches were dictated by the data collection methods used.

In the adult study respondents are asked by interviewers to sort a deck of approximately 240 cards bearing magazine logos into three categories:  read or looked into the past six months (yes - sure have), might have done so (not sure), and did not read or look into during the past six months (no - sure have not).  This "screen" or "filter" question reduces the number of magazines about which subsequent questions are asked.  Following the sorting procedure the surviving (read or might have read) magazines are once again presented, and respondents are asked how many issues of the average four are read or looked into (frequency) and are then asked whether each magazine was read or looked into during its most recent publication interval (recency) -- seven days for weekly magazines, 30 days for monthly magazines, and so on.

The TEENMARK research, because it was conducted by mail, did not use the screening procedure described above.  By omitting the screen in a mailed survey, MRI abided by the generally-held belief that such a procedure should be avoided.  Michael Brown stated this admonition best for readership surveys.  "One of these (readership surveys) involved a self-completion questionnaire and with this technique, an "invitation" to skip certain subsequent questions - as implied by the filter is usually avoided".

Instead, respondents were first asked to indicate the frequency of their readership of the average four issues, then to indicate whether they had read or looked into the magazine during its most recent publication interval.

Differences in readership results were expected and indeed were found in MRI's pretesting.  The explanations for these differences could be due to these factors:

1. During the personal interview in the adult study, recency questions are asked with reference to a specific previous date or day corresponding to the number of days earlier than the interview the publication interval occurred.  This was not possible with the mail survey, and may have lessened the accuracy with which respondents were able to answer. TEENMARK respondents were readily able, while answering the frequency question, to preview the following question set -- the recent reading questions -- and were also able, while responding to the recent reading questions, to refer back to their answers to the frequency question.  This would have allowed respondents to attempt to achieve perceived consistency between these question sets, if they were motivated to do so.  
   
2. In the adult study, after respondents "screen-in" by indicating readership of a magazine during the past six months, they are asked whether their readership of the average four issues is 0, 1, 2, 3, or 4.  In the Teen Study no screening question was asked and the zero frequency response cannot be distinguished from a non reader.

Since a major objective of the TEENMARK research is to provide readership data which may be combined with readership data from the adult study, two measures were employed to determine the extent of the differences in magazine readership, and to ensure results would be reported compatibly:

First, the inclusion of 18 and 19 year olds in the TEENMARK sample provided a direct means to compare results of the two studies, since respondents aged 18 and 19 are also included in the adult study.

Second, MRI has years of historical data on the relationship between the reading frequency and recency; the TEENMARK data were examined for consistency with these adult data relationships.

To make the TEENMARK magazine readership data compatible to the adult magazine readership data, the following procedure was undertaken:  the frequency and the recency questions were asked almost identically in both studies, yet, due to the methodological differences described above, produced different results; specifically the relationship between frequency and recency data was different for the teen and the adult studies. To render the TEENMARK readership data equivalent with adult data, six years of each magazine's historical frequency/recency relationship was calculated.  Within each frequency -- 1, 2, 3, or 4 of the average four issues -- the recency responses from the Teen Study were conformed to their historical relationships with frequency for each magazine.  The data were conformed separately for each of three age groups -- 12-14, 15-17, 18-19 -- to maintain response consistency within age cohorts.

The recency data for magazines not measured in the adult study were conformed to an averaged, historical frequency-recency relationship.

The resultant adjusted data correlate closely with the readership by 18 and 19 year olds of magazines also measured in the adult research.  This indicates that the restoration of the historical frequency-recency relationship overcomes to a great degree the differential results which arose from the employment of the separate data collection techniques.

Back to Table of Contents

The questionnaire contained five questions relating to the Internet and on-line service usage. The five questions released this year are worded as follows: · "In the last 30 days, have you accessed the Internet in any of the following places?" 

• "Which on-line service, if any, have you personally looked at in the last 30 days?" 
• "On average, how often do you access the Internet in a typical month?" 
• "Which, if any, of the following activities did you do on the Internet in the last 30 days?" 
• "Which, if any, of the following web sites did you visit in the last 30 days?"

