Market Research Directory - Glossary of Terms

American Association for Public Opinion Research.

Attitude, Awareness and Usage study. A common type of study used to understand how several brands are positioned in a particular product or service category. Competitive positioning can differ in terms of how often consumers of a product or service are aware of each brand, the extent to which they feel various attributes apply to each brand (i.e., attitudes), and usage of each brand (e.g., ever used, used recently, use most often, etc.).

American Marketing Association.

Analysis of Variance. A statistical term that can be used in many different contexts but often refers to testing for statistically significant differences between means for multiple subgroups. (See a statistical textbook for a formal, technical definition.)

A data file in ASCII format.

Involves conducting multiple "waves" of research to help understand changes over time in advertising awareness and various measures that may be impacted by advertising (e.g., brand awareness, brand attitudes, purchase interest, etc.). A common approach is to conduct a "pre-wave" before a new advertising campaign and then conduct one or more "post-waves" after the campaign has been in place for some time.

Asking survey respondents to indicate whether or not they are aware of specific, pre-identified items (e.g., brands, companies, images, message statements, etc.). In contrast, "unaided awareness" is based on asking respondents to name brands, companies, messages, or other items they can think of on their own (i.e., in an open-ended manner, without being "aided" by a list that they can select from).

Survey research where respondent identities are not recorded or tracked during the data collection process.


B

Business Reply Envelope. An envelope included in a survey mailing that respondents can use to return a completed questionnaire. The BRE has the correct return address already printed on it, and the survey respondent is not required to affix postage.

Business Reply Mail. Requires account with United Stated Postal Service and approved artwork for use with business reply envelopes.

The number of survey invitations sent is controlled by demographic subgroups and/or geographic regions to be in proper proportion relative to U.S. Census data (for a general population study) or relative to available data for a target population.

Data tabulations with separate columns showing results for selected subgroups. Significance testing is frequently included to highlight statistically significant differences between subgroups. Also sometimes referred to as "Banners" or "Tabs" or "Computer Tabulations".

A column heading of a banner tabulation representing a particular sub-group.

A type of graph where the data is displayed in the form of horizontal or vertical bars.

Creation of an internal or external benchmark comparison for future survey administrations. Can sometimes refer to results of a first "wave" of research, and comparisons are made to results from later "waves" of research.

Term can be used interchangeably with Baseline (see above).

A question with wording which influences many respondents to answer in a particular manner that may not be intended or may not be a valid reflection of their opinions.

A sample that is not representative of the survey population.

Testing of products or concepts where brand names and other identifying information are not revealed to study participants.

To the extent that brand perceptions and brand associations influence purchase interest and behavior, the brand itself can be thought of as having financial value.

A segment, subgroup, quota or sub-quota.

Discount postage rate for presorted standard mail.


C

Coding Accuracy Support System. Address-matching software that qualifies mailings for postage discounts with the U.S. Postal Service.

Computer Aided Telephone Interviewing. A CATI system is used by most telephone survey companies.

Change Of Address.

Cost Per Complete.

Cost Per Interview.

Comma Separated Values. Refers to a file format in which different data fields are separated by commas.

See Bucket above.

A study in which the entire target population is invited or included rather than drawing a sample.

The number of survey respondents from various demographic subgroups is controlled to be in proper proportion relative to U.S. Census data.

Type of survey question that includes a list of specific, pre-identified response options that a respondent can select from (e.g., possible answers are shown on the screen in a web survey or read during a telephone survey).

Number of completed surveys.

Length of time it takes a survey respondent to complete a survey.

Testing reactions to a description, image, audio clip, or video of a proposed product or service.

A statistical term that is typically used to refer to the interval created by adding and subtracting a margin of error from a survey result (e.g., from a percentage or mean). (See a statistical textbook for a formal, technical definition.)

A statistical term that would require a detailed technical definition to be precise but can be thought of roughly and intuitively as the opposite of the risk of error. For example, if the risk of error is 5%, the confidence level is 95%. (See a statistical textbook for a formal, technical definition.) Confidence intervals and statistical tests are commonly applied at the 95% or 90% confidence level.

Respondent identities are held in strict confidence and not disclosed to others. Answers to the survey are not linked to specific individuals, but are analyzed in aggregate only.

Technique used by market research firms to help understand how consumers make trade-offs between various features, benefits, pricing, and other attributes that can comprise a product or service.

Analysis of the degree to which changes in one variable are associated with changes in another. (See a statistical textbook for a formal, technical definition.)


D

Designated Marketing Area. A geographic area defined by ACNielsen Company.

Adjustment or removal of responses that are known to be incorrect or inconsistent or removal of respondents exhibiting such responses or otherwise exhibiting clearly unreliable response patterns.

Process of removing duplicate records when the same individual (or other type of unit that is supposed to be unique) appears more than once in a database or sample.

Gender, age, marital status, ethnicity, educational attainment, employment status, household income, and other personal characteristics.

