National Survey of Non-profit and Voluntary Organizations (NSNVO)
Detailed information for 2003
The National Survey of Nonprofit and Voluntary Organizations is aimed at improving our understanding of nonprofit and voluntary organizations and the challenges they face in pursuing their mandates.
Data release - September 20, 2004
The National Survey of Nonprofit and Voluntary Organizations is aimed at improving our understanding of nonprofit and voluntary organizations and the challenges they face in pursuing their mandates. This survey is sponsored by the Canadian Centre for Philanthropy.
The results of the survey will provide valuable insight on the size, scope and role of nonprofit and voluntary organizations in Canada and will be used by both government and the voluntary sector to develop policy initiatives.
The survey is a component of a larger project undertaken by the Voluntary Sector Initiative to develop a body of knowledge on the nonprofit sector in Canada. The objective of the Voluntary Sector Initiative is to improve the capacity of the sector to provide services and to improve the relationship between the sector and the federal government, with input from representatives of both groups.
- Social networks and civic participation
- Society and community
Data sources and methodology
The survey population was a list frame compiled from a set of lists from different sources. The target population was set to equal the survey population, thus smaller "grassroots" organizations are excluded from the coverage of the survey. The following sources were used to build the list frame: federally registered Canadian charities that submitted a T3010 return with the CRA for the year 2002; federally registered nonprofit organizations that submitted a T1044 return with the CRA for the year 2000; provincially incorporated nonprofit organizations for all provinces and territories in Canada except Manitoba; federally incorporated nonprofit organizations identified by Industry Canada; chambers of commerce and boards of trade in Canada identified by Industry Canada; and PAF non-integrated portion units as well as locations identified as nonprofit according to either of the SNA activity code or the profit indicator of the Business Register.
In order to eliminate duplicate records from the frame, algorithms compared the names of organizations, their addresses and other identifiers common to two or more sources, e.g., the business number.
The questionnaire for Phase I of the survey was developed very quickly as its purpose was simply to collect stratification information -- specifically revenue and ICNPO major activity. Members of the Voluntary Sector Research Consortium were consulted on the screening question to ensure that the organization reached was in fact in scope for the survey and on the revenue ranges used for the survey. Subsequently, questions to help measure duplication with other frame records were added in consultation with BSMD. Refining of the questionnaire continued through the beginning of collection in September 2002.
The first draft of the questionnaire for Phase II of the survey was designed by CCP based on the results of extensive interviewers conducted with approximately 300 organizations that were part of the nonprofit and voluntary sector. Subsequently, the Voluntary Sector Research Consortium and Statistics Canada made extensive revisions to the questionnaire. Both CCP and the Voluntary Sector Research Consortium did extensive testing of the questionnaire through both focus groups and one-on-one interviews. Statistics Canada also tested both the French and English questionnaires through the Questionnaire Design Resource Centre. Refining of the questionnaire continued through the beginning of collection in March 2003.
This is a sample survey with a cross-sectional design.
The frame was divided into two parts. If an organization had stratification information (province, indicator of main activity such as NAICS or other code, revenue) available, the organization was eligible for Phase II. If not, the organization was a candidate for Phase I. The purpose of Phase I was to collect stratification information as well as the Business Number.
In Phase I, 20 000 organizations were selected having been stratified by province (known for all records) and one of six activity groups according to a keyword search of the name of the organization. Of the 9078 respondents, 1053 organizations were found to be already represented in the Phase II portion according to the Business Number. Of the rest, 6145 either had no Business Number or had Business Numbers that were different from the organizations already assigned to Phase II; these organizations were eligible for Phase II. A further 1880 organizations had indicated that they had Business Numbers but could not provide them, of which 202 were chosen for Phase II, stratified by province.
The set of 127 567 organizations, for which stratification information was already available, were divided into two parts according to whether the revenue of the organization was at least $30 000. Using the StatMx sample allocation package developed by the Generalized Systems section of BSMD, the 115 786 organizations with revenue of at least $30 000 were stratified to one of three revenue size groups for each of the 143 combinations of province and ICNPO category. On this basis, StatMx determined that 12 928 units be included in the sample, of which 1351 had "small" revenue size. The other 74 635 "small" records were joined to the 11 781 records previously set aside (all of which had "small" revenue by definition), so that a further 1200 records would be chosen from the set of records with "small" revenues". A cumulative cubed root frequency approach, allowing for rounding, led to a further 1263 records to be included. GSAM was used to select the 14 191 (12 928 + 1263) units for inclusion in the direct-to-Phase II portion of the sample. The size of the Phase II sample was therefore 20 538.
Data collection for this reference period: 2003-03-17 to 2003-06-13
Responding to this survey is voluntary.
Data are collected directly from survey respondents.
The data were obtained using a Computer Assisted Telephone Interview (CATI) application during the collection period.
View the Questionnaire(s) and reporting guide(s) .
As part of the BLAISE application used for data capture, validity edits ensured eligible values of coded answers. Consistency edits ensured that some variables were related to a function of one or more other variables using inequalities. Reported total revenue was accepted if within 10% of the sum of the variables of revenue components.
As part of post-collection processing, some data were added or changed post-collection based on a case-by-case review of the comments of respondents as well as answers to questions of the type "Other: please specify". By way of a check on the data capture application, validity edits ensured eligible values of coded responses. If applicable, deterministic imputation was used if applicable to remedy ineligible values. Otherwise, variables were flagged for imputation. Ratios were used to verify the validity of reported employment, hours worked and salary expenses. Post-collection edits were programmed in SAS.
Edit and Imputation of the revenue variables:
If no values were missing and the total value of variables Q078_GO to Q090S_PG did not equate to the value of Q077_RE, then Q077_RE was changed to agree with the total. If one value was missing from the set of variables Q078_GO to Q090S_PG and Q077_RE was greater than the total, then the difference was imputed to the missing variable. If two or more values were missing from the set, then we applied hot deck imputation to select a donor record.
Edit and Imputation of variables other than revenue:
The remaining sets of variables requiring imputation were processed according to the skip patterns within the Phase II questionnaire. Using hot deck imputation, receiver records were matched to donor records based on ICNPO_ext, revenue category (high, medium, low), and any variables considered related to the variables to be imputed, notably the value of the variable determining whether or not the variables requiring imputation were to be asked.
ICNPO after edit and imputation:
At the request of CCP, the variable ICNPO_15 was created. Major group 02 was divided into 02U (universities and colleges) and 02E (other education); the subset of hospitals in major group 03 comprised new group 03H. Seventy-eight percent of the values of ICNPO major group used for stratification were unchanged after Phase II collection.
Each organization in the sample was assigned a weight to reflect the number of organizations it represented in the population. After collection, organizations responded, did not respond or were classified otherwise (most notably out of scope). The process had to account for the fact that the nonrespondent organizations might have been in scope or out of scope, depending on the answer that would have been given to the first question of the Phase II questionnaire. By way of adjustment, the weights of respondents and of organizations classified otherwise were increased.
Weights were specific to each stratum. Units that went directly to Phase II were assigned weights inverse to the probability of their selection for each combination of province, ICNPO and revenue category to which all units had been stratified prior to Phase II collection. Units that came from Phase I already had weights associated with the first phase. The 6145 take-all organizations coming from Phase I retained their Phase I weight, adjusted after Phase II collection for nonresponse. The weights of the 202 take-some organizations were increased so as to represent all 1880 (i.e., as if they had been take-all organizations), then halved. The reason for adjusting the weights of these units was because it was unknown whether any or all of these organizations, having responded in Phase I that they had either a Business Number or charity number (but could not provide it), were represented elsewhere on the sampling frame and were therefore duplicates. Their weights were halved in the absence of further knowledge of the likelihood of any of these units already having been represented on the sampling frame.
Estimation involved several domain variables, some of which were equal to variables that were collected while others were derived. Some domain variables consisted of categories of quantitative variables (e.g., revenue, employment), the definitions of which were agreed by both the external client and STC. It was decided not to calibrate the results to those of the Satellite Account because there were differences in their scope even after reconciliation, most notably the inclusion of provincially registered nonprofit corporations in the NSNVO survey.
Domain and outcome variables were coded specifically for use with the Generalized Estimation System (GES). Using GES, coefficients of variation were generated for quantitative point estimates. Similarly, standard errors were provided for estimates of proportions and ratios. Letter grades based on the measures of quality were assigned to each estimate.
During a data validation exercise undertaken to compare financial values from the NSNVO with the results of the Satellite Account, it was found that coverage differences between the two sets of estimates resulted in significant discrepancies. Specifically, the NSNVO included several types of public sector organizations, such as school boards, public schools, public transit organizations, public libraries, etc., which had been excluded from the Satellite Account in the implementation of international concepts. It was decided that these organizations also be excluded from the NSNVO.
