JP2010267035A - Investigation system and investigation method - Google Patents

Abstract

An object of the present invention is to reduce the processing load after obtaining survey results.
A survey system identifies a covariate item that is a common attribute item among attribute items indicating attributes of a survey target person in a reference survey and a correction target survey, and based on the value of the covariate item. The first tendency score indicating the probability of participation of the survey subject in the reference survey in the correction target survey, and the second tendency score indicating the probability of participation of the survey subject in the correction target survey in the correction target survey. calculate. Further, the survey system 10 is configured so that the distribution of the second tendency score of the survey target person in the actual survey approximates the distribution of the first tendency score based on the first tendency score and the second tendency score. Select the survey target in the actual survey from the survey target in the correction target survey.
[Selection] Figure 1

Description

  The present invention relates to a survey system for selecting a survey target person in a correction target survey such as a significant extraction survey using the Internet or the like, and a survey method in the survey system.

  In recent years, many surveys using the Internet (net surveys) have been conducted. The survey using the Internet requires less manpower and costs compared to the so-called visit survey, in which the investigator visits the survey subjects randomly selected from the Basic Resident Register, the electoral list, etc. In addition to being inexpensive, it has the advantage that survey results can be obtained quickly.

  However, when comparing various surveys such as visit surveys and online surveys, there is a problem that the characteristics of the survey subjects differ depending on the method of selecting the survey subjects, and as a result, the survey results also differ. In order to solve such a problem, a method has been proposed in which the result of another survey desired to be used as a reference is estimated from the result of one survey (see, for example, Non-Patent Document 1).

Takahiro Hoshino, Tadahiko Maeda, “Application to the sample survey by significant extraction of correction method using propensity score and proposal of covariate selection method”, Statistical Mathematics Vol. 54, No. 1, pp. 191-206, Statistics Mathematical Research Institute, 2006

  However, in the method described in Non-Patent Document 1 described above, after obtaining the survey result of the survey subject in one survey (hereinafter referred to as “correction target survey”), another survey (hereinafter referred to as a reference) , Referred to as “reference survey”), the survey results are corrected. The procedure for this correction is cumbersome and therefore requires a certain amount of time for processing after obtaining the survey results. The standard survey is not limited to surveys that ensure representativeness by random sampling, but includes all surveys to be used as standards.

  In view of the above problems, an object of the present invention is to provide a survey system and a survey method capable of reducing the processing burden after acquisition of survey results.

  In order to solve the above-described problems, the present invention has the following features. First, a first feature of the present invention is a survey system (survey system 10) that selects survey subjects in a predetermined survey, and includes a first attribute item indicating attributes of survey subjects in a standard survey, A covariate item specifying unit (covariate item specifying unit 152) for specifying a covariate item that is a common attribute item with the second attribute item indicating the attribute of the survey target person in the one correction target survey, and the reference survey Based on the value of the covariate item of the survey subject in, a first tendency score indicating the probability of participation of the survey subject in the reference survey in the correction target survey is calculated, and in the second correction target survey Based on the value of the covariate item of the survey target person, a tendency score calculation unit (inclination) that calculates a second tendency score indicating the probability of participation of the survey target person in the correction target survey in the second correction target survey Based on the score calculation unit 154), the first tendency score calculated by the tendency score calculation unit, and the second tendency score, from the survey target person in the second correction target survey in the actual survey A selection section (survey target selection section 156) for selecting a survey target person, wherein the selection section has a distribution of the second tendency score corresponding to the survey target person in the actual survey as a survey in the reference survey. The gist is to select the survey subject in the actual survey so as to approximate the distribution of the first tendency score corresponding to the subject.

  Such a survey system specifies a covariate item that is a common attribute item among the attribute items indicating the attributes of the survey target person in the reference survey and the first correction target survey, and sets the value of the covariate item as a value. Based on the first tendency score indicating the probability of participation of the survey subject in the reference survey in the correction target survey and the second tendency score indicating the probability of participation of the survey subject in the second correction target survey in the correction target survey. The tendency score is calculated, and based on the first tendency score and the second tendency score, the survey subject in the actual survey is selected from the survey subjects in the second correction target survey.

