Data quality, concepts and methodology: Concepts, methods and sources

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

The Cereals and Oilseeds Review is designed as a current source of grain marketing data on the major grains produced in Canada. Some of the data in this publication are also available on CANSIM, Statistics Canada's machine-readable database and retrieval system.

The majority of the data used in this publication are administrative in nature and are obtained from other areas within Statistics Canada or from other federal, provincial or international agencies. Most of the data for the major data series were produced with statistical purposes in mind; however, some were designed for program administration with statistical needs as a secondary objective. Every effort is made to ensure that administrative data are conceptually correct for the use to which they are put.

Much of the data obtained from administrative sources have been summarized from the financial transactions of individuals or companies. These summarized data are often subject to audit by independent professional accountants and/or are used to make payments to individuals. As a result, the quality of these data is considered to be good. The survey data used reflect typical Statistics Canada standards for quality assurance and, therefore, the quality of these data is also considered to be good.

However, it is important to note that both the administrative and survey data are subject to error. Administrative data may contain non-sampling error such as keying mistakes, while survey data may suffer from both non-sampling and sampling error. Users should also note that the quality of individual estimates may not be consistent between commodities or between provinces because the data sources and their quality may vary.

Information on inter-provincial movement of grain is limited. Data on inter-provincial canola movement are obtained from the Report of Crushing Operations. Data on inter-provincial purchases of feed grains by feed mills are obtained from the Feed Grain Purchases Survey. Both surveys are conducted by the Grain Marketing Unit of Statistics Canada.

The following text will discuss the most important grain marketing analysis tool - the supply and disposition tables and their components. Sections on Trade data, International data, Price data, Survey Estimates and the Revision Policies are also included.

Supply-disposition tables (S&Ds)

Overview

S&Ds or balance sheets are primary tools for grain market analysts. There are three types of S&Ds: farm, commercial and total. Farm S&Ds refer to grain produced and held on Canadian farms. Once the grain leaves the farms it enters the commercial system. Commercial S&Ds show the movement of grain from its receipt at primary elevators or process elevators to its final disposition at process or terminal elevators or in export or domestic markets. Commercial S&Ds are not published but are used for analysis purposes. Total (often called national) S&Ds bring together the data from both the farm and commercial S&Ds.

Farm S&Ds are available at the provincial level while commercial and total S&Ds are available only at the Canada level. The only exception is corn for which no farm S&Ds are produced; however, total supply and dispositions are produced for Canada, Ontario, Quebec and Other Provinces.

S&Ds are produced for a crop year, usually August 1 to July 31. The tables are revised after the release of farm stocks or production estimates, Canadian Grain Commission annual revisions or Statistics Canada trade data revisions.

Supply data

Production and farm stocks

Data on grain production and farm stocks are obtained from The Field Crop Reporting Series, Statistics Canada catalogue no. 22-002-X. Production and farm stock data comprise part of the supply of both the farm and national S&Ds.

Production data are estimated on a "field run" or dockage included basis. Three separate surveys in July, August and November provide data on the average yield and/or production of crops on farms. The published production estimates are obtained from the analysis of survey indicators, remote sensing data, consultation with field experts, agricultural tours and administrative data sources.

Farm stocks include marketable (whole, crushed or rolled) grain plus reserves for feed and seed as well as dockage on Canadian farms. Farm stocks are estimated at December 31, March 31 and July 31 for most grains and at August 31 for corn and soybeans. The supply and disposition analysis integrates data from various external sources such as grain deliveries from the Canadian Grain Commission and reconciles the various stock indicators.

Opening stocks are sometimes referred to as carry-in while ending stocks are sometimes called carry-out. The ending stocks for one crop year are the opening stocks for the next crop year. Opening stocks are part of the supply while closing stocks are part of the disposition in the S&Ds.

Production and farm stock data are collected by computer-assisted telephone interviews of large samples of Canadian farmers.

The survey indicators are analyzed and compared with provincial data or data from administrative sources such as the Canadian Grain Commission and the Canadian Wheat Board. S&D analyses are used extensively during the estimation process. For further information on the concepts, survey methods and data quality of the production or farm stock series please refer to Field Crop Reporting Series Reports 1 through 8 (Catalogue no. 22-002-X).

Commercial stocks

Commercial stocks comprise part of the supply and the disposition of the commercial and the national S&Ds. The total commercial stocks are calculated using stocks from the licensed system added to stocks in unlicensed positions. Wheat stocks in unlicensed positions are obtained from the Miller's Monthly Report, a Statistics Canada survey.

The licensed commercial stock data (or visibles) are produced by the Canadian Grain Commission (CGC) and exclude dockage. The data are received through regular reports from CGC licensees. The stocks consist of grain supplies held at licensed primary (country), process, transfer and terminal elevators, grain held in condominium storage and grain in-transit in rail cars and ships. The in-transit stocks are calculated for the Great Lakes, the railway Western Division and the railway Eastern Division.

Commercial stocks of special crops are obtained from Statistics Canada surveys of special crop companies. Commercial stocks of corn and soybeans are also obtained from Statistics Canada surveys of grain elevators.

Producer deliveries

Producer delivery data appear in the farm and the commercial S&Ds. The deliveries are a supply at the commercial level and a disposition at the farm level, thereby cancelling out for the total S&D (see Survey Estimates).

Western Canada

The largest portion of western delivery data are licensed deliveries published by the Canadian Grain Commission. Producer deliveries to licensed facilities are monitored by the Canadian Grain Commission. Licensees are required to provide weekly company summaries of the cash or storage ticket information to the Commission under the Canada Grains Act.

