Keyword search
Results
All (13)
All (13) (0 to 10 of 13 results)
- Articles and reports: 12-001-X197800254832Description: I.P. Fellegi and D. Holt proposed a systematic approach to automatic edit and imputation. An implementation of this proposal was a Generalized Edit and Imputation System by the Hot-Deck Approach, that was utilized in the edit and imputation of the 1976 Canadian Census of Population and Housing. This paper discusses that application, evaluating the strengths and weaknesses of the methodology with some empirical evidence. The system will be considered in relation to the general issues of the edit and imputation of survey data. Some directions for future developments will also be considered.Release date: 1978-12-15
- 2. Large scale imputation of survey data ArchivedArticles and reports: 12-001-X197800254833Description: Owners of small businesses complain about the quantity of forms they are required to collectors of statistics. Administrative data are an alternative source but do not usually include all the information required by the survey takers.The “Tax Data Imputation System” makes use of tax data collected from a large number of businesses by Revenue Canada and data obtained by sample survey for a small subset of these businesses. Survey data is imputed (estimated) for all the businesses not actually surveyed using a “hot-deck” technique, with adjustments made to ensure certain edit rules are satisfied. The results of a simulation study suggest that this procedure has reasonable statistical properties. Estimators (of means or totals) are unbiased with variances of comparable size to the corresponding ratio estimators.Release date: 1978-12-15
- Articles and reports: 12-001-X197800254834Description: Frames designed for continuous surveys are sometimes used for ad hoc surveys which require selection of sampling units separate from those selected for the continuous survey. This paper presents an unbiased extension of Keyfitz’s (1951) sample updating method to the case where a portion of the frame has been reserved for surveys other than the main continuous survey. A simple although biased alternative is presented. The scope under Platek and Singh’s (1975) design strategy for an area based continuous survey requiring updating is then expanded to encompass rotation of first stage units, establishment of a separate special survey sub-frame, and procedures to prevent re-selection of ultimate sampling units. The methods are evaluated in a Monte Carlo study using Census data to simulate the design for the Canadian Labour Force Survey.Release date: 1978-12-15
- 4. Alternative estimators in PPS sampling ArchivedArticles and reports: 12-001-X197800254835Description: Some estimators alternative to the usual PPS estimator are suggested in this paper for situations where the size measure used for PPS sampling is not correlated with the study variable and where data are available on another supplementary variable (size measure). Properties of these estimators are studied under super-population models and also empirically.Release date: 1978-12-15
- 5. Non-response and imputation ArchivedArticles and reports: 12-001-X197800254830Description:
The problems of dealing with non-response at various stages of survey planning are discussed with implications for the mean square error, practicality and possible advantages and disadvantages. Conceptual issues of editing and imputation are also considered with regard to complexity and levels of imputation. The methods of imputation include weighting, duplication, and substitution of historical records. The paper includes some methodology on the bias and variance.
Release date: 1978-12-15 - 6. The evolution of a national statistical agency ArchivedArticles and reports: 12-001-X197800254831Description:
A chronological account of the development of Canada’s central statistical agency is presented in this paper with emphasis on the importance of adapting the organization to the demands of the times.
Release date: 1978-12-15 - Articles and reports: 12-001-X197800154830Description: When a two-stage sample is used to collect data, the correlations between the sampled units make the ?2 test of independence invalid. Use of the ordinary ?2 tables generally results in a test which is greater than the desired level of significance. The effect of the sample design comes from two main areas: the sample size within PSU’s and the degree to which the characteristics are independent within each PSU. The effect of the sample size within PSU’s is greatest when there is no independence within each PSU, and diminishes as the degree of independence increases.Release date: 1978-06-15
- 8. Approximate tests of independence and goodness of fit based on stratified multi-stage samples ArchivedArticles and reports: 12-001-X197800154831Description: The impact on linear statistics of the sample design used in obtaining survey data is the subject of much of sampling literature. Recently, more attention has been paid to the design’s impact on non-linear statistics; the major factor inhibiting these investigations has been the problem of estimating at least the first two moments of such statistics. The present article examines the problem of estimating the variances of non-linear statistics from complex samples, in the light of existing literature. The behaviour of the chi-square statistic computed from a complex sample to test hypotheses of goodness of fit or independence is studied. Alternative tests are developed and their properties studied in simulation experiments.Release date: 1978-06-15
- Articles and reports: 12-001-X197800154832Description: This paper describes a survey design established to measure truck commodity flows in Peru. The article addresses the conceptual and operational features of the survey design as well as describing its elements and implementation techniques in the context of a pilot project. Finally, the paper illustrates how the results of this pilot might be used to design and implement a full-scale national survey.Release date: 1978-06-15
- 10. A comparison of correlated response variance estimates obtained in the 1961, 1971 and 1976 Censuses ArchivedArticles and reports: 12-001-X197800154833Description: The total variance of a survey estimate incorporates sampling variance, simple response variance and correlated response variance. The last component reflects the part of the total variance due to a common influence on a group of respondents. In the Canadian census, self-enumeration was adopted as the standard method of enumeration in the 1971 Census. One factor in favor of introducing this method was evidence, from the 1961 Census, that correlated response variance made an important contribution to the total variance of census estimates. Based on a study conducted using interpenetration of interviewers, this article compares correlated response variances from the 1961, 1971 and 1976 Censuses. The empirical results demonstrate that although the self-enumeration adopted in the 1971 Census did not completely remove the correlated response variance, this approach has considerably reduced the magnitude of this component of variance for almost all the characteristics examined.Release date: 1978-06-15
Data (0)
Data (0) (0 results)
No content available at this time.
