PureSpectrum’s Guide to Common Market Research Terminology

The features and functionality allow for audience targeting with custom profiling attributes. 

Market research that allows quick results and eliminates lengthy research processes.

Application Programming Interface. A set of tools used for building application software. In an online sample marketplace, this allows the buyer or seller to plug directly into a platform so it can be used directly on their own system.

When a respondent quits the survey before completion due to reasons including but not limited to technical glitches, disengaging survey design, outside distractions, etc.

The failure of a quality assurance test (IE knowledge test, red herring question, speeding, etc) established by the buyer.

How confident a researcher can feel about the data. Confidence is the sense of certainty that you have about the data. It is a statistical measurement from significant testing at different levels of confidence. The most widely used is at the 90% or 95% level. It helps to identify actual differences that can be reported with certainty. 

When a survey is not completed for a reason other than the respondent did not qualify. It can be calculated: Conversion = Completes / Completes + Incompletes that were not targeted.                                    

Foundational profiling attributes that can be used to target any respondent. Includes: Gender, Ethnicity & Race, Presence of Children in the Household, Marital Status, Employment Status, Education Attainment Status, etc.

Cost per completed interview in a survey. 

Way of presenting data to make it easily digestible. Cross tabulation is a quantitative research method appropriate for analyzing the relationship between two or more variables. Data about variables are recorded in a table or matrix. A sample is used to gather information about the variable. As an example, you may ask “what is your favorite color” and want to see if there are any similarities or differences if you look at “your favorite color” responses by age breaks or by gender.  

Data quality is measured by the information collected from the survey. This can be measured by the response time, red herring questions, and PureScore™. Responses that have high data quality are attentive, truthful, engaged, and from the proper targets. 

Quality measure that tracks a device to recognize if a person has already participated in a survey.

A theoretical number of completes a supplier may achieve on a quota/survey given the surveys Incidence rate (IR), length of interview (LOI), targeting, and field time. 

The amount of time an active survey is accepting responses. Field time is measured in days.

A willful misrepresentation for financial gain. PureSpectrum aims to catch fraud before respondents can corrupt buyer data or exploit suppliers. We check for fraud with device fingerprinting, bot checking, and blacklist checking.

Prevalence of a target, characteristic, or qualification in a targeted audience. For example, if 1 out of 10 people in a given population meets the criteria for a survey, there is a 10% incidence. It can be calculated: Incidence = Completes / Valid Clicks.

Combining multiple sample sources on panels via API into a single survey.

Short for Incidence Rate. The rate at which incidence of certain criteria occurs in a given population.

LOI stands for Length of Interview. How long will it take the average respondent to take the survey measured in minutes.

A MaxDiff question shows respondents a set of items, asking them to choose what is most and least important. When the results are displayed, each item is scored, indicating the order of preference. MaxDiff, short for maximum difference scaling, is sometimes referred to as best-worst scaling and is a way to distinguish between several important items.

Description of sample and fielding practices in a particular study.

Monadic testing is a type of survey research that introduces survey respondents to individual concepts in isolation. It is usually used in studies where independent findings for each stimulus are required, unlike in comparison testing, where several stimuli are tested side-by-side.

Preventing a respondent  from entering and answering the same survey from several different panels. 

Panel surveys are a type of longitudinal survey that measure consumer opinions, feelings, or thoughts over time. Panel research also allows for the collection of large amounts of information. For more information, read our article all about “what is a panel survey?”

Qualitative research is a type of social science research that collects and works with non-numerical data and that seeks to interpret meaning from these data that help understand social life through the study of targeted populations or places.  This type of research is usually collected by in-depth interviews (IDIs) or focus groups.

Quantitative research is the process of collecting and analyzing numerical data. It can be used to find patterns and averages, make predictions, test causal relationships, and generalize results to wider populations. Research is usually conducted today by online surveys.

The process by which we allow customers to reject and not pay for completed survey responses that they have identified as unacceptable.

A single survey taker.

How many respondents are needed to complete a survey.

There are several different types of significant testing done in research.  No matter what formula you use, it is to identify a difference in groups of data.  (Using a distance measure with a known probability distribution and a particular value of the distance measure is obtained. Most often in A/B testing the distance measure takes the form of a z score or t score which is compared to a specified rejection region.) These scores help researchers see if a claim should be supported or rejected.

A set of questions also known as a questionnaire is asked of a respondent in person, online, or over the phone.  

Is the outcome of significance testing.  Identifies if a difference between two or more groups of data is actually significant.  Shows a high probability of happening based on enough data to make confident assumptions. 

A project that repeats on a consistent cycle- weekly, monthly or quarterly. "Wave" is a similar type of study with the same concept, but might not use the calendar to set when each version will occur.

The security process of encrypting URLs that prevent the URL from being viewed by the respondent during the survey and being tampered with to display a faked completion.

A qualified respondent that is sent to the buyer’s survey that is not terminated by the buyer due to Buyer Security or Buyer Quality.