JP2013012120A - Questionnaire tabulation system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique for acquiring a tabulation result with high quality by removing responses made in a short time in a questionnaire via the Internet.SOLUTION: A questionnaire is divided into question patterns each of which is displayed on one screen of a terminal of each respondent. The transition order is determined depending on contents of responses. A questionnaire tabulation system measures a time required for response which is a time required by a respondent to respond to each question pattern, obtains a representative value of the time required for response of each respondent who responded to the question pattern for each question pattern, and obtains a response time index in the question pattern responded to for each respondent by dividing the time required for response by the representative value. The system obtains an average value by dividing the sum of the response time indexes by the number of the question patterns responded to and removes a prescribed ratio of responses in ascending order of the average values for the questionnaire tabulation.

Description

  The present invention relates to a questionnaire counting system, and more particularly to a questionnaire counting system for obtaining high-quality counting results in a questionnaire using the Internet.

  In recent years, survey methods for collecting data used for corporate product development and marketing by conducting questionnaires to respondents (sometimes called “monitors”) via the Internet have become widespread. This research method is also called net research. Companies that plan questionnaires according to their own purposes (hereinafter also referred to as “investigators”) may conduct their own surveys, but companies that conduct online research (hereinafter also referred to as “operators”) In some cases, surveys may be outsourced.

  In many cases, an operator who is engaged in net research has a member organization formed by registering each person's attribute information (gender, age, place of residence, etc.) on the monitor. Therefore, by entrusting the survey to the operator, it is possible to select a monitor according to the attribute information and improve the data analysis accuracy, and obtain a high-quality questionnaire result. This is also because the operator has know-how on how to create questions and extract results to extract useful answers.

  The operator creates questionnaire contents including a plurality of questions and an answer column for each question. There are various types of answers depending on the purpose, such as alternative, multiple choice, or free entry. The operator requests the monitor to participate in the questionnaire by e-mail or the like. Then, the monitor connects to the operator's website from a PC or mobile phone using a browser or a dedicated application, and answers the question online. When the number of responses reaches the specified number, the questionnaire is completed and the responses are counted.

Such survey methods, also called net research, have many advantages over traditional methods such as face-to-face and postal surveys. In other words, it is possible to call for registration as a monitor to a large number of people, such as using Internet advertisements. In addition, because the burden of monitors is relatively light and barriers to participation are low in Internet research, it is easy to expand the monitor population It is.
Furthermore, in Internet research, the process from the questionnaire request to the monitor to the answer by the monitor is completed online, and the answer is obtained as electronic data. As a result, it is possible to reduce the time required for conducting the questionnaire and counting the answers, and to reduce the cost.

  In addition, it is possible to narrow down the survey object based on the attribute information of the monitor and the result of the preliminary questionnaire, and it becomes possible to conduct detailed market research. In addition, there are various types of free-entry answers, such as being able to obtain a relatively large number of answers because of ease of input, and such free-entry answers in text data format, which are easy to analyze. There are advantages.

  Japanese Patent Application Laid-Open No. 2002-197232 (Patent Document 1) discloses a method for conducting a large-scale questionnaire using the Internet. Patent Document 1 describes a technique for increasing the number of respondents and improving quality by giving appropriate incentives to respondents to a questionnaire.

  Japanese Patent Laid-Open No. 2006-285691 (Patent Document 2) discloses a technique for improving the reliability of questionnaire results in an anonymous questionnaire on the Internet. Patent Document 2 describes a method of determining creditworthiness by classifying and organizing the contents that have been answered in the past while maintaining anonymity between the questioner and the respondent.

JP 2002-197232 A JP 2006-285691 A

  However, it is known that in Internet research, at a certain rate, low-quality answers are made without reading the question text well. An answer with low quality is, for example, a mutually contradictory answer to a plurality of questions, an answer made in a shorter time compared to other respondents, or the like.

  In particular, consider the background of responding in a short time. In the net research, a privilege is often given by an operator every time a monitor answers a questionnaire. As a privilege, there are points given according to the number of questions in the questionnaire, and the acquired points can be exchanged for prizes. In this way, using points as an inducement material, the number of monitors registered in the member organization is increased, and the willingness to answer individual questionnaires is increased.

  However, there is a person who answers a part of the monitor without looking closely at the question only for obtaining a privilege. For example, when selecting the first choice for all questions for a choice-type question or entering a meaningless short string (such as “Oh”) for a free-form question It is. In addition, as a result of a short-time answer, mutually contradictory answers may be made to the plurality of questions described above.