  Further, in selecting the survey subject, the survey system selects the survey subject in the actual survey so that the distribution of the second tendency score of the survey subject in the actual survey approximates the distribution of the first tendency score. is doing.

  Therefore, the survey subject in the actual survey has the same quality as the sample subject in the standard survey. Therefore, a survey result equivalent to the standard survey can be obtained by conducting a survey on the survey target person in the actual survey. In addition, as long as surveys are conducted on the survey subjects in the actual survey, it is necessary to revise the survey results after obtaining the survey results each time, even if the number of surveys reaches multiple times. The processing burden is reduced.

  A second feature of the present invention relates to the first feature of the present invention, wherein the selection unit sets a selection probability of a survey subject corresponding to the second tendency score for each second tendency score. Then, based on the selection probability, the gist is to select the survey subject in the actual survey from the survey subjects in the second correction target survey.

  A third feature of the present invention relates to the second feature of the present invention, wherein the selection unit selects a survey subject corresponding to the second tendency score as the value of the second tendency score increases. The gist is to lower the probability and increase the selection probability of the survey subject corresponding to the second tendency score as the value of the second tendency score is smaller.

  A fourth feature of the present invention is a survey method in a survey system for selecting a survey subject in a predetermined survey, wherein the survey system includes a first attribute item indicating an attribute of the survey subject in a standard survey, Identifying a covariate item that is a common attribute item with the second attribute item indicating the attribute of the survey target person in the first correction target survey, and the survey system includes the survey target person in the reference survey Based on the value of the covariate item, the first tendency score indicating the probability of participation of the survey target person in the reference survey in the correction target survey is calculated, and the survey target person in the second correction target survey Calculating a second tendency score indicating a probability of participation of the survey subject in the second correction target survey to participate in the correction target survey based on the value of the covariate item; and the survey system Selecting a survey subject in a real survey from survey subjects in the second correction target survey based on the calculated first trend score and the second trend score, The selecting step is performed so that the distribution of the second tendency score corresponding to the survey target person in the actual survey approximates the distribution of the first tendency score corresponding to the survey target person in the reference survey. The gist is to select the survey subjects in the actual survey.

  A fifth feature of the present invention relates to the fourth feature of the present invention, wherein the step of selecting calculates a selection probability of a survey subject corresponding to the second tendency score for each second tendency score. The gist is to set and select a survey subject in the actual survey from survey subjects in the second correction target survey based on the selection probability.

  A sixth feature of the present invention relates to the fifth feature of the present invention, and in the step of selecting, the larger the value of the second tendency score, the more the survey subject corresponding to the second tendency score is selected. The gist is to lower the selection probability and raise the selection probability of the survey subject corresponding to the second tendency score as the value of the second tendency score is smaller.

  According to the present invention, it is possible to reduce the processing load after obtaining the survey result.

It is a block diagram of the investigation system which concerns on embodiment of this invention. It is a figure which shows the process outline | summary of the investigation system which concerns on embodiment of this invention. It is a figure which shows an example of the reference | standard investigation attribute item, the correction | amendment object investigation attribute item, and a covariate item based on embodiment of this invention. It is a figure which shows an example of the reference | standard survey subject covariate value and correction | amendment target survey subject covariate value based on embodiment of this invention. It is a flowchart which shows operation | movement of the investigation system which concerns on embodiment of this invention.

  Next, an embodiment of the present invention will be described with reference to the drawings. Specifically, (1) Configuration of the survey system, (2) Operation of the survey system, (3) Action and effect, (4) Other embodiments will be described. In the description of the drawings in the following embodiments, the same or similar parts are denoted by the same or similar reference numerals.

(1) Configuration of Survey System FIG. 1 is a configuration diagram of a survey system according to an embodiment of the present invention.

  The survey system 10 shown in FIG. 1 is a survey subject to amendment, and the survey subjects in the survey using the Internet, etc. where the survey subjects participate in the survey at will are randomly selected from the Basic Resident Register or the electoral list. Select the survey subject in the actual survey from the survey subject in the amendment target survey so that it has the same quality as the sample subject in the standard survey such as the visit survey that surveys the extracted target subjects Is.

  The survey system 10 is a personal computer, for example. The investigation system 10 includes a control unit 102, a storage unit 103, a communication unit 106, a monitor 110, and an operation unit 116.