Back to Table of Contents

The radio listening question for weekday listening asks: "Please write in approximately how much time you usually spend listening to or hearing a radio on an average weekday, Monday through Friday (for each of the time periods shown below). Include all listening, whether in your home, car or any other place."

For each time period mentioned, the respondent was asked to write in the call letters of the stations and to indicate whether the station was an AM or an FM station.

The same questioning sequence was applied to Saturday and Sunday listening separately, with the respondent answering for the "average Saturday" or "average Sunday."

Back to Table of Contents

The questionnaire contained two sets of questions for television viewing. The first set asked about usual viewing on an "average weekday" and on an average Saturday and an average Sunday. The question was worded as follows:

"Please write in approximately how much time you usually spend watching television on an average weekday, Monday through Friday, for each of the time periods shown."

Time periods varied by time zone in which the teenager resided. Similar questions were asked for Saturday and for Sunday viewing, but the time periods were fewer than for the average weekday.

In addition to these questions, the survey measured teen viewing of specific programs. The recency and frequency of viewing, along with attention levels and place of viewing, were asked for daily (Monday through Friday) programs, weekly programs and weekend programs.

Back to Table of Contents

The question on cable television viewing was:

"About how many hours have you watched each of the following cable services in the past 7 days?"

In order to keep the data consistent with the adult study the teen responses were cleaned based on the earlier collected information on household subscription to cable television or ownership of a satellite dish.

Back to Table of Contents

H. PSYCHOGRAPHICS

The questionnaire contained nine categories relating to the teenager's interests and attitudes. The topics that were released this year are as follows:

• Internet/On-line Technology
• Finance
• Music
• Ads
• Media
• Yourself
• Beauty: Hair
• Beauty: Make-up
• Fashion

Back to Table of Contents

A limited number of demographic questions were asked of the teenagers. In addition to these, a number of responses by the adult in the same household regarding household data were carried over to the teen survey.

The collection of information on product use and consumption followed the demographic and media sections of the questionnaire. To be consistent with the adult study, questions common to the adult and teen study were asked in exactly the same format, including timeframe. Additional questions, particularly relevant to the teen market, were added to the product sections. As a result, not all of the product data can be merged with the adult study.

Back to Table of Contents

All data processing procedures are consistent with those employed by MRI for the adult study with one notable exception: because all data are collected by a single questionnaire, there is no data ascription. For a detailed description of other data processing procedures, please refer to the MRI Technical Guide for the Adult Study.

Back to Table of Contents

PROJECTION

The TEENMARK study is weighted and sample balanced to conform to estimates provided by the U.S. Bureau of the Census and Market Statistics.

The first stage, weighting, accounts for the differential probabilities of household selection for the study. The factors affecting these probability rates and the applied weights are discussed in detail in the Technical Guide for the Adult Study.

After weighting, the sample is balanced within sex on the following population parameters:

1. Top Ten Markets
2. Age
3. Race
4. Metropolitan/Non-Metropolitan Area within Census Region
5. Household Income
6. Education
7. County Size

The sample balancing procedure is the widely used technique first discussed by W. Edwards Deming in his book "Statistical Adjustment of Data". After sample balancing, the distribution of weights is inspected and the "outlier" weights are trimmed to reduce the sampling tolerance levels. The tables below compare the before and after balancing estimates.

Back to Table of Contents

All sample surveys are characterized by sampling tolerances. Sampling tolerance is the difference that can be expected between the results of a sample survey and the results of a full survey or census, using the same procedures and techniques. This is the difference due to the chance selection of one group of respondents or another. For a more extensive discussion of the differences in sampling tolerance levels, please refer to the Technical Guide for the Adult Study.

The calculated sampling tolerances for the media audiences are tabulated at the beginning of the audience volume. These are computed using a set of eight replicated subsamples of the total sample. The difference among the eight subsamples, which are chance differences, are used to estimate the sampling tolerances of the total sample.

The sample tolerances should be used to evaluate the precision of an estimate and the degree of confidence that can be placed in it.

Back to Table of Contents

For more information regarding Mediamark Research, contact MRI at info@mediamark.nopworld.com.
Mediamark Research is a NOP World company, a wholly-owned subsidiary of United Business Media.
Copyright  2005,  Mediamark Research