Respondents are asked to make a series of choices when presented with sets of product or service profiles that are configurations of various attributes (e.g., features, benefits, pricing, etc.). The configurations are typically varied in an experimentally controlled manner to allow statistical estimation of the "utility" (or relative value) of each attribute (or each attribute level when attributes are defined at multiple levels).

The result of an attempt to reach a potential respondent. This term is typically used in the context of telephone interviewing.


E

Qualification requirements for survey participation. Often referred to as "screening" criteria.

The heritage, nationality group, lineage, or country of birth of a person or the person 's parents or ancestors before their arrival in the United States. People who identify their origin as Spanish, Hispanic, or Latino may be of any race.

Comparison of survey results with baseline data derived from external sources such as nationwide or industry group benchmarks.


F

A multivariate statistical technique that can be thought of roughly and intuitively as helping to group inter-related attributes (or other types of measures) into distinct groupings that reflect major themes or "dimensions" in the data. Can be helpful for concisely summarizing relationships among a large number of attributes (or other types of survey measures). (See a statistical textbook for a formal, technical definition.)

Periodic review of all or a portion of the survey data during the data collection process. For telephone interviewing, monitoring early interviews can help to assess how well respondents are reacting to and interpreting question wording.

Period from start date to end date of data collection process.

A count (or percentage) of the number of times a response option for a question has been selected.


G

A technique used to identify differences or "gaps" between perceived performance vs. desired or "ideal" performance.

Post baby-boomers born between 1964 and the late 1970s or early 1980s. (Individuals in the "Baby Boomer" generation were born between 1946 and 1964.)

Also known as Millennials. Persons born after Generation X.

Open-ended responses that appear to be deliberately entered in an unintelligible manner.

Worldwide directory of marketing research firms and services. A wholly owned subsidiary of the New York American Marketing Association, publishes the GreenBook directories.


H

Primary decision-maker for a household (or only member of a household) or person who manages the affairs of the family living in a household.

Total gross income of all members of a household combined.


I

In Home Use Test. A product sample is provided for participants to use at home prior to survey administration.

Interactive Marketing Research Organization.

Percentage of those who respond to a survey invitation who meet the qualifications for participating in the full survey.

A statistical test that can be used in different contexts but often refers to testing for statistically significant differences between means among two subgroups. (See a statistical textbook for a formal, technical definition.)


L

A psychometric scale allowing a survey respondent to quantify his or her level of agreement, interest or other measure regarding various statements.

Sample comprised of telephone numbers that are listed (i.e., published / publicly available).

Study that involves taking measurements from the same respondents at different points in time.


M

Metropolitan Statistical Area.

The margin of sampling error can be thought of roughly as a gauge of how much survey results based on a sample could potentially differ from what would be obtained if every possible person who could qualify for the study were surveyed. (See a statistical textbook for a formal, technical definition.)

A firm that engages in the design of surveys, data collection, and analysis of survey data. Amplitude Research is an example of a Market Research Firm.

A statistical term that is often used interchangeably with "average." (See a statistical textbook for a formal, technical definition.)

A value such that approximately half of the respondents gave a lower (or equivalent) response and half gave a higher response. (See a statistical textbook for a formal, technical definition.)


N

Number of completes for a study.

National Change Of Address database managed by the United States Postal Service.

Survey question that tests for respondent attention by re-asking a prior question with the wording changed so that a negative (or low) response would be consistent with a prior positive (or high) response.

Take the percentage of respondents highly likely (e.g., giving a 9 or 10 rating on a ten-point likelihood scale) to recommend a brand, product, or service and then subtract the percentage not very likely to recommend (e.g., giving a rating of 1 through 6 on a ten-point scale).

When individuals are invited to participate in a survey but do not respond, their opinions are not reflected in the final data collected, and "bias" occurs if their absence significantly skews or otherwise alters the final results.


O

Client sharing of survey sample and respondent eligibility in a single survey instrument.

A survey question that does not have a set of pre-identified response choices and allows the respondent to enter verbatim text (in a web or regular mail survey) or provide an oral description (in a telephone survey) as the response.

When responses to a question or the reactions to statements, concepts, or other items are influenced by the order in which they are present to the survey respondents. For example, a statement may sometimes be evaluated differently if it is present first or last.

Extreme values that can be defined in different ways but often as more than three times the standard deviation plus or minus the mean (or average) response. (See a statistical textbook for a formal, technical definition.)


P

Testing of questionnaire with a small group of survey respondents prior to full launch.

Automated dialing of telephone numbers to reach survey participants for telephone interviews.

Research to collect new data that is not available (or accessible) from already existing sources.

Measurements related to lifestyles, attitudes, beliefs, and other measures that reflect a person's psychological state rather than physical, demographic, or geographic characteristics.


Q

Market research utilizing representative data collection techniques to collect feedback in a uniform, structured manner from a substantial sample of the target population. In contrast, "Qualitative Research" involves less structured, exploratory questioning and probing during small group discussions (e.g., "Focus Groups") or in-depth "one-on-one" interviews.