  The control part 102 is comprised by CPU, for example, and controls the various functions with which the investigation system 10 is provided. The storage unit 103 is configured by a memory, for example, and stores various types of information used for control and the like in the survey system 10.

  The communication unit 106 communicates with another communication device (not shown).

  The monitor 110 displays an image received via the control unit 102 and displays operation details. The operation unit 116 is configured by a numeric keypad, function keys, and the like, and is an interface used for inputting user operation details.

  The control unit 102 includes a covariate item specifying unit 152, a tendency score calculating unit 154, and a survey target person selecting unit 156.

  FIG. 2 is a diagram showing an outline of processing by the control unit 102 of the investigation system 10 according to the embodiment of the present invention. As shown in FIG. 2, the processing of the control unit 102 is divided into an operation (preliminary preparation) phase and an operation (main investigation) phase.

  In the operation (preliminary preparation) phase, the covariate item specifying unit 152 in the control unit 102 inputs and stores the survey results of the reference survey and the survey results of the correction target survey in the preliminary preparation stage for each survey target person. Stored in the unit 103. For example, the covariate item specifying unit 152 inputs the survey result of the reference survey and the survey result of the correction target survey in the preliminary preparation stage transmitted from another communication device via the communication unit 106. Alternatively, the covariate item specifying unit 152 inputs the survey result of the reference survey generated according to the operation of the operation unit 116 by the operator and the survey result of the correction target survey in the preliminary preparation stage.

  The survey results of the standard survey for each survey subject and the correction target survey in the preliminary preparation stage include the identification information (subject number) of the survey subject, the attribute items of the survey subject, and the answers, respectively.

  Next, the covariate item specifying unit 152 reads the survey result of the reference survey stored in the storage unit 103 and the survey result of the correction target survey in the preliminary preparation stage, and includes attribute items (reference survey attributes) included in the reference survey result. Item) and attribute items (correction target survey attribute items) included in the survey results of the correction target survey in the preliminary preparation stage.

  FIG. 3A is a diagram illustrating an example of the reference survey attribute item. The reference survey attribute items shown in FIG. 3A include the age, sex, annual income, number of trips, and savings amount of the survey target person. On the other hand, FIG. 3B is a diagram illustrating an example of the correction target survey attribute item. The correction target survey attribute items shown in FIG. 3B include the age, sex, annual income, number of trips, and the presence or absence of a driver's license of the survey target person.

  Next, the covariate item specifying unit 152 specifies an attribute item common to the extracted reference survey attribute item and the correction target survey attribute item as a covariate item candidate. When the reference survey attribute item is as shown in FIG. 3A and the correction target survey attribute item is as shown in FIG. 3B, the covariate item specifying unit 152 determines the age, sex, annual income, and number of trips. Is selected as a covariate item candidate.

  Further, the covariate item specifying unit 152 narrows down the covariate item candidates by a well-known method (see Non-Patent Document 1) and sets them as covariate items. For example, the following method is adopted.

  As a first step, the covariate item specifying unit 152 selects a covariate item candidate that has little change for each survey within the individual and can be continuously acquired in each of the reference survey and the correction target survey. To do.

  Next, as a second stage, the covariate item specifying unit 152 selects a covariate item candidate selected in the first stage that has a significant difference between the reference survey and the correction target survey. The significant difference is obtained by, for example, a t test. The t-test is a result of the reference survey (ratio%) corresponding to a predetermined covariate item candidate value (for example, age 20) and a correction target survey corresponding to the predetermined covariate item candidate value. This is a method of calculating a statistically significant difference from the result (ratio%) as a significant difference.

  Next, as a third step, the covariate item specifying unit 152 sets an item (answer) that needs to be corrected because there is a difference between the reference survey and the correction target survey, and goes through the first step and the second step. Regression analysis is performed using the selected covariate item candidate as an explanatory variable, and a standard partial regression coefficient in each of the reference survey and the correction target survey is calculated. Further, the covariate item specifying unit 152 selects a covariate item candidate for an item having a standard regression coefficient in the same direction (positive if positive, negative if negative) and a large absolute value in the reference survey and the correction target survey. To do.