Analysis (13)
Analysis (13) (0 to 10 of 13 results)
- Articles and reports: 12-001-X197800254832Description: I.P. Fellegi and D. Holt proposed a systematic approach to automatic edit and imputation. An implementation of this proposal was a Generalized Edit and Imputation System by the Hot-Deck Approach, that was utilized in the edit and imputation of the 1976 Canadian Census of Population and Housing. This paper discusses that application, evaluating the strengths and weaknesses of the methodology with some empirical evidence. The system will be considered in relation to the general issues of the edit and imputation of survey data. Some directions for future developments will also be considered.Release date: 1978-12-15
- 2. Large scale imputation of survey data ArchivedArticles and reports: 12-001-X197800254833Description: Owners of small businesses complain about the quantity of forms they are required to collectors of statistics. Administrative data are an alternative source but do not usually include all the information required by the survey takers.The “Tax Data Imputation System” makes use of tax data collected from a large number of businesses by Revenue Canada and data obtained by sample survey for a small subset of these businesses. Survey data is imputed (estimated) for all the businesses not actually surveyed using a “hot-deck” technique, with adjustments made to ensure certain edit rules are satisfied. The results of a simulation study suggest that this procedure has reasonable statistical properties. Estimators (of means or totals) are unbiased with variances of comparable size to the corresponding ratio estimators.Release date: 1978-12-15
- Articles and reports: 12-001-X197800254834Description: Frames designed for continuous surveys are sometimes used for ad hoc surveys which require selection of sampling units separate from those selected for the continuous survey. This paper presents an unbiased extension of Keyfitz’s (1951) sample updating method to the case where a portion of the frame has been reserved for surveys other than the main continuous survey. A simple although biased alternative is presented. The scope under Platek and Singh’s (1975) design strategy for an area based continuous survey requiring updating is then expanded to encompass rotation of first stage units, establishment of a separate special survey sub-frame, and procedures to prevent re-selection of ultimate sampling units. The methods are evaluated in a Monte Carlo study using Census data to simulate the design for the Canadian Labour Force Survey.Release date: 1978-12-15
- 4. Alternative estimators in PPS sampling ArchivedArticles and reports: 12-001-X197800254835Description: Some estimators alternative to the usual PPS estimator are suggested in this paper for situations where the size measure used for PPS sampling is not correlated with the study variable and where data are available on another supplementary variable (size measure). Properties of these estimators are studied under super-population models and also empirically.Release date: 1978-12-15
- 5. Non-response and imputation ArchivedArticles and reports: 12-001-X197800254830Description:
The problems of dealing with non-response at various stages of survey planning are discussed with implications for the mean square error, practicality and possible advantages and disadvantages. Conceptual issues of editing and imputation are also considered with regard to complexity and levels of imputation. The methods of imputation include weighting, duplication, and substitution of historical records. The paper includes some methodology on the bias and variance.
Release date: 1978-12-15 - 6. The evolution of a national statistical agency ArchivedArticles and reports: 12-001-X197800254831Description:
A chronological account of the development of Canada’s central statistical agency is presented in this paper with emphasis on the importance of adapting the organization to the demands of the times.
Release date: 1978-12-15 - Articles and reports: 12-001-X197800154830Description: When a two-stage sample is used to collect data, the correlations between the sampled units make the ?2 test of independence invalid. Use of the ordinary ?2 tables generally results in a test which is greater than the desired level of significance. The effect of the sample design comes from two main areas: the sample size within PSU’s and the degree to which the characteristics are independent within each PSU. The effect of the sample size within PSU’s is greatest when there is no independence within each PSU, and diminishes as the degree of independence increases.Release date: 1978-06-15
- 8. Approximate tests of independence and goodness of fit based on stratified multi-stage samples ArchivedArticles and reports: 12-001-X197800154831Description: The impact on linear statistics of the sample design used in obtaining survey data is the subject of much of sampling literature. Recently, more attention has been paid to the design’s impact on non-linear statistics; the major factor inhibiting these investigations has been the problem of estimating at least the first two moments of such statistics. The present article examines the problem of estimating the variances of non-linear statistics from complex samples, in the light of existing literature. The behaviour of the chi-square statistic computed from a complex sample to test hypotheses of goodness of fit or independence is studied. Alternative tests are developed and their properties studied in simulation experiments.Release date: 1978-06-15
- Articles and reports: 12-001-X197800154832Description: This paper describes a survey design established to measure truck commodity flows in Peru. The article addresses the conceptual and operational features of the survey design as well as describing its elements and implementation techniques in the context of a pilot project. Finally, the paper illustrates how the results of this pilot might be used to design and implement a full-scale national survey.Release date: 1978-06-15
- 10. A comparison of correlated response variance estimates obtained in the 1961, 1971 and 1976 Censuses ArchivedArticles and reports: 12-001-X197800154833Description: The total variance of a survey estimate incorporates sampling variance, simple response variance and correlated response variance. The last component reflects the part of the total variance due to a common influence on a group of respondents. In the Canadian census, self-enumeration was adopted as the standard method of enumeration in the 1971 Census. One factor in favor of introducing this method was evidence, from the 1961 Census, that correlated response variance made an important contribution to the total variance of census estimates. Based on a study conducted using interpenetration of interviewers, this article compares correlated response variances from the 1961, 1971 and 1976 Censuses. The empirical results demonstrate that although the self-enumeration adopted in the 1971 Census did not completely remove the correlated response variance, this approach has considerably reduced the magnitude of this component of variance for almost all the characteristics examined.Release date: 1978-06-15
Reference (0)
Reference (0) (0 results)
No content available at this time.
- Date modified: