Method and apparatus for the design and analysis of market research studies
First Claim
1. A database and reporting program allowing a computer to be turned into a tool that eases the task of designing custom research studies, by allowing the input and storage of all the necessary data elements needed in a research survey, and by analyzing those stored data elements and reporting the summary results of that analysis, through a method comprising the following steps:
- a. designing the questionnaire, allowing for a plurality of questions, each with a plurality of response categories, and prompting users to specify the following data elements at design time;
(i) the question label, (ii) the question text, (iii) whether the response categories to a question are to be read to respondents, not read to respondents, or read only if necessary, (iv) whether to list the response categories in the order they appear, to randomize them, or to rotate the list, (v) whether to have respondents indicate which response categories are applicable, or whether to have respondents rank response categories in some order (such as from most important to least important), (vi) the number of response categories to be ranked or to be allowed to be selected from the entire list of response categories, (vii) whether to allow answers other than those that appear on the list and how many such other answers to accept, (viii) whether to include a no/none/nothing response category, (ix) whether to include ‘
do not know’ and
‘
refused’
response categories (also known as unspecified responses), (x) any skip patterns associated with the response categories;
b. storing the data elements specified in step (a) of claim 1 for later analysis and reporting;
c. designing the data tabulation plan (tab plan) by prompting users to specify the following at design time and storing it in the database specified in step b of claim 1;
(i) the table label, (ii) whether to include or exclude unspecified responses in the table total, (iii) who should be asked the question and who should not be asked the question, (iv) whether the question is part of a battery of questions, such as attitudinal, psychographic, or behavioral questions, all using the same scale (see FIG.
2), to allow for summary tables when combined with a battery label, (v) whether the questions are scalar or categorical, 1. if the questions are scalar, which of the following statistics to calculate for the table;
mean, median, mode, standard deviation, standard error, 2. if the questions are categorical, whether to calculate the mode for the table;
d. providing a method for reordering the question order without having to manually readjust skip patterns (shown in FIG. 4 and described by module 1 in the code appendix);
e. outputting a questionnaire in any of the following formats, any of which could be viewed either on screen or by printing them out to read on paper;
(i) client copy-a version suitable for a market research client to review, (ii) an annotated field copy-a version suitable for field supervisors, CATI programmers and data analysts, (iii) an HTML version-a version suitable for posting to a web-site, although lacking in data collection functionality unless programmed by an experienced HTML/CGI survey programmer, (iv) an in-person intercept version-a version suitable for mall-intercepts, (v) a tab plan or tab plan script, (vi) an analysis plan;
f. producing banner and table syntax for use by data tabulation programs. g. populating the data-set with test data;
h. processing test data and tabulating it into tables;
i. statistically testing the tables from step (h), above, using the z-tests of means and proportions (see FIG. 5 and module 5 in the code appendix);
j. building a lexicon through user input and programmed prose syntax and storing the text thereby generated;
k. preparation of a detailed findings containing the following information;
i. question number, as assigned by the program, followed by the question title as a header, ii. a statement of the question'"'"'s base as a sub-header, iii. below the sub-header a table showing the total column (if proportional or weighted data), and the one or two sub-groups of specific interest for the study (e.g., customers vs. non-customers, or different clusters from a segmentation study), in tracking studies this is each wave'"'"'s total respondent column;
iv. detailed tabulations from one wave to the next of a tracking study which appear in the tables for that one wave, under the statistically-tested table appear any trends (if in a tracking study) or total, in descending order (if for a one wave study). The total appears in descending order on scaled questions (i.e., revenues), under the trends/total is the total (as above) if a tracking study. If not a tracking study, any instructive sub-group differences from the complete data tables, including all banner points studied, are discussed briefly;
v. An example of a report page appears below (a formatted version of this appears in FIG.
7);
Q1. Age of Respondent Note;
Asked of all respondents.ABWave 1 respondentsWave 2 respondents 18 to 2433%B25%25 to 3433%50%A35 and older33%B25%N600
Trends A greater proportion of respondents in Wave 2 than in Wave 1 are 25 to 34. This Wave'"'"'s Findings One quarter (25%) of Wave 2 respondents are 18 to 24. Half (50%) the Wave 2 respondents are 25 to 34. One quarter (25%) of Wave 2 respondents are 35 and older. Sub-Group Differences of Note Males (30%) are more likely to be 35 and older than females (20%). l. allowing the creator of a questionnaire design to produce an entire detailed findings section of a market research report, requiring only a modicum of market research experience to create credible market research results, through the method of claim 1, steps (a) through (k). m. preparing and displaying a complete preview of the resulting style of study results based on the questionnaire as asked prior to the fielding of surveys, including tables and a report populated by test data in steps (f) and (h) of claim 1. n. a questionnaire design module that requires analysts to think out all the possible outcomes of any given question, thereby increasing the quality and consistency of results by pulling together all the different stages of the market research process into one cross-indexed, interconnected database application program, as provided in this invention in step (b) of claim 1. o. What I further claim, in regards to claim 1, is that the questionnaire module, by specifying the base for each question, and then porting this information out to the banner specification program module, eliminates the problem of basing questions off of the wrong number of respondents, thereby greatly reducing error and the time loss associated with correcting that error, and so increasing the quality and accuracy of research outcomes, as well as eliminating rework hitherto required due to incompatible storage of this information between one phase of the market research process and another.
600
0 Assignments
0 Petitions
Accused Products
Abstract
This invention automates the most common processes in market research study design and analysis. By combining a database framework designed to hold all the data elements used in a market research survey, codifying heuristics actually used by analysts to make decisions about survey analysis, and building a lexicon to communicate summary results to lay readers. First, this provides a stringent framework for designing a questionnaire and study plan which will lead to clear tab plans and analysis plans. Second, by automating the research design and analysis process, it eliminates the human error associated with the intricacies of questionnaire design, tab plan design, and research report writing. Thirdly, it also greatly reduces the time necessary to rerun an entire report based on simple changes to any study element definitions to a few hours, a process formerly taking days or weeks. Lastly, it reduces the need for experienced researchers in the research design and analysis process, freeing analysts to spend a greater proportion of their time working on the more mentally challenging and demanding work of synthesizing the market research survey to produce credible findings, rather than the redundant work of data tabulation and summary reporting.
50 Citations
3 Claims
-
1. A database and reporting program allowing a computer to be turned into a tool that eases the task of designing custom research studies, by allowing the input and storage of all the necessary data elements needed in a research survey, and by analyzing those stored data elements and reporting the summary results of that analysis, through a method comprising the following steps:
-
a. designing the questionnaire, allowing for a plurality of questions, each with a plurality of response categories, and prompting users to specify the following data elements at design time;
(i) the question label, (ii) the question text, (iii) whether the response categories to a question are to be read to respondents, not read to respondents, or read only if necessary, (iv) whether to list the response categories in the order they appear, to randomize them, or to rotate the list, (v) whether to have respondents indicate which response categories are applicable, or whether to have respondents rank response categories in some order (such as from most important to least important), (vi) the number of response categories to be ranked or to be allowed to be selected from the entire list of response categories, (vii) whether to allow answers other than those that appear on the list and how many such other answers to accept, (viii) whether to include a no/none/nothing response category, (ix) whether to include ‘
do not know’ and
‘
refused’
response categories (also known as unspecified responses),(x) any skip patterns associated with the response categories;
b. storing the data elements specified in step (a) of claim 1 for later analysis and reporting;
c. designing the data tabulation plan (tab plan) by prompting users to specify the following at design time and storing it in the database specified in step b of claim 1;
(i) the table label, (ii) whether to include or exclude unspecified responses in the table total, (iii) who should be asked the question and who should not be asked the question, (iv) whether the question is part of a battery of questions, such as attitudinal, psychographic, or behavioral questions, all using the same scale (see FIG.
2), to allow for summary tables when combined with a battery label,(v) whether the questions are scalar or categorical, 1. if the questions are scalar, which of the following statistics to calculate for the table;
mean, median, mode, standard deviation, standard error,2. if the questions are categorical, whether to calculate the mode for the table;
d. providing a method for reordering the question order without having to manually readjust skip patterns (shown in FIG. 4 and described by module 1 in the code appendix);
e. outputting a questionnaire in any of the following formats, any of which could be viewed either on screen or by printing them out to read on paper;
(i) client copy-a version suitable for a market research client to review, (ii) an annotated field copy-a version suitable for field supervisors, CATI programmers and data analysts, (iii) an HTML version-a version suitable for posting to a web-site, although lacking in data collection functionality unless programmed by an experienced HTML/CGI survey programmer, (iv) an in-person intercept version-a version suitable for mall-intercepts, (v) a tab plan or tab plan script, (vi) an analysis plan;
f. producing banner and table syntax for use by data tabulation programs. g. populating the data-set with test data;
h. processing test data and tabulating it into tables;
i. statistically testing the tables from step (h), above, using the z-tests of means and proportions (see FIG. 5 and module 5 in the code appendix);
j. building a lexicon through user input and programmed prose syntax and storing the text thereby generated;
k. preparation of a detailed findings containing the following information;
i. question number, as assigned by the program, followed by the question title as a header, ii. a statement of the question'"'"'s base as a sub-header, iii. below the sub-header a table showing the total column (if proportional or weighted data), and the one or two sub-groups of specific interest for the study (e.g., customers vs. non-customers, or different clusters from a segmentation study), in tracking studies this is each wave'"'"'s total respondent column;
iv. detailed tabulations from one wave to the next of a tracking study which appear in the tables for that one wave, under the statistically-tested table appear any trends (if in a tracking study) or total, in descending order (if for a one wave study). The total appears in descending order on scaled questions (i.e., revenues), under the trends/total is the total (as above) if a tracking study. If not a tracking study, any instructive sub-group differences from the complete data tables, including all banner points studied, are discussed briefly;
v. An example of a report page appears below (a formatted version of this appears in FIG.