  Next, as the fourth stage, the covariate item specifying unit 152 further calculates the sum of square errors from the covariate item candidates selected through the first to third stages (the response of the reference survey and the correction target survey). In order to reduce the sum of squares of differences, for example, the candidate items of the covariate candidates are narrowed down by using a method such as stepwise or cross-validation, and the final covariate items are obtained. The covariate item specifying unit 152 causes the storage unit 103 to store the identification information of the covariate item specified in this way.

  FIG. 3C is a diagram illustrating an example of a covariate item. The standard survey attribute item shown in FIG. 3 (a) and the correction target survey attribute item shown in FIG. 3 (b) share the same age, sex, annual income, and number of trips. Is selected as a covariate item candidate. Further, the covariate item specifying unit 152 passes through the above-described first to fourth stages, and as shown in FIG. 3C, the covariate item specifying unit 152 shares attribute items other than gender among the age, gender, annual income, and number of trips. Identified as a variable item.

  After the operation (preparation) phase described above, the operation (main survey) phase is entered. In the operation (main survey) phase, the tendency score calculation unit 154 inputs the values of the covariate items of the reference survey and the correction target survey for each survey target person and causes the storage unit 103 to store them. Note that the correction target survey in the operation (main survey) phase may be the same survey as the correction target survey in the preliminary preparation phase in the operation (preliminary preparation) phase, or may be a different survey.

  For example, the tendency score calculation unit 154 inputs the values of the covariate items of the reference survey and the correction target survey transmitted from other communication devices via the communication unit 106. Or the tendency score calculation part 154 inputs the value of the covariate item of the reference | standard investigation and correction | amendment object investigation produced | generated according to operation of the operation part 116 of an operator. The covariate items include the survey subject's attribute items and responses.

  Next, the propensity score calculation unit 154 reads the identification information of the covariate item stored in the storage unit 103. Furthermore, the tendency score calculation unit 154 reads the result of the reference survey for each survey target person stored in the storage unit 103, and extracts the target number and the value of the covariate item included in the result of the standard survey. FIG. 4A is a diagram illustrating an example of values of covariate items for the survey target person of the standard survey.

  Similarly, the tendency score calculation unit 154 reads the survey result of the correction target survey for each survey target person stored in the storage unit 103, and includes the subject No. and the covariate item included in the survey result of the correction target survey. Extract the value. FIG. 4B is a diagram illustrating an example of covariate item values for the survey target person in the correction target survey.

  Furthermore, the tendency score calculation unit 154 calculates a first tendency score indicating the probability of participation of the survey target person in the reference survey in the correction target survey, and also indicates whether the survey target person in the correction target survey participates in the correction target survey. A second tendency score indicating the probability is calculated.

  Specifically, the tendency score calculation unit 154 uses the value of the covariate item for the reference survey and the correction target survey for each survey target person as an explanatory variable, and includes identification information (here, 0) indicating that the survey is a standard survey, The identification information (in this case, 1) indicating that the survey is subject to correction is the objective variable, logistic regression analysis, etc. is performed, and the first tendency score corresponding to each survey subject in the standard survey and each subject in the correction subject survey A second tendency score corresponding to is calculated. The first tendency score and the second tendency score are any values between 0 and 1.

  Further, the tendency score calculation unit 154 associates each first tendency score with the corresponding subject No. of the survey target person and stores it in the storage unit 103, and stores each second tendency score in the corresponding survey target person's subject. The subject No is associated and stored in the storage unit 103.

  Next, the tendency score calculation unit 154 specifies the distribution of the first tendency score corresponding to each survey target person of the reference survey and the distribution of the second tendency score corresponding to each research target person of the correction target survey. The reference survey and the correction target survey are different in the method of selecting the survey target person, and the distribution of the first tendency score and the distribution of the second tendency score are different.

  Based on the distribution of the first tendency score and the second tendency score calculated by the tendency score calculation unit 154, the survey target person selecting unit 156 selects the survey target person in the actual survey from the survey target persons in the correction target survey.

  At this time, the survey target person selecting unit 156 causes the distribution of the second tendency score corresponding to each survey target person in the actual survey to approximate the distribution of the first tendency score corresponding to each survey target person in the reference survey. Select the survey target in the actual survey from the survey target in the correction target survey.