  When answers with low quality are mixed in this way, the reliability of the questionnaire is lowered, which adversely affects corporate activities using the tabulated results. In addition, when an operator conducts a questionnaire, the implementation of a high-quality questionnaire leads to the maintenance of trust. Therefore, it is necessary to exclude low-quality answers from the aggregated results. Hereinafter, such an operation is also referred to as “data cleaning”.

One of the data cleaning methods is a method of excluding a predetermined percentage of responses according to the surveyor's designation in the order of short response time. This is a method that takes advantage of the characteristic that it is easy to measure the time required for an answer because the net research is made by a response via the Internet and the communication is managed by a computer.
More specifically, first, the total time required for answering the questionnaire (total time required for answering) is calculated for each monitor. Then, the ratio of the monitor to be deleted is designated (for example, 3%), 3% of answers from the shorter total answer required time are excluded, and the remaining answers are totaled.

However, this data cleaning method causes a problem when the number of questions changes depending on the contents of the questionnaire. Hereinafter, a specific example will be described.
FIG. 3 shows a screen when the monitor answers a questionnaire from the browser of the PC, and a transition order thereof. This questionnaire includes six sets of questions and answer columns (Q1 to Q6). As the answer column, there are a radio button for selecting one of the options, a check box capable of selecting a plurality of options, an answer column for a free-entry question, and the like.

When the monitor connects to the operator's website, selects a questionnaire and starts answering, the total time required for answering begins. First, the monitor selects either the male or female radio button for Q1 asking for sex in FIG. 3A, and clicks the “Next” button. Then, the operator's server determines the content of the reply, and if “1. male” is selected, the process proceeds to Q2, and the screen of FIG. 3B is generated and returned. In this screen, Q3 to Q5 are also described in addition to Q2. The monitor reads Q2 to Q5, selects an answer, and clicks the “Next” button. Then Fig. 3 (
Q6 is displayed on the screen of c). On the other hand, if “2. Female” is selected in Q1, the process proceeds directly to the screen of FIG. When the monitor responding to Q6 clicks the “Done” button, the questionnaire is completed. At the end of the survey, the total response time is measured.

  At this time, the number of questions for the monitor who answered “1. Male” to Q1 is six questions Q1 to Q6, while the number of questions for the monitor who answered “2. Female” is only two questions Q1 and Q6. . Assume that it takes 1 minute to properly grasp the contents of one question and give a high-quality answer, and 30 seconds to give a low-quality answer without reading the question well. Then, a male monitor that gives a high quality answer is 6 minutes, a male monitor that gives a low quality answer is 3 minutes, a female monitor that gives a high quality answer is 2 minutes, and a female monitor that gives a low quality answer is 1 minute Total response time is required.

  In this situation, when data cleaning is performed to exclude answers in the order of the shortest total response time (for example, the top 3%), answers from female monitors are excluded even though they are not actually low-quality answers. I tend to. As a result, the tabulation result may be biased, for example, in line with the concept of male monitors. On the other hand, the data of male monitors who answered low quality remains in the tabulation.

  As another data cleaning method, there is a method of setting a minimum answer time for a question for each question type. When the total response time required for the monitor is less than the total minimum response time, the monitor's answer is excluded from the totalization. However, in this case, individual conditions for each question, such as the length of the question sentence, were not considered, and it could not be said that high-quality answers remained.

  As described above, the conventional data cleaning method may not be able to exclude low quality answers. Such a situation is not preferable from the viewpoint of the investigator conducting business activities using the tabulation results and from the viewpoint of ensuring the quality of the questionnaire by the operator.

  The present invention has been made in view of the above problems, and its object is to provide a technique for obtaining high-quality tabulation results by excluding answers made in a short time in a questionnaire via the Internet. There is.

  In order to achieve the above object, the present invention employs the following first configuration. That is, it is a system that obtains answers to a questionnaire from a plurality of respondents via the Internet and excludes a predetermined percentage of the answers, and the questionnaire is a unit displayed on the screen of the respondent's terminal. Each of which is divided into a plurality of question patterns each including at least one question, and the transition order of the question patterns is determined according to the contents of the answers. A communication means for distributing and displaying on the screen, obtaining a reply from the respondent, a measuring means for measuring a time required for the answerer to answer each question pattern, and the plurality of the means After obtaining responses from respondents, the representative value calculation that calculates the representative value of the time required for answering all the respondents who answered the question pattern for each question pattern After the representative value is calculated, for each respondent, the answer required time of the respondent is divided by the representative value calculated by the representative value calculating means in all question patterns answered by the respondent. An average value calculation means for calculating an average value by calculating an answer time index based on the value and dividing the sum of the answer time indexes of the respondents by the number of answer patterns, and after the average value is calculated A means for comparing the average values of the respondents, excluding the predetermined percentage of the answers in ascending order of the average value, and calculating a questionnaire based on the answers that are not excluded. It is a questionnaire totalization system characterized.

In such a questionnaire totalization system, the time required for answering is stored for each question pattern, which is a unit answered by an answerer (monitor). Then, for each question pattern, a representative value (for example, a median value) of the required response time is calculated. The response time for each respondent is normalized by dividing it by the calculated representative value to become the response time index, and examines the tendency of response time regardless of the length of the question text or the answer format. be able to. Furthermore, an average value is calculated by dividing the sum of the answer time indexes by the number of question patterns answered by the respondent, and an exclusion target is determined in ascending order of the average value. That is, it is determined that respondents with smaller average response time indices tend to respond in a shorter time. As a result, even if the number of questions varies depending on the respondent, it is possible to determine whether the answer is a short time on the same basis. That is, there is no problem that respondents whose total response time has been shortened due to the small number of questions are easily excluded from the totalization. Then, by collecting questionnaires based on the answers that are not excluded, it is possible to obtain a high-quality aggregation result.

  Moreover, in the questionnaire totalization system which takes said 1st structure, the following 2nd structures can also be taken. That is, the average value calculation means sets the answer time index when the value obtained by dividing the answer required time by the representative value is smaller than 1, and sets the answer time index when the value is 1 or more. It is a questionnaire totalization system.

  In such a questionnaire totalization system, when the required response time is longer than the representative value, the response time index is uniformly 1. Thereby, in the calculation of the answer time index, it is possible to suppress the influence of an abnormal value due to a long interruption of the answer. In addition, since it is empirically considered that the long-term response has a low influence on the quality of the survey, it is possible to obtain a high-quality total result even if this processing is performed.

  Moreover, in the questionnaire totalization system which takes said 1st or 2nd structure, the following 3rd structures can also be taken. That is, after the aggregation means has aggregated the questionnaire, the contents of the answers used for the aggregation are confirmed, and further includes confirmation means for determining whether to exclude the answers whose contents have been confirmed from the aggregation, Instead of answers that are determined to be excluded by the user, among the answers that are excluded by comparison of the average values, those that have a large average value are used as the answers that are used for the aggregation, and the questionnaire is totaled again It is a questionnaire totalization system.

  In such a questionnaire tabulation system, it is possible to check the answers used for tabulation by, for example, individual visual confirmation or filtering by the operator, and to exclude those judged to be inappropriate, and to improve the tabulation quality. It becomes possible to make it higher. As a result, when the number of responses used for aggregation becomes less than the predetermined number, the aggregation means selects the responses that have already been excluded and that have a relatively large average value and is considered far from the short-term responses, and re-aggregates By performing the above, it is possible to maintain a high-quality count result.

  According to the present invention, it is possible to provide a technique for obtaining a high-quality counting result by excluding answers made in a short time in a questionnaire via the Internet.

FIG. 1 is a block diagram illustrating the configuration of a questionnaire counting system. FIG. 2 is a flowchart showing a procedure of the entire process of the questionnaire counting system. FIG. 3 is a diagram showing an example of screen transition displayed on the screen of the monitor terminal. FIG. 4 is a flowchart showing a data cleaning procedure. FIG. 5 is a diagram illustrating an example of values calculated in the data cleaning procedure. FIG. 6 is a flowchart illustrating the overall processing procedure of the second embodiment.

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings. However, the constituent blocks described below and their relative arrangements should be appropriately changed according to various conditions of the system to which the invention is applied, and the scope of the present invention is limited to the following description. It is not a thing.
In the following embodiment, it is assumed that the operator performs a questionnaire to the monitor (respondent) in response to a request from the researcher, and performs data cleaning and aggregation processing. However, it is possible to apply the present invention to the aggregation of questionnaires regardless of who implements it.

