Survey on Financing of Co-operatives

Detailed information for 2007




One Time

Record number:


The objective of this survey is to provide data on the challenges and issues that co-operatives in Canada face in securing financing, as well as other data on the characteristics of these organizations.

Data release - December 3, 2008


This survey gathers information about the amount and types of financing available to co-operatives, as well as the challenges co-operatives face in obtaining this financing. This will allow Agriculture and Agri-Food Canada and Human Resources and Social Development Canada to assess the state of the co-operative sector in Canada and its needs in terms of government policies or programs.


  • Business performance and ownership

Data sources and methodology

Target population

The frame provided to us from Agriculture and Agri-Food Canada included all non-financial co-operatives operating in Canada, with the exception of housing co-operatives and feeder financing co-operatives. Housing co-operatives were excluded from the survey because housing co-operatives are quite different from other types of co-operatives, to the extent that including them in the survey would have skewed the results. Similarly, feeder financing co-operatives exist primarily to provide financing to other co-operatives, and as such including them in the survey would have skewed the results.

Instrument design

The questionnaire used for the Survey on Financing of Co-operatives, 2007 was based on the questionnaire used for the Survey on Financing of Small and Medium Enterprises, 2007. The questionnaire content was developed jointly by AAFC, HRSDC and SC. The Survey on Financing of Co-operatives, 2007 was conducted using a computer-assisted telephone interview (CATI).

The CATI application was developed by specialists, and tested by both subject matter and operational staff.


This is a sample survey with a cross-sectional design.

The sample was designed to provide estimates of population proportions for the following domains of interest:

. Estimates by Region with a +/- 8% margin of error
. Estimates by Co-operative Subgroup with a +/- 8% margin of error, and
. Estimates by Co-operative Type with a +/- 8% margin of error.

In order to satisfy these requirements, a total sample size of 1,400 drawn under a stratified simple random sampling design was determined to be sufficient (taking into account a response rate of 60%). It is important to note that this sample size is calculated based only on one-dimensional domains and would not guarantee reliable estimates at more detailed levels or for any other combinations of the above domains (e.g. Province by Co-op Subgroup). In the end, the client provided funding for a total of 1,800 units which was the size of the sample selected.

Data sources

Data collection for this reference period: 2008-02-11 to 2008-03-31

Responding to this survey is voluntary.

Data are collected directly from survey respondents.

Information for this survey is collected using a computer-assisted telephone interview (CATI). The CATI interview consists of a questionnaire that will last on average no more than fifteen (15) minutes.

The CATI system developed for this survey includes on-line edits to maximize data quality and allow interviewers to correct any problems with the information provided by the respondent during the initial interview. Such edits ensure that, for example, questions with multiple response categories that only demand one response receive only one response.

A maximum of five (5) attempts will be made to conduct an interview with each co-operative in the sample.

Collection is undertaken by the Operations and Integration Division (OID) of Statistics Canada.

View the Questionnaire(s) and reporting guide(s) .

Error detection

The processing phase of the survey was for the most part concerned with applying consistency edits and validity edits to the data reported. Consistency edits ensure that data reported in one question does not contradict information reported in another question. Validity edits ensure that the data reported is valid (i.e. that percentage values reported do not exceed 100%, that values that are supposed to sum up do in fact sum up, that skip patterns are followed, etc.).

If edits failed, the data for the co-operative in question was examined to determine whether the edit could be manually resolved. In instances where the edit failure could be resolved, the data in question were adjusted accordingly. Otherwise, the data in question were flagged for imputation.

Many of the questions on the survey included "Other (Please Specify)" as a response category, where the respondent could indicate a response other than those previously listed. These responses were reviewed and where possible were recoded to a previous category (i.e. if the respondent had listed "Computer network" as an "another purpose" in Question A.7, it was recoded to "Computer hardware and software"). In many cases, however, the "other" responses were clearly different than the listed responses. In a few cases, there were enough similar distinct "other" responses to allow for the creation of an additional response category.


Imputation was done to complete data for records that had given erroneous responses or only partial responses for the questionnaire.

Nearest neighbour donor imputation was used to correct partial non-response present in the completed questionnaires. This imputation method consists of replacing one or more missing values from a respondent, called receiver, by values provided by one or more respondents, called donors. Nearest donor(s) are determined based on size (revenue and number of employees closest to the receiver's) and other characteristics (co-operative type and subgroup identical to those of the receiver). The values selected for imputation are first verified to pass pre-established edit rules before being inserted into the recipient record. Statistics Canada's generalized system Banff was used to implement imputation.

In general the imputation rates are fairly low. All imputation rates except one are below 30%.


Statistics Canada's Generalized Estimation System (GES) was used for producing estimates. Initial sample weights were adjusted to account for refusals and other cases of complete non-response.

Estimates were produced for 71 domains of interest defined based on stratification variables (e.g. region, industry) as well questionnaire variables such as number of employees and characteristics of the co-operative.

Disclosure control

Statistics Canada is prohibited by law from releasing any data that 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.

Data Sharing Agreement

Statistics Canada has entered into a data sharing agreement under Section 12 of the Statistics Act with the Co-operatives Secretariat of Agriculture and Agri-food Canada for the sharing of information from this survey. The Co-operatives Secretariat of Agriculture and Agri-food Canada will keep the information confidential and use it only for statistical purposes.

The microdata file to be provided to the Co-operatives Secretariat of Agriculture and Agri-food Canada per the data sharing agreement that exists for this project had all identifiers, such as name of business removed.

In our output tables, any cell which had less than 6 respondents was suppressed. Additionally, smaller industries were collapsed into a residual category at the provincial level, such that for most industries, only national estimates were produced.

Data accuracy

Overall, the survey response rate was 66.4%, which is an acceptable response rate for Statistics Canada business surveys.

Data response error may be due to questionnaire design, the characteristics of a question, inability of the respondent to provide correct information, misinterpretation of the questions or definitional problems. These errors are controlled through careful questionnaire design and the use of simple concepts and consistency checks.

For this survey, non-responding records were dealt with by adjusting the weights assigned to the responding records, such that one responding record might also represent other non-responding units with similar characteristics (i.e. size, province, industry) as the responding record.

Processing error may occur at various stages of processing such as data entry, editing and tabulation. Measures have been taken to minimize these errors.

Date modified: