Analysis

Description:

The Canadian Census of Construction (COC) uses a complex plan for sampling small businesses (those having a gross income of less than $750,000). Stratified samples are drawn from overlapping frames. Two subsamples are selected independently from one of the samples, and more detailed information is collected on the businesses in the subsamples. There are two possible methods of estimating totals for the variables collected in the subsamples. The first approach is to determine weights based on sampling rates. A number of different weights must be used. The second approach is to impute values to the businesses included in the sample but not in the subsamples. This approach creates a complete “rectangular” sample file, and a single weight may then be used to produce estimates for the population. This “large-scale imputation” technique is presently applied for the Census of Construction. The purpose of the study is to compare the figures obtained using various estimation techniques with the estimates produced by means of large-scale imputation.

Description:

In the presence of unit nonresponse, two types of variables can sometimes be observed for units in the “intended” sample s, namely, (a) variables used to estimate the response mechanism (the response probabilities), (b) variables (here called co-variates) that explain the variable of interest, in the usual regression theory sense. This paper, based on Särndal and Swensson (1985 a, b), discusses nonresponse adjusted estimators with and without explicit involvement of co-variates. We conclude that the presence of strong co-variates in an estimator induces several favourable properties. Among other things, estimators making use of co-variates are considerably more resistant to nonresponse bias. We discuss the calculation of standard error and valid confidence intervals for estimators involving co-variates. The structure of the standard error is examined and discussed.

Description:

The procedure of subsampling the nonrespondents suggested by Hansen and Hurwitz (1946) is considered. Post-stratification prior to the subsampling is examined. For the mean of a characteristic of interest, ratio estimators suitable for different practical situations are proposed and their merits are examined. Suitable ratio estimators are also suggested for the situations in which the Hard-Core are present.

Description:

Missing survey data occur because of total nonresponse and item nonresponse. The standard way to attempt to compensate for total nonresponse is by some form of weighting adjustment, whereas item nonresponses are handled by some form of imputation. This paper reviews methods of weighting adjustment and imputation and discusses their properties.

Description:

In this paper, different types of response/nonresponse and associated measures such as rates are provided and discussed together with their implications on both estimation and administrative procedures. The missing data problems lead to inconsistent terminology related to nonresponse such as completion rates, eligibility rates, contact rates, and refusal rates, many of which can be defined in different ways. In addition, there are item nonresponse rates as well as characteristic response rates. Depending on the uses, the rates may be weighted or unweighted.

Description:

Using the optimal estimating functions for survey sampling estimation (Godambe and Thompson 1986), we obtain some optimality results for nonresponse situations in survey sampling.

Description:

Multiple imputation is a technique for handling survey nonresponse that replaces each missing value created by nonresponse by a vector of possible values that reflect uncertainty about which values to impute. A simple example and brief overview of the underlying theory are used to introduce the general procedure.

Description:

Statistics Canada has undertaken a project to develop a generalized edit and imputation system, the intent of which is to meet the processing requirements of most of its surveys. The various approaches to imputation for item non-response, which have been proposed, will be discussed. Important issues related to the implementation of these proposals into a generalized setting will also be addressed.

Description:

The analysis of survey data becomes difficult in the presence of incomplete responses. By the use of the maximum likelihood method, estimators for the parameters of interest and test statistics can be generated. In this paper the maximum likelihood estimators are given for the case where the data is considered missing at random. A method for imputing the missing values is considered along with the problem of estimating the change points in the mean. Possible extensions of the results to structured covariances and to non-randomly incomplete data are also proposed.

Description:

For periodic business surveys which are conducted on a monthly, quarterly or annual basis, the data for responding units must be edited and the data for non-responding units must be imputed. This paper reports on methods which can be used for editing and imputing data. The editing is comprised of consistency and statistical edits. The imputation is done for both total non-response and partial non-response.

Description: Release published in The Daily – Statistics Canada’s official release bulletin

Description: Release published in The Daily – Statistics Canada’s official release bulletin

Description: Release published in The Daily – Statistics Canada’s official release bulletin

Description: Release published in The Daily – Statistics Canada’s official release bulletin

Description: Release published in The Daily – Statistics Canada’s official release bulletin

Description:

This infographic denotes the process taken by Statistics Canada in collaboration with data providers and other stakeholders to review and standardize the collection of data on unfounded incidents through the Uniform Crime Reporting Survey, and the release of data to the public.

Description:

Celebrate Statistics Canada’s centennial by looking back on our journey with Canada.

Description: Release published in The Daily – Statistics Canada’s official release bulletin

Description: Release published in The Daily – Statistics Canada’s official release bulletin

Description: Release published in The Daily – Statistics Canada’s official release bulletin

Description: In this paper, we look for presence of heterogeneity in conducting impact evaluations of the Skills Development intervention delivered under the Labour Market Development Agreements. We use linked longitudinal administrative data covering a sample of Skills Development participants from 2010 to 2017. We apply a causal machine-learning estimator as in Lechner (2019) to estimate the individualized program impacts at the finest aggregation level. These granular impacts reveal the distribution of net impacts facilitating further investigation as to what works for whom. The findings suggest statistically significant improvements in labour market outcomes for participants overall and for subgroups of policy interest.

Description: Inspired by the two excellent discussions of our paper, we offer some new insights and developments into the problem of estimating participation probabilities for non-probability samples. First, we propose an improvement of the method of Chen, Li and Wu (2020), based on best linear unbiased estimation theory, that more efficiently leverages the available probability and non-probability sample data. We also develop a sample likelihood approach, similar in spirit to the method of Elliott (2009), that properly accounts for the overlap between both samples when it can be identified in at least one of the samples. We use best linear unbiased prediction theory to handle the scenario where the overlap is unknown. Interestingly, our two proposed approaches coincide in the case of unknown overlap. Then, we show that many existing methods can be obtained as a special case of a general unbiased estimating function. Finally, we conclude with some comments on nonparametric estimation of participation probabilities.

Description: We provide comparisons among three parametric methods for the estimation of participation probabilities and some brief comments on homogeneous groups and post-stratification.

Description: Beaumont, Bosa, Brennan, Charlebois and Chu (2024) propose innovative model selection approaches for estimation of participation probabilities for non-probability sample units. We focus our discussion on the choice of a likelihood and parameterization of the model, which are key for the effectiveness of the techniques developed in the paper. We consider alternative likelihood and pseudo-likelihood based methods for estimation of participation probabilities and present simulations implementing and comparing the AIC based variable selection. We demonstrate that, under important practical scenarios, the approach based on a likelihood formulated over the observed pooled non-probability and probability samples performed better than the pseudo-likelihood based alternatives. The contrast in sensitivity of the AIC criteria is especially large for small probability sample sizes and low overlap in covariates domains.

Description: Non-probability samples are being increasingly explored in National Statistical Offices as an alternative to probability samples. However, it is well known that the use of a non-probability sample alone may produce estimates with significant bias due to the unknown nature of the underlying selection mechanism. Bias reduction can be achieved by integrating data from the non-probability sample with data from a probability sample provided that both samples contain auxiliary variables in common. We focus on inverse probability weighting methods, which involve modelling the probability of participation in the non-probability sample. First, we consider the logistic model along with pseudo maximum likelihood estimation. We propose a variable selection procedure based on a modified Akaike Information Criterion (AIC) that properly accounts for the data structure and the probability sampling design. We also propose a simple rank-based method of forming homogeneous post-strata. Then, we extend the Classification and Regression Trees (CART) algorithm to this data integration scenario, while again properly accounting for the probability sampling design. A bootstrap variance estimator is proposed that reflects two sources of variability: the probability sampling design and the participation model. Our methods are illustrated using Statistics Canada’s crowdsourcing and survey data.

Description: In this rejoinder, I address the comments from the discussants, Dr. Takumi Saegusa, Dr. Jae-Kwang Kim and Ms. Yonghyun Kwon. Dr. Saegusa’s comments about the differences between the conditional exchangeability (CE) assumption for causal inferences versus the CE assumption for finite population inferences using nonprobability samples, and the distinction between design-based versus model-based approaches for finite population inference using nonprobability samples, are elaborated and clarified in the context of my paper. Subsequently, I respond to Dr. Kim and Ms. Kwon’s comprehensive framework for categorizing existing approaches for estimating propensity scores (PS) into conditional and unconditional approaches. I expand their simulation studies to vary the sampling weights, allow for misspecified PS models, and include an additional estimator, i.e., scaled adjusted logistic propensity estimator (Wang, Valliant and Li (2021), denoted by sWBS). In my simulations, it is observed that the sWBS estimator consistently outperforms or is comparable to the other estimators under the misspecified PS model. The sWBS, as well as WBS or ABS described in my paper, do not assume that the overlapped units in both the nonprobability and probability reference samples are negligible, nor do they require the identification of overlap units as needed by the estimators proposed by Dr. Kim and Ms. Kwon.

