Ad Hoc Research – A one-time research project designed for a particular purpose (as opposed to being conducted regularly or as part of a larger research program).
Adaptive Conjoint – An interactive conjoint design, ideal for situations in which the number of attributes exceeds what can reasonably be done with more traditional conjoint methods. Adaptive conjoint gives priority to attributes that are most relevant to the respondent and avoids respondent fatigue by focusing on fewer attributes at a time.
Aided Recall– A questioning approach that attempts to stimulate a respondent’s memory with clues about an object of interest.
Analysis of Variance (ANOVA)– A method of testing metric variables against a single dependent categorical measure to determine whether or not the means differ across all groups.
Applied Research – Research designed specifically to solve a particular problem or objective (as opposed to theoretical or exploratory research).
Attitudes – Mental states used by individuals to structure the way they perceive their environment and to guide the way which they respond. A psychological construct comprised of cognitive, affective, and intention components.
Attitude, Awareness & Usage Study (AAU) – A longitudinal (tracking) study that measures the attitudes, awareness, and usage levels for a product or brand.
Attitudinal Scaling – Questioning technique where respondents rate an object (brand, product, service, etc.) on a pre-defined scale, such as “extremely valuable”, “somewhat valuable” and “not valuable.”
Attribute – A characteristic or property of an object or person.
Augment – To increase the number of respondents included in a segment or sub-segment beyond what a random sample would provide.
Awareness – The proportion of people who are familiar with a product, brand name, or trademark.
Balanced Scales – A scale used in questionnaires where the number of positive and negative categories is equal.
Banner Point – The heading for a single column of data in cross-tabulations.
Base – The sample size or number of respondents used to compute percentages in a data table.
Bayesian Statistics – Statistics that incorporate prior knowledge and accumulated experience into probability calculations. Bayesian analysis can be applied to a number of traditional analyses such as regression and the estimation of conjoint utilities.
Behavior – The past and present overt responses of subjects.
Benchmark – A control source against which you compare the area you’re studying. For example, you may compare the results of a study in one state to the results of the nation as a whole.
Bias – A systematic tendency of a sample to misrepresent the population. Biases may be caused by inadequate sampling of the population, survey or item non-response, interviewing techniques, wording of questions, data entry, etc.
Bipolar Scale – A rating scale that uses symmetrical and opposing end points (e.g. agree/disagree, good/bad) to capture attitudes or evaluations. Bipolar scales typically invoke at least 5 scale points, and use escalating modifiers (e.g. somewhat, very) to describe values as they deviate further from a midpoint. The midpoint value represents a point of indifference or equality between the opposing end points.
Blind Study – A study where all brand identity is removed or hidden, most frequently to avoid bias.
Brand – A specific name, symbol or design which is used to distinguish a product or service from competitors.
Brand Associations – Attributes, values, or other meanings (images, emotions, etc.) that are attached or associated with a brand.
Brand Equity – The concept wherein the brand is considered an asset insofar as it can be sold or bought for a price. A powerful brand is said to have high brand equity.
Brand Extension – A branding or marketing strategy where a new product or service is launched under the existing brand name of a well-developed product or service.
Bulletin Board Groups – Online qualitative research that utilizes bulletin boards or message boards to have discussions, read and comment on postings, share files, etc.
Business to Business (B2B) – Describes the market where a business is selling it’s product or service to another business (as opposed to consumers).
Business to Consumer (B2C) – Describes the market where a business is selling its products direct to consumers.
Buying Behavior – The process purchasers go through when deciding which products or services to buy.
Buying Intent – The likelihood that a respondent will purchase a product or service.
CART/CHAID Analyses – Two tree-based methods for segmenting respondents that maximize differences with respect to an outcome (dependent) variable. Tree-based methods evaluate each potential predictor and divide the sample by the characteristic that maximizes differences across respondents. Splitting stops when certain statistical criteria are met, leaving the segments defined in terms of the nodes created from each split. These methods are capable tools for sifting through a large amount of information to identify powerful classification variables, and are thus used often in data mining efforts.
Case Study – A comprehensive description and analysis of a single situation.
Causal Relationship< – A precondition influencing a variable of interest, or, in more strictly, a change in one variable that produces a change in another variable.