  Specifically, the survey target person selecting unit 156 calculates the selection probability of the survey target person for each survey target person corresponding to the second tendency score. Here, the survey subject with a large second tendency score value has a high probability of participating in the correction target survey, and the survey subject with a small second tendency score value has a low probability of participating in the correction target survey. On the other hand, it is generally considered that the survey target person in the standard survey has a low probability of participating in the correction target survey.

  Therefore, the survey target person selecting unit 156 reduces the selection probability of the survey target person corresponding to the second tendency score, while reducing the selection probability of the survey target person having the large second tendency score, while the survey target person having a small second tendency score. Increase the selection probability.

  Next, based on the selection probability calculated for each survey target person in the correction target survey, the survey target person selecting unit 156 selects a survey target person in the actual survey from the survey target persons in the correction target survey. Further, the survey target person selecting unit 156 stores the target person No of the selected survey target person in the storage unit 103.

  Thereafter, the actual survey is performed on the survey subject corresponding to the subject No. stored in the storage unit 103.

(2) Operation of Survey System Next, the operation of the survey system 10 will be described. FIG. 5 is a flowchart showing the operation of the survey system 10.

  In step S <b> 101, the covariate item specifying unit 152 in the control unit 102 inputs the result of the reference survey and stores it in the storage unit 103. In step S <b> 102, the covariate item specifying unit 152 inputs the survey result of the correction target survey in the preliminary preparation stage, and stores it in the storage unit 103.

  In step S103, the covariate item specifying unit 152 reads the reference survey result and the survey result of the correction target survey stored in the storage unit 103, the attribute items included in the reference survey result, and the preliminary preparation. Covariate items are specified based on attribute items included in the survey results of the correction target survey at the stage. Further, the covariate item specifying unit 152 causes the storage unit 103 to store identification information of the specified covariate item.

  In step S104, the tendency score calculation unit 154 indicates the probability of participation of the survey target person in the reference survey in the correction target survey based on the value of the covariate item included in the results of the standard survey and the correction target survey. The tendency score and the second tendency score indicating the probability of participation of the survey target person in the correction target survey in the correction target survey are calculated and stored in the storage unit 103.

  In step S <b> 105, the tendency score calculation unit 154 specifies the distribution of the first tendency score corresponding to each survey target person of the reference survey and the distribution of the second tendency score corresponding to each research target person of the correction target survey. .

  In step S106, the survey target person selecting unit 156 approximates the distribution of the second tendency score corresponding to each survey target person in the actual survey to the distribution of the first tendency score corresponding to each survey target person in the standard survey. In addition, the survey subjects in the actual survey are selected from the survey subjects in the amendment survey.

(3) Action / Effect The survey system 10 according to the embodiment of the present invention specifies a covariate item that is a common attribute item among the attribute items indicating the attributes of the survey target person in the reference survey and the correction target survey. , Based on the value of the covariate item, a first tendency score indicating the probability of participation of the survey target person in the reference survey in the correction target survey, and the participation tendency of the survey target person in the correction target survey A second tendency score indicating the probability is calculated, and based on the first tendency score and the second tendency score, the survey subject in the actual survey is selected from the survey subjects in the correction target survey. Note that the covariate items specified here have a wide application range and can be used continuously.

  Further, in selecting the survey subject, the survey system 10 determines that the survey subject in the actual survey so that the distribution of the second tendency score of the survey subject in the correction target survey approximates the distribution of the first trend score. Is selected.

  Therefore, the survey subject in the actual survey has the same quality as the sample subject in the standard survey. Therefore, a survey result equivalent to the standard survey can be obtained by conducting a survey on the survey target person in the actual survey. In addition, as long as the survey is conducted for the selected survey subjects, it is necessary to revisit the survey results after obtaining the survey results each time, even if the number of surveys reaches multiple times. And the processing burden is reduced.

(4) Other Embodiments As described above, the present invention has been described according to the embodiment. However, it should not be understood that the description and drawings constituting a part of this disclosure limit the present invention. From this disclosure, various alternative embodiments, examples and operational techniques will be apparent to those skilled in the art.

