# GLOSSARY

To assist all those working in the market research and opinion industry ESOMAR maintains a comprehensive glossary of market research and marketing terms.

A |

B |

C |

D |

E |

F |

G |

H |

I |

J |

K |

L |

M |

N |

O |

P |

Q |

R |

S |

T |

U |

V |

W |

X |

Y |

Z
A means of segmenting a population of interest using life-cycle stage, income and occupation variables.

A general term that can be used in two similar ways:

- The importance a brand holds for different groups of consumers. It is an indication of emotional closeness to, or distance from a brand.
- The extent to which a brand comes readily to mind (eg first mentioned in an answer to an awareness question), or the most frequently mentioned in connection with a set of associations.

A sub-group of the population of interest that has been selected for study.

The efficiency and the effectiveness of the sampling approach at reaching the units specified in the research brief

A description of the ideal (or the intended) sample for a research project.

Sample Selection Error occurs in experiments when a bias is introduced into the way in which experimental units are assigned to groups.

The number of sample units to be included in the sample.

An individual member of the sample.

The error in a survey caused by using a sample to estimate the value of a parameter in the population.

A list of the population of interest that is used to draw the sample in a survey, eg a telephone directory or a list of members of a profession.

A type of non-sampling error in a survey caused by a sampling frame containing either more or less of a particular type of potential respondent, compared with the population of interest.

A set of instructions given to interviewers concerning the selection of participants for a research project. These instructions must be followed exactly to ensure accurate respondent selection.

The size of the step between selected units in systematic sampling.

When sample units can be included more than once in the sample.

When sample units cannot be included more than once in the sample.

A survey where as many members as possible of a population of interest are contacted.

The extent to which a scale is valid and reliable, ie free from bias and random error.

The degree to which a scale is free from random error and can reproduce consistent results.

The manipulation of scale variables to ensure comparability with other scales and enable comparisons to be made. The most frequently used scale transformation procedure is standardisation.

Scale Validity is whether or not a scale measures what it is intended to measure. A scale may be biased by the way in which a question is worded and therefore the results it produces are not valid.

Scales are measurement devices that allow respondents to report the degree of their opinions. Scales are usually in the form of statements or numbers. Pictures may also be used - see

Pictorial Scale.
The data recorded by bar-code scanners.

A type of consumer panel where participants use a bar-code scanner to record purchases.

A plot of the values of two variables for all the cases or observations.

A numerical value assigned to a response or an observation.

The procedure of asking specific questions to determine whether respondents are eligible to participate in a particular research study. This is done at the very beginning of an interview.

The questions at the beginning of an interview or questionnaire to ensure that a potential respondent is eligible for the survey.

Data that have already been collected and published for another research project (other than the one at hand). There are two types of secondary data: internal and external secondary data.

The process of dividing markets into groups of consumers who are similar to each other, but different to the consumers in other groups.

A type of non-sampling error where the sample units are selected for treatment in a particular way that produces a different profile to the population. Selection bias can be introduced by researchers and/or by respondents (putting themselves into groups to which they aspire to belong, but they do not currently belong).

Self administered is where a respondent completes a survey without the assistance or involvement of a researcher, eg a self-completion questionnaire sent through the mail.

A situation where the true answer to a question does not reflect favourably on the respondent or it contravenes a social group norm. In order to avoid bias in the data, one should try to avoid putting respondents in self-effacing situations if possible.

A type of non-sampling error that occurs when respondents who chose to participate in some research are systematically different to the intended sample. This type of bias is caused by some types of respondent replying to a survey invitation more than others.

A type of categorical, non-comparative scale with two opposing adjectives separated by a sequence of unlabelled categories, eg Good 1 2 3 4 5 6 7 Poor.

An interview or questionnaire is semi-structured when many or all of the questions are open-ended, allowing the interviewer to clarify and probe the answers given by respondents.

The theory of signs and symbols in language and the meanings they convey. In research it is used to identify and evaluate the true meaning behind consumers' linguistic responses, to decode their cultural frames of reference and behaviours. It employs specialist techniques to overcome the problems of conditioned or expected responses (Social Group Norms) and provides a deeper understanding of consumers' motivations.

The extent to which a research technique can identify differences between sub-groups of participants.

A type of focus group where the same participants are convened several times over a period of time. The subsequent discussions may or may not be on the same subject.

An examination of some or all aspects of products that are perceived by the five senses.

A projective technique where participants are asked to complete a number of incomplete sentences that are provided by the researcher.

Sequential Monadic Evaluation is where two or more products or ideas are each evaluated on their own, one after the other. The sequence is usually rotated to minimise order bias (ie respondents favouring an object or idea because of its position in a list).

Sequential Sampling is where the sample units are drawn one by one or in groups and the results of the drawing at any stage decide whether sampling is to continue.

The extent to which a particular brand will be thought of in relation to a specific product category.

The percentage of all promotion in a specific product category that is accounted for by a particular brand.

A type of prompt material in the form of cards with images that are shown to participants in research studies.

The maximum probability of rejecting a true null hypothesis (or committing a type I error), equal to one minus the confidence level.

An analysis of sample data to determine whether the data supports a hypothesis about the population from which it was drawn.

A type of probability sample where all units in a population of interest have an equal, known and non-zero chance of being selected.