Description: In some of non-probability sample literature, the conditional exchangeability assumption is considered to be necessary for valid statistical inference. This assumption is rooted in causal inference though its potential outcome framework differs greatly from that of non-probability samples. We describe similarities and differences of two frameworks and discuss issues to consider when adopting the conditional exchangeability assumption in non-probability sample setups. We also discuss the role of finite population inference in different approaches of propensity scores and outcome regression modeling to non-probability samples.

Description: Pseudo weight construction for data integration can be understood in the two-phase sampling framework. Using the two-phase sampling framework, we discuss two approaches to the estimation of propensity scores and develop a new way to construct the propensity score function for data integration using the conditional maximum likelihood method. Results from a limited simulation study are also presented.

Description: Nonprobability samples emerge rapidly to address time-sensitive priority topics in different areas. These data are timely but subject to selection bias. To reduce selection bias, there has been wide literature in survey research investigating the use of propensity-score (PS) adjustment methods to improve the population representativeness of nonprobability samples, using probability-based survey samples as external references. Conditional exchangeability (CE) assumption is one of the key assumptions required by PS-based adjustment methods. In this paper, I first explore the validity of the CE assumption conditional on various balancing score estimates that are used in existing PS-based adjustment methods. An adaptive balancing score is proposed for unbiased estimation of population means. The population mean estimators under the three CE assumptions are evaluated via Monte Carlo simulation studies and illustrated using the NIH SARS-CoV-2 seroprevalence study to estimate the proportion of U.S. adults with COVID-19 antibodies from April 01-August 04, 2020.

Description: Our comments respond to discussion from Sen, Brick, and Elliott. We weigh the potential upside and downside of Sen’s suggestion of using machine learning to identify bogus respondents through interactions and improbable combinations of variables. We join Brick in reflecting on bogus respondents’ impact on the state of commercial nonprobability surveys. Finally, we consider Elliott’s discussion of solutions to the challenge raised in our study.

Description:

This study documents the methodological and technical challenges that are involved in performing analysis on small groups using a sample survey, oversampling, response rate, non-response rate due to language, release feasibility and sampling variability. It is based on the 1999 General Social Survey (GSS) on victimization.

Description:

Low income cut-offs (LICOs) are intended to convey the income level at which a family may be in straitened circumstances because it has to spend a greater portion of its income on the basics (food, clothing and shelter) than does the average family of similar size. The LICOs vary by family size and by size of community.

This publication provides a brief explanation of how the LICOs are derived and updated annually. In addition, it provides on a historical basis, LICOs for different family sizes by size of area of residence. LICOs are calculated based on the spending patterns of families on basic 'necessities' - food, shelter and clothing - as collected from the Survey of Household Spending (formerly referred to as the Family Expenditure Survey (FAMEX)).

Description:

This study was initiated to test the validity of probabilistic linkage methods used at Statistics Canada. It compared the results of data linkages on infant deaths in Canada with infant death data from Nova Scotia and Alberta. It also compared the availability of fetal deaths on the national and provincial files.

Description:

The second edition of Historical statistics of Canada was jointly produced by the Social Science Federation of Canada and Statistics Canada in 1983. This volume contains about 1,088 statistical tables on the social, economic and institutional conditions of Canada from the start of Confederation in 1867 to the mid-1970s. The tables are arranged in sections with an introduction explaining the content of each section, the principal sources of data for each table, and general explanatory notes regarding the statistics. In most cases, there is sufficient description of the individual series to enable the reader to use them without consulting the numerous basic sources referenced in the publication.

The electronic version of this historical publication is accessible on the Internet site of Statistics Canada as a free downloadable document: text as HTML pages and all tables as individual spreadsheets in a comma delimited format (CSV) (which allows online viewing or downloading).

Description:

The framework described here is intended as a basic operational instrument for systematic development of statistical information respecting the evolution of science and technology and its interactions with the society, the economy and the political system of which it is a part.