AI Open-End™ terms

AI Open-End™ process

Reports and LiveSlides

Panel and Statistics

Study design

Study stages

GroupSolver's unique AI-powered open-ended question. Respondents answer open-ended questions in their own words. Then they evaluate other survey answers which highlight audience commonalities. Our AI calculates statistical support and organizes responses into key insights (themes).

The AI Open-End™ Manager tab allows users to review and manually edit themes and answers. Users can also access the AutoThemes feature here.

GroupSolver’s capability that automatically places similarly worded answers into color-coded themes. This feature is found in the "AI Open-End™ Manager" tab of the GroupSolver dashboard.

"Cluster mean sim" describes how close semantically are the contents of a single theme. This score is the average of all pairwise cosine similarity scores within the answers of a given theme cluster. The cosine similarity score can range from 0 to 1, where a score of 0 means no similarity at all, and a score of 1 means that the two answers are identical.

The standard deviation of the similarities for all answers within the theme. A theme with similar answers will have a large Cluster Mean Sim and a small Cluster SD sim.

The likelihood that a respondent who views one answer would agree with another.

This is the list of answers that represents key responses to the question and it excludes duplicate answers, gibberish or otherwise low-value responses. Typically, IdeaPool size ranges up to 75 answers, which receive evaluations from respondents.

Theme in GroupSolver is a group of answers with similar meaning. AutoThemes is the feature in which the AI automatically creates groups of similar answers and color codes them.

The Consolidation phase happens when a respondent is answering an AI Open-End™ question within a GroupSolver study. Respondent might be asked if a similar, previously given answer could replace their verbatim. This allows GroupSolver to remove duplicate answers. Respondents will only go through this phase if the AI believes there are answers similar enough to be categorized together but not similar enough to be combined automatically.

The Evaluation phase happens when a respondent is answering an AI Open-End™ question within a GroupSolver study. During this phase, a respondent will view previous verbatim responses from other respondents and be asked to select between "agree," "disagree," "neutral," or "unclear answer." We use evaluation data to calculate support.

The Ideation phase of a GroupSolver study refers to when respondents are asked to provide their unaided response to an AI Open-End™ question. GroupSolver filters will automatically check verbatims for gibberish, profanity, and whether or not the verbatim holds value (such as if someone was to write in "I don't care").

A representative response that has passed GroupSolver filters and that embodies the meaning of a group of responses received during ideation.

The Consensus Solution™ visualization displays correlation between one answer and all other respondents' answers.

The Download tab allows you to download specific study data, which you can then manipulate and customize, upload into varying analysis tools of your choice, and/or see all verbatim data in a clear-cut format.

Estimated percentage of the sample that would agree with an answer.

The IdeaCloud is a variation of the "word cloud" concept. Rather than showing keywords, we offer themes or answers, allowing you to quickly recognize which answers are top-of-mind with your audience.

The IdeaCluster visualization is a reporting tool that allows you to quickly see how answers or themes of an AI Open-End™ question correlate with one another. IdeaCluster reduces correlation matrix from n-dimensions to a 2-domensional chart.

Importing data using the Import tab allows you to upload additional data into the GroupSolver Dashboard based on a unique identifier. You can connect past data for tracker studies, upload demographics and other information from an existing customer list, create complex segments in Excel, and import these segments into the GroupSolver Dashboard.

IntelliSegment is a helpful visual report that allows you to find the differences or similarities between two or four distinctrespondent segments among answers to an AI Open-End™ question.

Number of survey participants who provided an answer extracted from their verbatim response.

Our unique QuantQual methodology lets you connect the what to the why by estimating the impact of open-ended answers on quantitative data.

P-Value is the probability of obtaining test results at least as extreme as a result observed, assuming the null hypothesis is correct.

A normalized statement derived from one or more verbatim responses given by respondents.

A segment is a set of respondents grouped by their demographics or by their answers within the study. You can create segments around any answers to a question asked within a study.

This is a conservative support estimate based on the Bayesian Statistical model. It is defined as the lower bound of the 90% credible interval. There is a 95% probability that the support is at least as high as the percentage shown.

Estimated percentage of the sample that would agree with a theme.

LiveSlides is a powerful tool that allows you to create shareable presentation with live data while the survey is still running.

The number of respondents who abandoned the study before completion.

The panel of respondents you can order online directly from the GroupSolver platform.

The Incidence Rate, or the IR, measures the percentage of respondents who are eligible to participate in a study. The actually achieved IR of the particular study is calculated by dividing the number of completes by the number of respondents entering the study.