Census – A survey of an entire population or universe.
Chi-Square Statistic – There are a variety of applications and definitions of the chi-square statistic, but the most common usage is to determine the degree of independence between two nominally scaled variables.
Closed-end Question – Questions that require a respondent to select from a pre-defined set of responses.
Cluster – In a sampling context, a cluster is defined as a category assigned to a neighborhood based on the assumption that the households share certain demographic, social, and economic characteristics. In a segmentation context, a cluster is a homogenous group of respondents who are defined through cluster analysis.
Cluster Analysis– A technique for segmenting respondents without using a predictor (dependent) variable. It identifies segments using a variety of data, including attitudinal, usage, or preference inputs. Cluster analysis uses one of algorithms to group people who are maximally similar to one another and maximally different from other groups. Cluster analysis is best viewed as an exploratory technique, since it is impossible to determine the “right” number of segments for any given market.
Coding – The process of assigning numeric values to responses from open-ended or other-specify questions.
Comparative Scales – A scale that requires one object (brand, product, service, etc.) to be compared to other objects.
Completes – Completed research interviews.
Completion Rate – The percentage of qualified respondents who complete a survey in its entirety.
Computer-Aided Self-administered Interviewing (CASI) – A computer-based interview that respondents complete themselves, usually at a pre-determined location.
Computer-Aided Telephone Interviewing (CATI) – Interviews conducted over the telephone, where the interviewer is using a computer-based program.
Computer-Aided Web Interviewing (CAWI) – Interviews conducted over the internet.
Concept Description – A brief description of a new product or service.
Concept Test – A test of a product concept where the concept is evaluated by a sample of the target segment.
Conceptual Mapping – A moderation technique in which participants are asked to place the names of products or services on a grid. How they group the items on the diagram is used to stimulate discussion.
Confidence Intervals – The range around a survey result for which there is a defined statistical probability that it contains the true population parameter.
Confidence Level – The probability that a particular confidence interval will include the true population value.
Conjoint Analysis – A broad collection of techniques that use experimental designs to derive the relative worth or value of a product based on respondents’ willingness to trade-off each component of the offering. This term typically refers to methods that use rank-ordering or rating scales to evaluate preference of attribute combinations.
Consideration Set – All the alternatives that potential buyers would consider in their next purchase of the product or service.
Convenience Sample – A non-random sample of respondents, selected based off of convenience.
Copy Testing – The process of using market research to evaluate the effectiveness of advertising, typically against metrics such as memorability, appeal and impact.
Correlation – A number between +1 and 1 that reflects the degree to which two variables have a linear relationship.
Cross-Tabulation – Breaking out survey results by different groups of respondents (by age, gender, etc.).
Custom Marketing Research – Customized marketing research to address specific projects for corporate clients.
Customer Value Analysis (CVA) – Analysis of customer satisfaction within an organization as compared with a competing organization’s customer satisfaction. CVA is used to determine profitable customer relationships, areas where the organization and competing organizations succeed /fail, the profitability per customer, the overall customer value of a firm and customer value for its competition, the value and cost of potential customers, and the cost of retaining customers. Gives a detailed picture of the organization’s strengths and weaknesses, as well as customer expectations.
Data Collection – The process of collecting research data.
Data Mining – Using statistical and advanced software to query large sets of data.
Data Processing – Organization of data for the purpose of producing desired information; involves recording, classifying, sorting, summarizing, calculating, disseminating, and storing data.
Data Visualization – The techniques used to communicate data or information by encoding it as visual objects contained in graphics. The goal is to communicate information clearly and efficiently to users.
Database – An organized store of data, usually within a computer.
Deduping – Identifying and removing duplicate records in a data file.
Demographics – Description of the vital statistics or objective and quantifiable characteristics of an audience or population. Demographic designators include age, marital status, income, family size, occupation, and personal or household characteristics, such as age, sex, income, or educational level.
Dependent Variable – A symbol or concept expected to be explained or caused by the independent variable. It is the variable measured on each subject to determine whether its value is affected by the independent variable. Also known as criterion variable.
Depth Interview – A discussion between the moderator and interviewee. This technique avoids groupthink and enables deep probing of the subject matter.