7);
Q1. Age of Respondent Note;
Asked of all respondents.A B Wave 1 respondents Wave 2 respondents 18 to 24 33% B 25% 25 to 34 33% 50% A 35 and older 33% B 25% N 600
600
Trends A greater proportion of respondents in Wave 2 than in Wave 1 are 25 to 34. This Wave'"'"'s Findings One quarter (25%) of Wave 2 respondents are 18 to 24. Half (50%) the Wave 2 respondents are 25 to 34. One quarter (25%) of Wave 2 respondents are 35 and older. Sub-Group Differences of Note Males (30%) are more likely to be 35 and older than females (20%). l. allowing the creator of a questionnaire design to produce an entire detailed findings section of a market research report, requiring only a modicum of market research experience to create credible market research results, through the method of claim 1, steps (a) through (k). m. preparing and displaying a complete preview of the resulting style of study results based on the questionnaire as asked prior to the fielding of surveys, including tables and a report populated by test data in steps (f) and (h) of claim 1. n. a questionnaire design module that requires analysts to think out all the possible outcomes of any given question, thereby increasing the quality and consistency of results by pulling together all the different stages of the market research process into one cross-indexed, interconnected database application program, as provided in this invention in step (b) of claim 1. o. What I further claim, in regards to claim 1, is that the questionnaire module, by specifying the base for each question, and then porting this information out to the banner specification program module, eliminates the problem of basing questions off of the wrong number of respondents, thereby greatly reducing error and the time loss associated with correcting that error, and so increasing the quality and accuracy of research outcomes, as well as eliminating rework hitherto required due to incompatible storage of this information between one phase of the market research process and another.
-
-
2. Additionally, I claim that this program provides the advanced data editing tools, including questionnaire renumbering, uniform data cleaning, and wave-to-wave comparison tools necessary to automate a custom research study.
a. Automated question and skip pattern renumbering: - what I further claim is that the questionnaire module, by specifying skip patterns based on question labels rather than question numbers, is an invention which eliminates the hitherto not uncommon problem of respondents being asked the wrong questions simply due to last minute changes in question numbering, and further, provides for last minute changes to skip patterns after all question numbering and skips have been set.
b. Uniform reassignment of mis-punched questions;
what I also claim is that the data editing capabilities of the program are such that if ranked questions are not perfectly completed by respondents, a module contained within the program will strip out the wrong responses and fairly (using a uniform random distribution) reapportion the responses, a data consistency problem hitherto corrected on an ad hoc basis by programs written by skilled database programmer-analysts.c. Wave-to-wave comparisons;
what I also claim is that the data tabulation program module, by allowing wave-to-wave comparisons, allows for the first time, an accessible, accurate, and cost-effective means to run tracking studies and to place wave-to-wave tables next to each other by means of a method other than manual labor.
- what I further claim is that the questionnaire module, by specifying skip patterns based on question labels rather than question numbers, is an invention which eliminates the hitherto not uncommon problem of respondents being asked the wrong questions simply due to last minute changes in question numbering, and further, provides for last minute changes to skip patterns after all question numbering and skips have been set.
-
3. What I also claim is that the lexicon implemented prose report generation method of my invention creates a template format of reporting such that rerunning studies with slightly different banner point definitions has only a minimal impact (i.e., requiring one to two hours of labor) compared with current techniques and methods of so doing, which usually takes two to three weeks.
a. What I also claim in regards to claim 3 is a method of automating the writing of the detailed findings and the strategic findings without the assistance of a skilled and trained research analyst by asking users to input words that relate the response categories to the question title or full text of the question and then combining that with its preprogrammed lexicon to produce prose reports. (See in FIG. 6.) b. I also claim in regards to claim 3 is that the application of the z-test to banner point definition by using automation and a random decision-making algorithm to determine splits in populations, which were unknown a priori, provides a new method for finding the optimal population splits on which to report, enabling the summary prose reports generated by to identify the differences normally requiring a trained analyst.
Specification