<System configuration>
FIG. 1 is a block diagram for explaining the configuration of a questionnaire totalization system 1 (hereinafter also referred to as “system”) according to the present invention.
The server 2 is an information processing device managed by a questionnaire implementer. The server 2 includes a control unit 21, a communication unit 22, a totaling unit 23, and a storage unit 24 as functional blocks.

  The control means 21 controls and operates each block in the server based on a user's instruction (for example, an operator's operator) using the program or the console 3. Various arithmetic processes are performed. For example, regarding the contents of the questionnaire, the combination of questions to be presented on the screen of the monitor side terminal is determined according to a predetermined transition flow and the answer contents of the monitor. The communication means 22 communicates with the terminal 5 (5a to 5c) via the Internet 4. The communication means 22 is performed by providing information on a page including a question and an answer column for connection from the browser of the terminal 5 such as a PC or a mobile phone, and receiving the answer. At this time, the server 2 functions as a Web server. The contents of the received answer are stored in the storage means 24 by the control means 21 together with information relating to the answer time. The counting means 23 processes the response contents of the monitor after data cleaning, which will be described later, into a format desired by the investigator.

  For example, the server 2 is configured by a computer in which a CPU operates in accordance with the contents of a program called up on a memory. However, the actual server configuration is not limited to a single computer, and a plurality of computers may be used in conjunction with each other depending on the function and load. Further, each means included in the server 2 does not necessarily have to be physically configured separately, but can be considered as a functional block realized by a program module corresponding to each means using hardware resources of a computer. May be. As the storage unit 24, various storage devices that are usually used in computers can be used as long as the information about the answer from the monitor can be stored until the questionnaire is completed.

The console 3 is an interface between the server 2 and the user, and is provided for checking information created or stored by the server on the screen or giving an instruction using an input means such as a keyboard. The console 3 and the storage unit 24 may be provided separately from the server 2 or may be formed as an integral unit.
The terminal 5 is an information processing apparatus that has a function of connecting to the Internet 4 and displays a question to the monitor and is used when the monitor inputs an answer. For example, a PC or a mobile phone can be used as the terminal 5. The display on the terminal and the answer are performed on a general browser or an application provided by the operator.

[Example 1]
In the present embodiment, a method for excluding answers made in a short time by performing data cleaning when a questionnaire having a relatively simple structure is executed and counted in the above-described questionnaire counting system will be described step by step. .

<Composition of questionnaire>
The questionnaire used in this example is the same as described above with reference to FIG. This questionnaire is divided into three parts: Q1 common to all monitors, Q2 to Q5 for answering only males, and Q6 for joining all monitors.
Hereinafter, a screen displayed at a time on a browser such as a monitor PC is also referred to as a “page”. If it can be displayed by scrolling, it will be within the range of one page even if it is outside the display range of the browser. As shown in Q2 to Q5 of FIG. 3B, a plurality of questions and answer columns may be included in one page. Of course, some questionnaires display only one question per page.

  Furthermore, a combination of questions displayed on one page is referred to as a “question pattern”. This questionnaire has three question patterns, which are pattern A (Q1), pattern B (Q2 to Q5), and pattern C (Q6), respectively. The monitor that answered “1. male” to Q1 transitions to patterns A to B to C. On the other hand, the monitor that replied “2. Female” to Q1 directly replies to patterns A to C.

<Procedure for the entire questionnaire>
The overall processing flow according to the present embodiment will be described with reference to the flowchart of FIG.

(Basic survey information setting)
In step S201, the operator sets basic information for the questionnaire. That is, based on the survey details designated by the surveyor, the conditions of the questionnaire, the contents of the questions and answers, the form of aggregation and delivery, etc. are determined. The survey details are created by the surveyor determining and instructing according to the purpose of the questionnaire, or selecting from options provided by the operator. The basic information of the questionnaire is normally set by the operator of the operator operating the setting tool from the console, and is the same processing as the questionnaire totaling system according to the prior art.

The basic information set by the operator includes the number of answers (number of samples) used for tabulation. In many cases, the number or ratio of answers (oversamples) to be obtained in excess of the number used for tabulation is prepared in preparation for some problem.
The contents of the questionnaire are mainly the number of questions, the text of each question, the answer column and answer format, the combination of questions in the page (question pattern), and the page transition conditions.
In addition, the period for conducting the questionnaire, the conditions for the preliminary survey for narrowing down the respondents to the target desired by the surveyor, and the benefits given to the respondents are also determined.

(Data cleaning condition setting)
In step S202, the operator sets data cleaning conditions. In this embodiment, it is assumed that the answers to be excluded are designated by a ratio. For example, when 3% is specified, 3% of answers are excluded from the collected objects in order of increasing value indicating the answer time of the respondent. This ratio can be changed according to the request of the investigator. The higher the ratio to be excluded, the higher the quality. On the other hand, the required number of answers increases, so the cost and time also increase. The designation method based on this ratio is the same as in the prior art.

  The ratio of data cleaning to the oversample determined in step S201 and the short-time answer determined in step S202 may be determined continuously. In this case, the data cleaning target is set within the oversample. Alternatively, when the ratio to be excluded by data cleaning is determined later, it may be checked whether the number of answers to be excluded exceeds the number of answers secured as an oversample.

(Survey)
In step S203, a questionnaire is started. The questionnaire is conducted in the same way as the conventional method. In other words, recruiting monitors to participate by e-mail, etc., narrowing down the target as needed, and having them participate in this survey.

  Explain the flow when the monitor, who is a male, answers the questionnaire. The monitor connects to the operator's website from the browser of the terminal, inputs his / her information and password as necessary, logs in, and then selects a questionnaire to start answering. When the question displayed on the monitor is read and answered, the content of the answer is transmitted to the server of the operator. The server generates the next screen according to the contents of the answer and returns it to the monitor.

  More specifically, after the questionnaire is started, the screen of FIG. 3A corresponding to the pattern A is displayed. From this point, measurement of the time required for replying by Mr. A in pattern A is started. At this time, the control means operates as the measurement means of the present invention. Here, the term “response time” is used as the time required for a monitor to answer one question pattern.

  First, Mr. A selects the male radio button for Q1 asking for sex in FIG. 3A, and clicks the "Next" button. Then, the server of the operator determines that the response content is “1. male”, and generates and returns the screen of FIG. At this point, Mr. A's answer required time in pattern A is stored in the storage means, and measurement of the answer required time in pattern B is started.

  Next, read Q2 to Q5, select an answer, and click the “Next” button. Then, the screen of FIG. 3C corresponding to the pattern C is displayed. At this time, the required response time for Pattern B of Mr. A is stored in the storage means, and measurement of the required response time for Pattern C is started. After answering Q6, clicking the “Done” button completes the questionnaire for Mr. A, and the time required for Mr. A to answer in Pattern C is stored. At this point, the end process for Mr. A, for example, the provision of a privilege, the process of changing the questionnaire to the answered status, and the like are performed.

  The system continues such processing until the number of responses exceeds the predetermined value determined in the above-described preprocessing, that is, the number of responses and the number of oversamples (step S204 = YES). If the number exceeds the predetermined number, the acceptance of the questionnaire is terminated and the process proceeds to the data cleaning process. A series of processes from the start to the end of the questionnaire may be automatically performed by the system, or the operator may perform a trigger operation at an appropriate timing.

(Data cleaning execution and questionnaire aggregation)
In step S205, data cleaning is performed on the acquired answer according to the conditions. Details of this processing will be described later. Further, in step S206, the tabulating means tabulates the questionnaire and processes the questionnaire into a desired format.
By executing the above steps, the questionnaire of the requested researcher is completed.

<Data cleaning>
The flow of the data cleaning process of this embodiment will be described. The step number corresponds to the flowchart of FIG. In addition, Fig. 5 shows an example of the calculated values of the male monitor A and the female monitor B when processing is performed along this flow, and an example of the median response time required for all monitors including these two persons. is there.

In step S401, question patterns appearing in the questionnaire are classified. Although which question pattern each monitor answers dynamically changes, the question pattern that can appear is determined at the stage of the questionnaire creation, so this process can also be performed at the stage of step S201 in FIG. In this embodiment, the male monitor A responds to patterns A to B to C, and the female monitor Oto responds to patterns A and C.

  In step S402, for each question pattern, the response time required for each monitor that answers the pattern is calculated. Since the required response time is stored in the storage means at the time of conducting the questionnaire (step S203), the storage means table may be referred to using the identification information of the question pattern to be processed as a key. As a result, as shown in FIG. 5, the time required for answers from Mr. A, Mr. Oto and other monitors is obtained for each question pattern.