Discrete Choice Analysis – A variation of conjoint analysis that uses respondent choices rather than rankings or ratings to express product preferences. Discrete choice provides a more realistic context for many product decisions, and also allows the opportunity to express a lack of interest through a “None of these” option. Although discrete choice has traditionally been limited to deriving utilities at the aggregate level, recent advances have made it possible to create utilities for each respondent. This results in better overall estimation as well as opportunities for preference-based segmentation using choice-based techniques.
Discrete Variable – A quantitative variable that can assume a finite or at most a countable number of values such as the number of children in a family.
Discriminant Analysis – Similar to regression, but predicts group membership rather than strength of association against a metric variable. This technique is a useful way to find the characteristics that best define differences across groups. It can also be used in conjunction with cluster analysis to derive equations that allow the segmentation to be applied consistently to future observations.
Discussion Board – Or an online threaded discussion, is a private online forum in which users share information, ideas, and opinions. Respondents are invited to log-in at their leisure to answer questions. A trained moderator is present and clients can observe results in real time.
Discussion Guide – Also called a Moderator’s Guide, an outline of topics to cover or questions to ask during qualitative focus groups or interviews.
Dropout Rate – The percentage of qualified respondents who begin a survey but do not complete it.
Dyads – A focus group conducted with two people.
Estimate – A numerical value obtained from a statistical sample and assigned to the population parameter.
Ethnography or Observational Research – The observational study of human behavior in it’s natural environment.
Executive Interviews – Executive decision-makers use a different set of criteria from those of consumers. An in-depth interview provides extensive knowledge of the executive decision-making process. Phone or in-person interviews are conducted to complete these interviews.
External Validity – The extent to which causal relationships measured in an experiment can be generalized to outside persons, settings and times.
Factor – An underlying construct defined by a linear combination of variables.
Factor Analysis – An approach that, like cluster analysis, identifies relationships without using an outcome (dependent) variable. Grouping related characteristics instead of related people, factor analysis reveals unobserved “dimensions” that underlie a larger number of observed variables. This technique can either identify a subset of variables to represent these dimensions, or derive new variables that are composites of the original variables associated with each dimension. In either case, subsequent analyses (e.g. regression or cluster) can benefit from variable reduction.
Familiarity – Level of knowledge of a brand or product.
Focus Group – A research discussion group between eight to twelve people conducted by a moderator and designed to create a somewhat open, free flowing discussion about a general topic.
Focus Group facility – The central location where focus groups are held.
Group Dynamics – The interaction among people in a group. An effective moderator can enable group dynamics to promote helpful discussion by various techniques, as well as minimize the potentially negative effects of group dynamics.
Independent Variable – In an experimental setting, a variable that is controlled or manipulated by the researcher. In most multivariate analyses, however, these are simply the variables used to predict an outcome (or dependent) variable. Also known as predictor variables.
Infographic – A visual image such as a chart or diagram used to represent information or data.
In-Depth Interviews (IDIs) – A qualitative research technique that involves conducting intensive individual interviews to explore their perspectives on a topic.
In-Home/Observational Research – Often times, people act differently than how they say they act, therefore, it is important to view people doing activities or using products in their home or office environment to better understand their behavior. This enables a client to develop better solutions for their customers. Asking questions is always worthwhile, but observing behavior adds another rich layer of understanding.
In-Store Interviews – Asking people questions at the point of purchase is a valuable tool used to understand the reasons and motivations of a specific decision. In a facility setting, the respondent may forget why they selected a specific brand at that crucial moment. By intercepting shoppers we can gain an edge and discover market-actionable insights.
Insights Association – A trade organization for market research which outlines guidelines and code of ethics for researchers.
Intercept – A recruitment method in which an interviewer stops people in a mall or other public location and administers a survey.
Interviewer – The person responsible for recruiting participants for a focus group or the person administering a questionnaire.
Interviewer Error – An error that results from conscious or unconscious bias in the interviewer’s interaction with the respondent.
Key Drivers Analysis – A statistical approach to prioritizing the relative impact of various factors on an outcome (dependent) variable of interest. Key Drivers Analysis (KDA) is often used to compare the importance of a feature or attribute against the strength of performance. While it is typically synonymous with Regression Analysis, KDA can be conducted using a variety of Multivariate Analysis.