A target number of survey respondents desired who often must meet specific qualifications. Often the plural form "Quotas" can be used similarly to the term "Sub-quotas."

A computer software process that is used to automatically shut down a cell (or subgroup or quota or sub-quota) of an online survey once a pre-determined number of completions has been obtained.


R

Random Digit Dialing. A sample of telephone numbers derived from randomly generating telephone numbers from particular telephone exchanges..

African American, Asian, White / Caucasian, or other category used by the U.S. Census Bureau to define "race." Although "race" is not technically the same as "ethnicity," the terms are sometimes used interchangeably in common practice. However, it is now common to consider Hispanic to be an ethnicity and not a race.

Telephone interviewing involving the use of randomly selected listed sample.

Using a process that selects individuals (or other units) in a random manner from a target population.

A statistic computed by subtracting the minimum value given from the maximum value given.

Often used to refer to the original survey response data before or without any cleaning or inspection for accuracy or consistency. Sometimes used to refer to untabulated respondent-level data files in Excel, SPSS, CSV, or ASCII format.

File of contact information for potential survey participants.

Percentage of invited survey participants who decline to take part in a research study.

A sample that properly reflects the mix of characteristics in the target population. Ideally achieved by true random sampling, or (when true random sampling is not feasible) by a sampling procedure that includes all types of individuals from the target population in proper proportion.

Reduction in accuracy and reliability of response data as survey length becomes excessive.


S

The number of survey completes for a research study.

The margin of sampling error can be thought of roughly as a gauge of how much survey results based on a sample could potentially differ from what would be obtained if every possible person who could qualify for the study were surveyed. (See a statistical textbook for a formal, technical definition.)

Survey questions used to determine who is and who is not qualified to complete a full survey.

Obtaining and analyzing publicly available or otherwise accessible data that already exists.

Research designed to uncover subgroups or "segments" of a target population that are distinct in important ways (i.e., that have implications for marketing and/or advertising planning). Segmentation research can range from the very complex (e.g., applying advanced multivariate statistical techniques to a large number of attitudinal and behavioral measures) to fairly simple applications (e.g., identifying simple subgroups that differ significantly on an overall measure of purchase interest).

Applying statistical tests to identify when differences are large enough to be statistically significant. "Statistically significant difference" would require a detailed technical definition to be precise but can be thought of roughly and intuitively as sufficient to be highly confident that a true difference exists in the target population, even though the results being tested are based on a sample, and are therefore subject to "sampling error."

Can be thought of roughly and intuitively as sufficient to be highly confident that a true difference exists in the target population, even though the results being tested are based on a sample and are therefore subject to "sampling error."

Sending invitations to a partial sample of potential survey participants at the start of data collection often to test incidence, the response rate, or other reactions to the survey before proceeding with a full launch.

Data file in a format that can be used with SPSS computer software.

A statistical measure of dispersion (i.e., the extent to which responses / values differ from respondent to respondent). (See a statistical textbook for a formal, technical definition.)

When a respondent does not appear to be paying attention to actual question wording and continues selecting the same response option for numerous questions.

Subdivision or subgroup of a "Quota" that may be assigned a target number of desired respondents.

Length of a telephone interview or web survey in minutes, or number of pages in a regular mail survey.

Procedures followed to conduct a study.

The entire population of consumers (or other individuals) under study. A survey is typically conducted among a sample drawn from the study (or target) population. However, if the target population is small, all members of the target population may be invited to participate in a survey.

A person invited to participate in a survey who does in fact participate by answering a minimum number of questions.


T

Basic summary of response data typically displaying frequencies and other statistics for the total sample (i.e., not including cross-tabulation of results by subgroups).

Involves conducting multiple "waves" of research to help understand changes over time. (See Advertising Tracking Research earlier as one example.)


U

United States Postal Service.

Measuring awareness based on asking respondents to name (or write in) brands, companies, messages, or other items they can think of on their own (i.e., without being "aided" by a list to select from). A common approach is first to measure "Top of Mind" unaided awareness, which focuses on the first brand / company / message that comes to a respondent's' mind, and then to measure "Other unaided awareness," which refers to other brands / companies / messages that a respondent can think of after their "Top of Mind" response.


V

Responses to open-ended questions that are recorded "as is" without editing or coding.

Assignment of codes to verbatim responses based on common themes.


W

When the actual number of completions for particular subgroups differs (either intentionally for various reasons or unintentionally) from what would be expected from a true random sample of the target population, the survey results can be adjusted by "weighting" the data. Essentially, "weighting" assigns more "weight" or influence to the types of respondents who are under-represented in the survey and less "weight" or influence to the types of respondents who are over-represented in the survey.


Z

A statistical test that can be used in different contexts but often refers to testing for statistically significant differences between percentages among two subgroups. (See a statistical textbook for a formal, technical definition.)