Almost all organizations in the frame that went directly to Phase II and not included in Phase I had a Business Number. It was possible for IEAD to verify whether these organizations were included or excluded from the Satellite Account by matching the NSNVO organizations to their set of organizations based on BN. The 3954 organizations that IEAD identified as to be excluded were removed from the Phase II portion of the frame and all sampling weights were recalculated as though the 3954 organizations had never been included. The size of the Phase II sample dropped from 20 538 to 19 163. Note that the results presented within the methodology report have already been adjusted to reflect this change.
Organizations that were included in the Phase I portion of the frame lacked one of both of revenue and a variable from which ICNPO could be derived. As these organizations were mostly from provincial lists and had no BN, IEAD could not identify organizations to be excluded from the frame using BN. However, IEAD was able to use the BN provided by respondents to Phase I to identify units to exclude. A set of keywords was also used to assist with the identification of units to be excluded. The outcome status of 113 organizations targeted for exclusion was changed to "out of scope". The sampling weights of these organizations represented 916 units on the Phase I portion of the frame that were expected to have the same status.
Because of the inclusion of provincially registered nonprofit corporations in the sampling frame, the NSNVO enabled coverage to be supplemented for certain types of organizations that were not well-represented in the data sources used to build the Satellite Account, such as sports and recreation organizations, business and professional organizations and unions. Based on the results of the NSNVO, Satellite Account estimates were adjusted to account for these coverage deficiencies.
Statistics Canada is prohibited by law from releasing any data which would divulge information obtained under the Statistics Act that relates to any identifiable person, business or organization without the prior knowledge or the consent in writing of that person, business or organization. Various confidentiality rules are applied to all data that are released or published to prevent the publication or disclosure of any information deemed confidential. If necessary, data are suppressed to prevent direct or residual disclosure of identifiable data.
The sponsor of the survey, the Canadian Centre for Philanthropy, was provided with a copy of the microdata file for the survey, per the data sharing agreement in place between CCP and Statistics Canada.
No public use microdata file is available for the National Survey of Nonprofit and Voluntary Organizations at this time.
Any estimate to which fewer than seven observations contributed was suppressed. For the most part, the only tabulations produced from the survey have been two dimensional tables, limiting the risk of disclosure. When tabulations have been produced using three dimensional tables, such as the appendices to the Cornerstones of Community report, additional care has been taken to ensure that no organizations were identifiable by their ICNPO.
Revisions and seasonal adjustment
This methodology does not apply to this survey.
The response rate was about 77%--a strong one for business surveys, in the experience of Statistics Canada.
Domain and outcome variables were coded specifically for use with the Generalized Estimation System (GES). GES is a SAS based application for producing estimates and quality indicators by domain (domains can be derived from any collected variable, as opposed to a stratification variable). Using GES, coefficients of variation (CVs) were generated for quantitative point estimates. Similarly, standard errors (SEs) were provided for estimates of proportions and ratios. Consistent with STC policy, letter grades based on the measures of quality were assigned to each estimate (please see the document below).
An unreliable estimate is subject to a high degree of sampling error. As with other sample surveys, NSNVO estimates were based on information collected from a sample of organizations. Somewhat different findings might have been obtained if all organizations in the sampling frame had been subject to the same survey method (e.g., interviewed using the same questionnaire, assignment of ICNPO and data processing). The difference between the estimates obtained from the sample and the values that would have been obtained from a census of all organizations on the frame is the sampling error.
It is standard practice to indicate the magnitude of the sampling error for estimates from a sample survey. The standard error of the estimate, derived from the survey results, is the basis for measuring the size of sampling errors. However, because of the large variety of estimates from a survey, the standard error is usually expressed relative to the estimate to which it pertains. This measure, expressed as a percentage, is known as the coefficient of variation (C.V.). It is obtained by dividing the standard error of the estimate by the estimate itself. For example, if a sample statistic of .78 has a standard error of 0.03, then the coefficient of variation of the estimate is calculated as: (.03/.78) * 100% = 3.8%.
Estimates at all levels of quality have been produced for the NSNVO, although users are forewarned not to rely on estimates that have been indicated to be used with caution.
No estimates have been produced that have been based on fewer than seven observations.