Level of Significance – The stated probability at which a hypothesis is either accepted or rejected. In most cases, differences are accepted only if they occur at the 95% level or higher. This is interpreted to mean that, by accepting the statistical difference, we are observing an actual population difference in 95 out of 100 cases. In the other 5 cases, the observed difference is actually attributable to random error.
Likert Scale – One of the most common approaches to capturing attitudes or opinions, Likert Scales use a Bipolar Scale that captures the strength of agreement or disagreement with multiple items that combine to form an empirically validated concept. The term is generally used to describe any single question using a 5-point scale, although this obscures distinctions with other scale constructs and anchors.
Logistic Regression – A variation on OLS regression that predicts a binary outcome such as agreement (yes vs. no) or purchase (buy vs. not buy). Discriminant analysis can also be used to predict dichotomous group membership, but “logit” is generally preferred due to its broader applicability. Logit output indicates whether each predictor variable increases or decreases the probability of the outcome.
Logit – A variation of logistic regression that is typically applied to the analysis of respondent choices. Multinomial logit (MNL) is the preferred method for synthesizing the impact of multiple-predictor variables on a categorical outcome involving more than two choices, such as choice-based conjoint data. As with logistic regression, MNL output indicates the impact of each characteristic on the probability of brand choice.
Market – The total of all individuals or organizations that represent potential buyers.
Marketing Research – The specification, gathering, analyzing, and interpretation of information that links the organization with its market environment.
Market Segmentation – The development and pursuit of marketing programs directed at subgroups or segments of the population that the organization could possibly serve.
MaxDiff Analysis – An approach to prioritizing many features or attributes by asking individuals to identify the most and least important items. People are often better at identifying extremes (“maximum differences”) than at making granular assessments. By repeatedly identifying the best and worst items among a subset of features, we are able to derive individual ratings that are discriminating and free from scale usage bias. Advanced applications of MaxDiff will implement anchoring techniques to overlay resulting scales with meaningful thresholds.
Menu-Based Conjoint – A variation of Choice Modeling that applies to decisions that include multiple stages, bundles or customization. Menu-Based Conjoint (MBC) examples include simulating restaurant orders or car configurations, but MBC can be applied to any purchase model that requires multiple choices, customization, or product cannibalization. MBC is more complex than tradition choice modeling, which has implications for base size and project scope.
Methodology – The research procedures used; the section of the final report in which the researcher outlines the approach used in the research, including the method of recruiting participants, the types of questions used, etc.
Mini Groups – A focus group of generally four to six people.
Mobile Surveys – A method of data collection means by using functions of mobile phones, smartphones and tablets. It makes use of strengths from mobile communication and applies these strengths to research purposes.
Multiple Classification Analysis (MCA) – An extension of regression analysis that is typically used in satisfaction research. Unlike regression, it allows the analyst to find exceptional (non-linear) relationships that demonstrate the “penalty” or “reward” associated with various levels of each predictor variable. MCA requires larger sample sizes than regression, and is most valuable when examining the impact of variables with discrete values, such as attribute ratings.
Mutually Exclusive – Events are said to be mutually exclusive if they have no intersection.
Non-Response Bias – An error due to the inability to elicit information from some respondents in a sample, often due to refusals.
Non-Reponse Error – An error that occurs due to non-participation of some eligible respondents in the study. This could be due to the unwillingness of the respondents to participate in the study or the inability of the interviewer to contact the respondent.
Normal Distribution – A continuous distribution that is bell shaped and symmetrical about the mean.
Observation – A data collection method where the relevant behaviors are recorded.
On the Street Interviews – A qualitative method in which a respondent is stopped on the street and asked for their feedback. Usually lasts between 5-10 minutes.
One-on-One or Depth Interview – A discussion between the moderator and interviewee. This technique avoids group-think and enables deep probing of the subject matter.
Online Communities (MROCs) – A targeted group of people who are recruited into a private online venue to participate in research-related activities over an extended period of time.
Perceptual Map – A spatial representation of the perceived relationships among objects in a set, where the objects could be brands, products, or services.