A type of laboratory experiment that aims to imitate real life, where respondents are selected, interviewed and then observed making or discussing their purchases. Simulated test markets can lead to mathematical models used to forecast factors such as awareness, trial, sales volumes, impact on other products etc.

A model composed of mathematical and logical relationships designed to represent an actual system and indicate how the real system would react in various circumstances.

A type of research design where one sample is drawn from the population of interest only once.

A measurement scale that gathers opinions about an object on a single dimension.

A question that asks respondents for only one answer.

Data on a variety of behavioural habits from consumers in the same households, such as TV watching, reading and shopping habits.

Part of the marketing planning process that deals with understanding the environment and the market (by identifying opportunities and threats) and assessing a firm's competitive position (by identifying its strengths and weaknesses).

Refers to the symmetry of a distribution. A skewed distribution is one where the mean, the median and the mode have different values, whereas in a symmetrical distribution they all have the same value.

A question used to guide an interviewer through a survey to different questions (ie skipping some questions), depending on the answers given.

A type of pictorial scale that consists of a balanced series of facial expressions that are often used in children’s research to obtain satisfaction ratings.

The process of removing fluctuations in an ordered series of data.

A type of non-probability sampling where initial respondents are selected at random and subsequent respondents are then selected by referrals or information from the earlier respondents.

The tendency for respondents to give answers that are socially desirable or acceptable, that may not be accurate.

Expected overt modes of behaviour that constitute culturally acceptable ways of behaving in specific situations. Participants in market research projects may well bias their responses in order to conform to expected behaviour.

Statistics that describe social rather than economic variables (eg birth rate, life expectancy at birth and doctors per thousand of the population).

A method of dividing a population of interest into groups usually based on income and occupation of the head of household, although other variables can also be used. The ESOMAR social grades (A, B, C1, C2, D, E1, E2 & E3) are based on the terminal education age and occupation of the main income earner.

Information that is based on the occupation and income level of the head of household.

A type of true experimental design where test units are randomly allocated to two experimental groups and two control groups. One of the experimental groups and one of the control groups is measured. Both experimental groups are then exposed to a treatment. Afterwards both experimental and control groups are measured. Six measurements are taken in all and the design aims to account for pre-testing bias and pre-test manipulation interaction bias.

A procedure where a sample is divided into two halves and each receives a slightly different questionnaire.

A technique in advertising research that involves placing an advertisement in one form in half of the copies of a given publication and in another form in the other half. The purpose is to compare the relative effectiveness of the two forms of advertisement.

A measure of how many respondents can quote a brand name without any assistance on behalf of the interviewer.

When two variables appear to change together but there is no causal relationship between them.

A measure of variability (or dispersion) of a distribution and it is equal to the square root of the variance.

A measure of the amount of sampling error present and it is equal to the standard deviation.

A means of classifying businesses based on the products or services they provide.

A scale transformation procedure that involves manipulating data from different types of scales so that they can then be compared. It consists of subtracting the sample mean from each score and dividing by the standard deviation.

A type of categorical, non-comparative scale that is similar to a semantic differential scale except that instead of having two opposing adjectives, there is only one adjective in between a sequence of positive and negative categories. Respondents indicate how relevant each single characteristic is to an object in question.

A type of pre-experimental design where there are two groups; an experimental group and a control group. The test units are not assigned randomly between both groups. The experimental group is exposed to a treatment and afterwards, both groups are measured.

A type of experimental design that allows for the statistical analysis and control of external variables. Examples of statistical design are: randomised blocks, Latin square and factorial design.

Involves a comparison of the sampling errors generated by different sampling procedures. A statistically efficient sampling procedure is one that produces fewer sampling errors for the same sample size than other procedures.

Refers to whether some research results genuinely reflect a population of interest in some way or whether the results could occur by chance. Statistical significance is determined by comparing the research results with the values defined by the confidence interval.

A process where there is an element of random variation.

A set of boards that illustrate the main points of a proposed TV commercial, details of dialogue and sound effects are also included.

A projective technique where participants are asked to provide the conclusion to a number of incomplete stories provided by the researcher.

A type of probability sample where the units in a population of interest are divided into mutually exclusive and collectively exhaustive strata and a (proportionate or disproportionate) random sample is drawn from each stratum.

A set of line charts where the data are successively aggregated over the series. The magnitudes of each variable are represented by the areas between the line charts.

A form of observation study where the behaviour of interest and the method of observation are clearly specified in advance of the study.

Questions where the choice alternatives for respondents are all specified.

A row heading in banner format tabulations. It is usual practice to have the dependent variables as stubs.

A sample of a sample, which may or may not be selected using the same approach as the original sample.

Sales under the guise of research and it is one of the reasons why potential participants in market research projects are reluctant to take part.

A type of non-sampling error where someone other that the intended respondent participates in the research and this produces a systematic difference between the resulting sample and the intended sample.

Involve a (statistically) large number of interviews with respondents, using pre-designed questionnaires.

Where the findings and costs of a research project are shared (partially or fully) among a number of clients.

A type of brain storming where the participants are from a variety of different backgrounds are asked to solve a specific problem. The approach is intended to increase the creativity of ideas produced.

An error that affects measurements in a consistent way.

A type of probability sample where every nth unit is included in the sample from a list of the population of interest. The value of n is calculated by dividing the number of units in the population of interest by the required sample size.

Facebook Twitter