  In the embodiment described above, the survey target person selecting unit 156 calculates the selection probability of the survey target person for each survey target person corresponding to the second tendency score, and based on the selection probability, the survey target in the correction target survey The survey subject in the actual survey is selected from the researcher, and the selection probability of this survey subject is that the distribution of the second tendency score of the survey subject in the actual survey is the first tendency score of the survey subject in the reference survey It may be set so as to approximate the distribution of.

  In addition to online surveys, for example, group interviews, so-called street catch surveys with passersby as survey subjects, and other survey data taken completely separately The present invention can be applied.

  Thus, it should be understood that the present invention includes various embodiments and the like not described herein. Therefore, the present invention is limited only by the invention specifying matters in the scope of claims reasonable from this disclosure.

  INDUSTRIAL APPLICABILITY The investigation system and the investigation method of the present invention can reduce the processing load after obtaining the investigation result, and are useful as an investigation system and an investigation method.

  DESCRIPTION OF SYMBOLS 10 ... Survey system, 102 ... Control part, 103 ... Memory | storage part, 106 ... Communication part, 110 ... Monitor, 116 ... Operation part, 152 ... Covariate item specific | specification part, 154 ... Trend score calculation part, 156 ... Investigation subject selection Part

Claims (6)

A survey system for selecting survey subjects in a predetermined survey,
A covariate item that is a common attribute item of the first attribute item indicating the attribute of the survey target person in the reference survey and the second attribute item indicating the attribute of the survey target person in the first correction target survey is specified. A covariate item identification part;
Based on the value of the covariate item of the survey target person in the standard survey, a first tendency score indicating the probability of participation of the survey target person in the standard survey in the correction target survey is calculated and a second correction Trend score calculation for calculating a second tendency score indicating a probability of participation of the survey subject in the second correction target survey in the correction target survey based on the value of the covariate item of the survey target in the target survey And
Selection for selecting a survey subject in a real survey from survey subjects in the second correction target survey based on the first trend score and the second trend score calculated by the trend score calculation unit With
The selection unit is configured so that the distribution of the second tendency score corresponding to the survey target person in the actual survey approximates the distribution of the first tendency score corresponding to the survey target person in the reference survey. A survey system that selects survey subjects for the actual survey.   The selection unit sets, for each second tendency score, a selection probability of a survey target person corresponding to the second tendency score, and based on the selection probability, a survey target in the second correction target survey The survey system according to claim 1, wherein a survey target person in the actual survey is selected from a person.   The selection unit decreases the selection probability of the survey subject corresponding to the second tendency score as the value of the second tendency score increases, and the value of the second tendency score decreases as the value of the second tendency score decreases. The survey system according to claim 2, wherein the selection probability of the survey subject corresponding to the tendency score is increased. A survey method in a survey system for selecting survey subjects in a predetermined survey,
The survey system is a common attribute item that is a common attribute item of the first attribute item indicating the attribute of the survey target person in the standard survey and the second attribute item indicating the attribute of the survey target person in the first correction target survey. Identifying a variable item;
The survey system calculates a first tendency score indicating the probability of participation of the survey subject in the standard survey in the correction target survey based on the value of the covariate item of the survey subject in the standard survey. Based on the value of the covariate item of the survey subject in the second correction target survey, a second tendency score indicating the probability of participation of the survey subject in the second correction target survey in the correction target survey is calculated. A calculating step;
The survey system selects a survey subject in a real survey from survey subjects in the second correction target survey based on the calculated first tendency score and the second tendency score; With
In the selecting step, the distribution of the second tendency score corresponding to the survey subject in the actual survey is approximated to the distribution of the first tendency score corresponding to the survey subject in the reference survey, A survey method for selecting survey subjects in the actual survey.   The selecting step sets, for each second tendency score, a selection probability of a survey subject corresponding to the second tendency score, and based on the selection probability, a survey in the second correction target survey The survey method according to claim 4, wherein a survey target person in the actual survey is selected from target persons.   In the selecting step, as the value of the second tendency score is larger, the selection probability of the survey subject corresponding to the second tendency score is lowered, and as the value of the second tendency score is smaller, the 2 The survey method according to claim 5, wherein the probability of selecting the survey subject corresponding to the tendency score is increased.

Images