Pop-Up – A form of surveying that shows a pop-up window invitation while respondents are on a website.
Population – The collection of all objects that are of interest to the statistician. The elements of a population may be called units or subjects. Also known as the universe.
Positioning – Location of a brand or product in consumers’ minds relative to competitive products.
Pretest – The presentation of a questionnaire in a pilot study to a representative sample of the respondent population in order to discover any problems with the questionnaire prior to full-scale use.
Price Sensitivity Meter – A series of questions used to determine perceived normal price, penetration price, highest reasonable price, and lowest reasonable price.
Primary Data – Data collected to address a specific research objective.
Primary Research – Conducting research to collect new data to solve a marketing information need. See also secondary research.
Probability Sampling – Any sampling method where the probability of any population element’s inclusion is known and is greater than zero.
Probing – A follow-up technique for getting complete responses to open-ended questions by asking.
Profile Analysis– The comparison of evaluations of the alternatives in a consideration set, on the important and determinant attributes.
Projectability – The capability of research results to be extrapolated to the larger universe, on the assumption that the sample is representative of the total.
Projection – An estimate, based on assumptions about future trends in births, deaths, and migration, of a demographic characteristic such as population or number of households. Forecasts and projections are terms that are often used interchangeably.
Psychographics – Research that attempts to explain behavior by analyzing people’s personality traits and values. Often associated with lifestyle research.
Qualitative Research – Research designed primarily for exploratory purposes, such as getting oriented to the range and complexity of consumer activity, clarifying the problem, and identifying likely methodological problems. Examples include focus groups, case studies, and one-on-one interviews.
Quantitative Research – Research conducted for the purpose of obtaining empirical evaluations of attitudes, behavior or performance. Designed to generate projectable numerical data about a topic.
Questionnaire – A set of questions designed to generate data necessary for accomplishing the objectives of the research project.
Quota< – A defined, required number of units.
Quota Sampling – A non-probability sampling method that is constrained to include a minimum from each specified subgroup in the population, regardless of their actual probability of inclusion.
Random Variable – A variable whose value is determined by the outcome of an experiment in which the outcome is subject to chance.
Randomization – A procedure in which the assignment of subjects and treatments to groups is based on chance. Randomization ensures control over the extraneous variables and increases the reliability of the experiment.
Regression Analysis – A fundamental and versatile research technique that seeks to explain an outcome (dependent) variable in terms of multiple predictor (independent) variables. This analysis reveals the nature and strength of the relationship between each predictor variable and the outcome, independent of the influence from all other predictors. The term typically refers to Ordinary Least Squares (OLS) regression, which models a linear relationship among variables.
Relevance – A criterion used to judge whether a market research study acts to support strategic and tactical planning activities.
Reliability – The random error component of a measurement instrument.
Representative Sample – A sample in which each unit has a known probability of selection that is accounted for either through sampling or weighting. A representative sample is obtained using probability, or random sampling.
Research Objectives – A precise statement of what information is needed, consisting of the research question, the hypothesis, and the scope or boundaries of the research.
Research Process – The series of stages or steps underlying the design and implementation of a marketing research project, including the establishment of the research purpose and objectives, information value estimation, research design, and implementation.
Respondent – The individual from whom data are collected. Also called participant, unit, unit of analysis, subject or experimental unit.
Respondent Fatigue – When a respondent begins to lose interest in a survey (due to length, complicated questions, etc.) and can provide invalid or inaccurate responses.
Response Bias – The tendency of respondents to distort their answers systematically for a variety of reasons, such as social desirability or prestige seeking.
Response Error – Error that occurs due to the respondents providing inaccurate information (intentionally or unintentionally). This might be due to the inability of the respondent to comprehend the question or misunderstanding the question due to fatigue or boredom.
Sampling Frame – A listing of population members from which the sample is drawn.
Sampling Frame Error – Error that occurs when the sample is drawn from an inaccurate sampling frame.
Sampling Unit – Any type of element that makes up a sample, such as people, stores, or products.
Screener – Questions used to capture the appropriate respondents for a particular research solution.
Screening Sample – A representative sample of the population being studied that is used to develop or pretest measurement instruments.
Secondary Data – Data collected for some purpose other than the present research purpose.
