## 2003 Household Survey on Land：Sampling Procedure and Reliability of Estimate

### 1 Outline of Sampling Design

The 2003 Housing and Land Survey was conducted based on the stratified two-stage sampling method, with the first sampling unit of enumeration districts designated in the 2000 Population Census, and the secondary sampling unit being dwellings within the enumeration unit districts based on sampled enumeration districts.

The sampled enumeration districts were decided applying different sampling rates by region in order to achieve sufficient accuracy in the survey results for each of prefectures, urban area (all cities), rural area (all towns and villages), each of cities and of wards in 14 large cities, each of large and other metropolitan areas, and each of towns and villages having a population that meets or exceeds a certain level, respectively.

Furthermore, among the enumeration districts of the 2000 Population Census, districts where a prison or a detention house, etc. is located (enumeration districts with the suffix number 5), districts for the Self-Defense Forces (with the suffix number 6), districts for US military forces stationed in Japan (with the suffix number 7) and water-surface districts (with the suffix number 9) are excluded from the subjects of sampling.

### 2 Sampling Method of Enumeration Unit Districts

##### (1) Stratifying the enumeration districts of the Population Census

Based on the results of the 2000 Population Census, criteria were established in 24 strata (see Table 1), mainly in terms of “the number of converted households”, “distribution of households by tenure of the dwelling” and “ratio of general households having a member aged 65 years or older”.

##### (2) Sampling Process

Systematic sampling at a fixed sampling rate was conducted on every city, ward, and town or village having a population of 15,000 or more, and on every prefecture for towns and villages having a population of less than 15,000.

The sampling rate was decided as per Table 2 by size of population (see note) in order to distribute the number of samples necessary for each region of the result tables as mentioned above. The enumeration unit districts were distributed into A or B at random, so that the ratio of A to B was about 6 to 1.

##### (3) Designation of the enumeration districts where Questionnaire B is distributed

In order to achieve sufficient accuracy of results for the survey items in Questionnaire B, the enumeration districts where Questionnaire B is distributed were designated using systematic sampling at a fixed sampling rate in every prefectural capital city, ordinance-designated city, and other areas for both Kanagawa Prefecture and Fukuoka Prefecture, and every prefectural capital city and other areas for prefectures other than Kanagawa and Fukuoka.

##### (4) Sampling of Enumeration Unit Districts

In general, sampled enumeration districts are considered enumeration unit districts. However, enumeration districts having 70 or more dwellings were divided into two or more so that one district contains roughly 50 dwellings. One of the two (or more) districts is then sampled at random as an enumeration unit district.

NOTE: | The populations of municipalities used when applying the sampling rate is based on the 2000 Population Census in the administrative districts as of August 1, 2002 (if information on abolition, establishment or consolidation of municipalities from August 2, 2002, to October 1, 2003, was obtained during sampling of designated enumeration districts, such information was included). However, the result tables are expressed in accordance with the administrative districts as of October 1, 2003. |

### 3 Method of Estimate of Results

The survey results of every enumeration unit district of the 2003 Housing and Land Survey were multiplied by the reciprocal number of the sampling rate applicable to surveyed dwellings within the enumeration unit district (total number of dwellings within the enumeration unit district/number of surveyed dwellings within the enumeration unit district) or, in cases where enumeration unit districts were established by dividing a sampled enumeration district, by the number of divisions and by the reciprocal number of the sampling rate (Table 2) applicable to the sample enumeration district (includeing the enumeration unit district). They were then multiplied uniformly by the ratio so that they conform to the total population by municipality as of October 1, 2003.

In the Household Survey on Land, the aggregate estimates in Questionnaire B of the 2003 Housing and Land Survey are expressed using the following formulas for calculations targeting ordinary households.

### 4 Reliability of the Estimate

The survey results of the 2003 Housing and Land Survey are estimates obtained by sampling survey and, therefore, contain sampling errors. Thus, the results do not always coincide with results that would be obtainable in a complete survey.

Table 1 of Reference 3 at the end of this volume shows the ratio of standard error for the estimates used in this report. Here, the ratio of standard error is a guide for showing the range of existence of a value, which would be obtainable in a complete survey (hereinafter referred to as the “true value”). That is, with the estimate in the center and a range equivalent to the standard error (the estimate multiplied by the ratio of standard error) on both sides, the true value is expected to be within the entire range at 68% of probability. With a range equivalent to 2 times the standard error on both sides, the true value is now expected to be within the entire range at 95% of probability.

It should be mentioned that the ratio of standard error by major item (Table 1 of Reference 3) was calculated using the formula shown below. The ratio of standard error by size of estimate (Table 2 of Reference 3) is an evaluation made on an average basis by fitting a curve to the size of the estimate and the ratio of standard error that corresponds to this size.

The codes in the formula are as follows:

### ATTACHED TABLES

**Results for Japan**

Table 1 | Criteria and Codes for Stratification of enumeration districts (EDS) for the Population Census | ViewDownload |
---|---|---|

Table 2 | Sampling rate of sampled enumeration districts | ViewDownload |

Table 3 | Standard error rate for each main item (%) | ViewDownload |

Table 4 | Standard error rate for each estimated value (%) | ViewDownload |

**Results for Prefectures**

Table 1 | Criteria and Codes for Stratification of enumeration districts (EDS) for the Population Census | ViewDownload |
---|---|---|

Table 2 | Sampling rate of sampled enumeration districts | ViewDownload |

Table 3 | Standard error rate for each main item (%) | ViewDownload |

Table 4 | Standard error rate for each estimated value (%) | ViewDownload |

**Results for Ordinance-Designated Cities**

Table 1 | Criteria and Codes for Stratification of enumeration districts (EDS) for the Population Census | ViewDownload |
---|---|---|

Table 2 | Sampling rate of sampled enumeration districts | ViewDownload |

Table 3 | Standard error rate for each main item (%) | ViewDownload |

Table 4 | Standard error rate for each estimated value (%) | ViewDownload |

**Results for Cities with Prefectural Government**

Table 1 | Criteria and Codes for Stratification of enumeration districts (EDS) for the Population Census | ViewDownload |
---|---|---|

Table 2 | Sampling rate of sampled enumeration districts | ViewDownload |

Table 3 | Standard error rate for each main item (%) | ViewDownload |

Table 4 | Standard error rate for each estimated value (%) | ViewDownload |