The last time the study data was processed and statistics were recalculated. Live study information is continuously updated. Once the study has been solved, some data will no longer need updating. Editing segments and changes in AI Open-End Manager will trigger recalculation even when the study is solved.

The LOI stands for Length of Interview. You'll notice this industry-standard term used throughout the dashboard. It denotes the length of time respondents take to complete a study.

Overquota stands for the number of over-quota terminations. This termination occurs when a respondent is disqualified from a study because quotas have already been filled.

A redirect allows the client to add a link that creates a flow from one survey platform to another, thus enabling the respondents to seamlessly continue from one survey platform to another. You can find this feature in Panel settings in the Audience tab.

Self-distribution panel is automatically created when you start designing your study. It contains a link to your study that you can share with the respondents.

This refers to a unique query string that enables you to pass data to another website or survey.

With blocks, you can improve your navigation in study flows by organizing parts of the study into collapsible groups. You can add questions, messages, or logic modules to the block. In the study design, each block has a name, and a slightly transparent blue box separates its content from the rest of the study. When creating LiveSlides from a study, blocks used in the flow will also be used to organize the LiveSlides presentation.

The Code is a numerical value automatically assigned to each option in the choice questions. You can change the option code for each option even after the study has been solved.

"Complete" endpoints will end the study session for respondents and will include their partial data in the final data pool.

Unlike “Percent Quota” that controls sample composition throughout fielding, “Completes Quota” only sets the maximum limit regardless of when that limit may be reached. Once the maximum value of the quota is met, respondents who meet the specified criteria will be terminated from the study.

Use conditions to direct respondents down different paths in the survey, terminate them, or ask for elaboration on how they answered choice questions.

Endpoints are points in the study where respondents finish their journey. They allow you to control who qualifies to take your study by terminating the study session for respondents who do not meet specific criteria. These are commonly used in the Screener section of your study. Other endpoints include overquotas, quality terminates or completes..

The History filter allows you to control the visibility of past questions and responses within a study.

When adding a new module to your survey flow, you can select from templates in your library. You can save your templates (private library) or choose from common questions we prepared for you (public library).

The Libraries tab on the account dashboard stores all the templates within your account for future use.

A module is an element within the study, such as the question, message, or condition that can be added to the session flow.

The Module Randomizer allows you to select modules to randomize in your study.

"Overquota" endpoints will end the study session for those respondents who exceeded a quota set in the study conditions. Their responses will not be included in the results.

This allows you to control the composition of a sample by setting percent of respondent meeting specific criteria allowed to enter the study during the fielding of the study.

Proceed Modules allow you to control the pace that respondents go through the study by stopping the respondent's progress until they confirm to continue.

"Live preview" allows you to preview the study from the respondent's perspective before you launch your study and start collecting data.

"Quality terminate" will end respondent’s session for being of "poor quality." Their data will not be counted toward the final data pool and they will not affect the incidence rate of the study.

Quotas allow you to control your sample by setting maximum number of respondents that meet specific criteria who are allowed to complete the study.

You can create any block of the study or full study into a template to be used in you the future work.

Terminate endpoint will end the study session for a respondent. Data from a terminated respondent will not be included the final data pool.

A track refers to the different branches or paths of a survey that respondents might take depending on filters or quotas incorporated into the survey design.

The Track randomizer allows you to select tracks that can be presented to respondents in random order. You can indicate how many of the tracks a respondent can enter.

Welcome screen is the first screen that respondents see. You can add the logo or a picture, name, and description and change the color style visible to the respondents.

When using the GroupSolver dashboard, your studies may become cluttered and disorganized. Changing the study status allows you to organize your studies more efficiently. By archiving a study, you are saying you are finished working through the data, and the study is no longer needed in the dashboard. The system will not delete the data, but you will only have access to data download.

This refers to one of the five "stages" a study may take. A “collecting data” study means it has begun to collect respondents. When the study is in the "collecting data" stage, you can only perform minor adjustments in the study design. It is possible to edit the wording of the question/option to allow for correcting possible grammatical errors. You can also edit the option codes to fix potential mistakes in the study design.

Studies that have yet to be launched are in the draft stage. These are new studies still in the building process or older studies that have never been launched.

You can pause a study that is collecting data. No new respondents will be allowed to enter the survey until you resume collection, however, respondents already in the study will be allowed to complete their session.

A solved study is a study that has finished data collection. You can mark the study as “solved” any time, even before your targeted sample is reached.