  In step S403, for each question pattern, the median answer required time is calculated. The group of answer required times inquired for each question pattern in the previous step is arranged in order, and the value located in the center is obtained. In the table of FIG. 5, the bottom row corresponds to this. Here, the median value, not the average value, is obtained as the representative value of the required response time because the influence of the abnormal value increases when the average value is used. As an abnormal value, for example, a long interruption of the answer due to leaving the monitor can be considered. However, although the median is most preferable, other representative values can be used. At this time, the control means operates as the representative value calculation means of the present invention.

  In step S404, for each monitor, a value obtained by dividing the time required when the monitor answers the question pattern by the median value obtained in the previous step is obtained. Hereinafter, this quotient is also referred to as “answer time index”. For example, if the median value of pattern A is 3.9 seconds and the response time required for pattern A of male monitor A is 2 seconds, the response time index is 2 ÷ 3.9 = 0.51. This process is performed for every question pattern answered by the monitor for each monitor.

  When the required response time is equal to or greater than the median value, that is, when the quotient is 1 or more, the response time index is preferably 1. For example, the time required for replying to pattern A by female monitor Otsu is 4 seconds, and the answer time index is 4 / 3.9 = 1.03, which is replaced with 1. This is to prevent the interruption of the monitor answer from affecting the determination of the exclusion target. By doing so, for example, an abnormal value due to the monitor leaving for a long time can be eliminated. Compared to typical values, short-time responses are irresponsible responses and often have quality problems, whereas long-term responses are empirically considered to have no such problems. This is because it is not necessary to consider how long it takes.

  In step S405, for each monitor, all the answer time indexes of the question patterns answered by the monitor are totaled. Then, the total value is divided by the number of pages (number of question patterns) to calculate the average value of the answer time index. At this time, the control means operates as an average value calculation means of the present invention.

In step S406, the average values of the monitors are compared. For example, Mr. A has an average value of 0.76 and Mr. Oto has an average value of 0.95. Comparing both names shows that Mr. A has a higher tendency to answer in a shorter time. Then, the answers are excluded at a ratio determined by the pre-processing in ascending order of the average value, and aggregation is performed.
By executing the above steps, a total result based on high-quality answers that have been subjected to data cleaning can be obtained.

In such a questionnaire totalization system, the response time of each monitor is standardized by being divided by the calculated representative value to be an answer time index. Therefore, it is possible to examine the tendency of the response time regardless of the length of the question sentence or the answer format.
Further, when the required response time is longer than the median (representative value), by setting the index to 1, the influence of the long-time response can be eliminated.

  Further, for each monitor, the average value is calculated by adding all the answer time indexes and then dividing by the number of question patterns. Answers are excluded in ascending order of average value. Therefore, regardless of the number of question patterns, in other words, the number of questions answered, it can be determined whether the monitor tends to answer for a short time. As a result, since the number of questions is small and the total response time is short, it is no longer necessary to exclude answers that have been determined to be low quality, and high-quality tabulation can be performed.

  In addition, since the investigator or the operator can specify the data cleaning by the ratio excluded from the aggregation as in the conventional method, it is possible to improve only the quality of the aggregation without feeling uncomfortable with the operation.

In this embodiment, a questionnaire with a relatively simple configuration has been described as an example, but the same method can be applied to a questionnaire including a more complicated question pattern configuration and screen transition. For example, even when a certain question is included in a plurality of question patterns, data cleaning can be performed by summarizing the question pattern classification and the time required for answering by the above-described method.
In addition, the questionnaire may have a number of answers called “assignment” to be secured according to the attribute of the monitor. For example, a questionnaire is conducted by determining conditions such as gender, age, and residential area, and specifying the required number of each. However, even in such a case, the present invention can be applied without any problem and data cleaning can be performed.

[Example 2]
In this example, in order to further improve the quality of the counting results in the questionnaire counting system, a method of checking the contents of the answers used for counting and excluding those judged to be inappropriate and replacing them with other answers Will be described.
The structure of the questionnaire used in the present embodiment is the same as that described above with reference to FIG. Moreover, although the structure of a questionnaire totaling system is the same as what was demonstrated above using FIG. 1, the function which excludes a specific answer from totaling is added so that it may mention later. The answer is excluded manually from the console 3 or by logic.

<Procedure for the entire questionnaire>
The overall processing flow according to the present embodiment will be described with reference to the flow of FIG. The processing unique to this embodiment starts when the questionnaire is tabulated in the first embodiment. In the figure, the processing in steps S201 to S206 is the same as that in the first embodiment, and a description thereof will be omitted.

  In step S601, a user (for example, an operator operator) uses the console 3 to visually confirm the contents of the answers used in the aggregation in step S206. Then, the answer determined to be inappropriate is excluded from the aggregation target. The criteria to be excluded can be arbitrarily determined, and can be determined in consideration of, for example, the contents of answers that are freely entered. With this process, it is possible to exclude an answer that should satisfy the quality in terms of the time required for the answer but should not be included in the aggregation when viewed from other criteria. At this time, the console operates as input means and as confirmation means of the present invention.

  In step S602, it is confirmed whether the number of answers subject to aggregation as a result of manually removing inappropriate answers in the previous step is less than a preset value. In normal processing, since it is considered that tabulation is performed based on the number of answers set in advance, S602 = YES.

In step S603, from the answers excluded in the data cleaning process in step S205, one having a relatively large average value of the answer time index is selected as an object to be counted again. This process is repeated for the number of excluded answers. If there are insufficient extra answers at this stage, the operator may be notified accordingly.

  In step S604, the counting means re-counts the questionnaire according to the surveyor's desired format using the newly selected answer and the answer that has not been manually deleted.

  In the flow of FIG. 6, the answer is excluded by visual confirmation from the console and manual deletion. However, as another method, there is a method of filtering whether or not inappropriate terms and contents are included in the response contents converted into text data. Alternatively, it is conceivable that the contents of the preliminary survey of the monitor to be investigated stored in the storage means and the answer contents of this questionnaire are compared by logic to examine whether there is a contradiction. At this time, the control means operates as confirmation means of the present invention.

  As described above, in the questionnaire totalization system according to the present embodiment, the operator confirms the answers used for the totalization, or examines it with logic, and can exclude those determined to be inappropriate. As a result, the quality of aggregation can be improved. As a result, when the number of responses used for aggregation becomes less than the predetermined number, the aggregation means selects the responses that have already been excluded and that have a relatively large average value and is considered far from the short-term responses, and re-aggregates By performing the above, it is possible to obtain a high-quality count result.

  In the above description, the present invention is specified as a questionnaire totalization system having each functional block. However, the present invention can be specified as a questionnaire totaling server having each of the above functional blocks and communicating with a monitor terminal, or can be specified as a questionnaire totaling method in which the information processing apparatus sequentially executes the above processing. . It can also be specified as a program that causes the information processing apparatus to sequentially execute the above processing.

  2: server, 21: control means, 22: communication means, 23: counting means, 24: storage means, 5: terminal

Claims (3)

A system that collects responses to questionnaires from multiple respondents via the Internet, excludes a certain percentage of responses,
The questionnaire is a unit displayed on the screen of the respondent's terminal, and is divided into a plurality of question patterns each including at least one question, and the transition order of the question patterns depends on the contents of the answers. Is determined,
The question pattern is distributed to the respondent's terminal and displayed on the screen, and communication means for obtaining an answer from the respondent;
A measuring means for measuring a response time, which is a time required for respondents to answer each question pattern;
After obtaining answers from the plurality of respondents, for each question pattern, representative value calculating means for calculating a representative value of the time required for answering all respondents who answered the question pattern;
After the representative value is calculated, based on a value obtained by dividing the answer required time of the respondent by the representative value calculated by the representative value calculating means in all question patterns answered by the respondent for each respondent. An average value calculating means for calculating an average value by determining an answer time index and dividing the sum of the answer time indexes of the respondents by the number of question patterns answered,
After the average value is calculated, the average values of the respondents are compared, the predetermined percentage of answers are excluded in ascending order of the average value, and the questionnaire is tabulated based on the answers that were not excluded. Aggregation means;
Questionnaire totalization system characterized by having. The average value calculation means is characterized in that if the value obtained by dividing the answer required time by the representative value is smaller than 1, that value is set as an answer time index, and if it is 1 or more, 1 is set as an answer time index. The questionnaire totalization system of Claim 1 to do. The aggregation means further includes a confirmation means for confirming the contents of the answers used in the aggregation after the questionnaire is aggregated, and determining whether to exclude the answers whose contents are confirmed from the aggregation,
The tabulation means, in place of the answer determined to be excluded by the user, again as the answer using the larger average value among the answers excluded by the comparison of the average value as the answer for the questionnaire The questionnaire totalization system according to claim 1 or 2, characterized by performing.

Images