Secondary Research – Analyzing information from previously conducted research projects. See also primary research.
Segment – Portion selected on the basis of a special set of characteristics. Also used to describe the outcome of a segmentation, such as with cluster analysis.
Segmentation – A type of advanced analysis that divides respondents into groups who are similar to each other and differentiated from other groups.
Semantic Differential Scale – A rating scale that uses opposing terms or statements to anchor end points of a Bipolar Scale. Semantic Differential scales differ from Bipolar Scales in that the end points do not necessarily have to be symmetrical with one another, and the scale itself is based on agreement or association with the opposing terms rather than identifying different levels of the terms themselves.
Sensitivity – The ability of a measurement instrument to discriminate among meaningful differences in the variable being measured.
Significance Level – The probability of obtaining the evidence if the null hypothesis were true.
Significant Difference – In mathematical terms, difference between tests of two or more variables. The significance difference varies with the confidence level desired.
Simple Random Sampling – A sampling method in which each population member has an equal chance of being selected.
Skip Pattern – A form of survey logic that uses answers to previous questions to determine whether or not respondents should be asked certain questions.
Snowball Samples – A form of non-probability sampling where respondents provide referrals for additional research respondents.
Standard Deviation – The square root of variance.
Standard Error – The standard deviation of a distribution of sample means; the square root of the variance of the sampling distribution.
Standard Industrial Classification (SIC) System – A uniform numbering system developed by the US Government for classifying industrial establishments according to their economic activities.
Statistic – Any of several characteristics of a sample.
Statistical Control – Adjusting for the effects of confounded variables by statistically adjusting the value of the dependent variable for each treatment condition.
Stub – A row heading in data tables or tabulations.
Survey Method – A method of data collection, such as telephone or personal interview.
Syndicated Research – Studies in which the sponsoring research company defines the audience to be surveyed and the interval between studies and the questions to be asked. Clients share the same results and costs.
Test Marketing – The introduction of a new product in selected test cities that represent a typical market, so that results of the performance in these markets can be projected on a national basis.
Tests of Significance – Tests for determining whether observed differences in a sample are sufficiently large as to be caused by something other than mere chance.
Top of Mind Awareness – The very first response to open-ended questions such as brand awareness.
Topline – A brief summary of the preliminary results of the research.
Tracking Studies – Monitoring the performance of things like advertisements or brands by regular surveying of the audience.
Trade-Off Approach – A method of collecting data for trade-off analysis in which the respondent is asked to rank each combination of levels of two attributes from most preferred to least preferred.
Transcript – The written account of a focus group or other qualitative interview (such as one-on-one interviews).
Triads – A focus group conducted with three people.
Unipolar Scale – A rating scale that captures the presence or absence (e.g. not at all vs. completely) of attitudes or evaluations. Unipolar scales can invoke as few as 3 scale points, and use escalating modifiers (e.g. somewhat, very) to describe values as they deviate further from a baseline level. Midpoint values are not used the same as in a Bipolar Scale and should be avoided other than to identify a halfway point between presence and absence.
Universe – Also know as the population, the universe is the entire group from which respondents are selected.
Usability Test – Test of a website, software, or other product to determine its effectiveness for the user or customer.
Utility – In trade-off analysis, the worth or value of each level of each attribute relative to the other levels.
Variable – Any characteristic that can be measured on each unit of the population.
Variability – Differences in the measurement of variables.
Variance – A measure of dispersion based on the degree to which elements of a sample or population differ from the average element.
Verbatim – Transcribing the provided response of a respondent word-for-word.
Web Based Research – A method of qualitative or quantitative research that takes place via the web. For qualitative research, this often occurs in real time and lends itself to geographically dispersed segments of the market that would otherwise be difficult to recruit for a traditional focus group. For quantitative research, questionnaires are written for the Internet and delivered to the respondent through email invitations, pop-up surveys, or banner recruitment.
Weighting – A procedure by which each response in the database is assigned a number according to some specified rule. Probability weighting is a theoretically-based method by which respondents are given greater or less weight based on their probability of inclusion in the sampling scheme. Post hoc weighting is an atheoretical method that attempts to account for various types of bias by weighting the sample to match an external reference on a variety of characteristics.