These deliveries are considered complete since they represent all deliveries reported on cash tickets or storage tickets to all licensees. There is no sampling or benchmarking. However, some licensees report only on a monthly basis and there is occasional non-reporting. The data collection, edit and publication methods employed by the Canadian Grain Commission have remained fairly consistent; therefore, the final data are comparable over time although there can be significant time lags. Further, a lack of analysis on a regular basis during the crop year can produce some large anomalies. The Grain Marketing Unit regularly analyzes the data and does consult with the Commission on apparent problems.

Provincial grain delivery data published by the Canadian Grain Commission represent the province where the licensed facilities are located. The data do not represent the province where the grain was grown or where the producers live. It is known that some producers cross provincial borders to deliver their grain to elevators but little data are available on the quantities of grain in question except for canola. Estimates of inter-provincial canola movement to crushing plants are obtained from the monthly survey of crushers –Report of Crushing Operations and are used to adjust the provincial canola deliveries.

Unlicensed deliveries for western Canada are estimated by the Grain Marketing Unit and are added to the licensed delivery totals. These unlicensed deliveries represent deliveries to unlicensed facilities (feed mills, distillers, ethanol and biodiesel plants and direct exports).

Deliveries to western unlicensed feed mills are estimated from the Feed Grain Purchases Survey conducted by the Agriculture Division of Statistics Canada and supplemented by trade information. This survey also provides information on inter-provincial movement related to feed grain purchases by feed mills. Historically, deliveries to western unlicensed feed mills were estimated on the basis of the Annual Survey of Manufactures or with deliveries to designated purchasers under the Western Grain Stabilization Program.

Eastern Canada

The Canadian Grain Commission produces delivery data for licensed facilities in eastern Canada; however, a significant portion of the eastern grain moves through unlicensed channels. Therefore, the Grain Marketing Unit uses a variety of other sources to produce the eastern delivery data that are published in the farm supply-disposition tables.

In Ontario, deliveries of corn, soybeans, wheat and canola are derived from administrative check-off data maintained by the Ontario Corn Producers' Association, the Ontario Soybean Growers' Association, the Ontario Wheat Producers' Marketing Board and the Ontario Canola Growers' Association. In 2009, the wheat, corn and soybean associations combined to form Grain Farmers of Ontario.

Under the Ontario Soybean Growers' regulations, all soybean sales must be reported to the board. Most of the sales are reported by grain dealers.

The Ontario Corn Producers' Association check-off system came into effect in 1984/85. All commercial buyers of corn must be licensed by the Ontario Ministry of Agriculture, Food and Rural Affairs and must deduct a levy for every tonne of corn they purchase. Seed corn, corn fed on farms, inter-farm sales and sales out of the province are excluded.

Historically, the Ontario Wheat Producers' Marketing Board was the sole selling agency for wheat grown in Ontario and sold outside the farm sector. Ontario wheat marketing data were obtained from the Board's administrative data. As of 2003 and after a phase-in period that began in 2000, farmers have the option of direct marketing their crop both within Ontario and for export without an exemption certificate. However, the Board is still able to provide the deliveries of both Board and non-Board wheat because of a revision to the Board's legislation that requires license fees to be collected on all wheat marketed, except farm to farm sales. Further, there are now requirements under the Grain Financial Protection Program of the Farm Product Payment Act for the Board to report sales of non-Board wheat to AGRICORP, the Ontario crop insurance corporation.

In Quebec, the quantity of wheat milled was used historically to estimate deliveries. Data are now obtained from La Fédération des producteurs de cultures commerciales du Québec. This is the same source for the barley delivery data.

The oat deliveries originate from the Millers Monthly Report.

No marketing data are available for the Atlantic provinces.

Imports

Import data are a supply component in the national S&D. Imports are discussed in more detail under the Trade Data section.

Disposition data

Seed

Seed data are included in both the farm and the national S&Ds.

Seed requirements are based on average producer seeding rates multiplied by the area seeded. The average seeding rates are updated with an occasional Survey of Seeding Progress conducted by the Field Crop Reporting Unit of Statistics Canada.

Human food and industrial use

Human food and industrial use data are a component of the national S&Ds.

For the cereals, the human food data are mainly collected from the survey Millers Monthly Report and the Survey of Grain Used for Industrial Purposes. An adjustment is made to remove flour exports from the domestic use totals to avoid double counting. Historically, data from the Annual Survey of Manufactures supplemented the human food component of the S&Ds. The human food component is usually estimated on a current basis and is updated when the survey data become available. Since human food use tends to be fairly stable from year to year and the quantity is relatively small, the effect of an estimation error is considered limited. Further information on the millers' survey and the survey of grain used for industrial purposes may be found in the section Survey Estimates.

Industrial use of cereals for ethanol and biodiesel production is obtained from the Survey of Grain Used for Industrial Purposes and from the Survey of Commercial Stocks of Corn and Soybeans. Further information on these surveys may be found in the section Survey Estimates.

Industrial use data for the oilseeds are obtained from a monthly survey of Canadian oilseed crushers-Report of Crushing Operations conducted by the Grain Marketing Unit of Statistics Canada. Further information on the crushers' survey may be found in the section Survey Estimates.

Loss in handling

Loss-in-handling data are included in the national supply-disposition tables only.

The 'loss-in-handling' category includes drying loss, outturn loss (the difference between the loading and unloading weights of ships or railcars), fire loss, losses due to unusual circumstances such as train derailments and Maritime disasters. This category includes gains in the net weight of grain due to overages from weighovers and dockage shipped in flaxseed and canola within allowable tolerances. These adjustments data are compiled annually by the Canadian Grain Commission from information reported by licensees. During the crop year, losses in handling are estimated by the Grain Marketing Unit on advice from the Canadian Grain Commission.

Feed, waste and dockage

Feed, waste and dockage data are calculated residually in the S&Ds. The data are analyzed to ensure they relate to indicators such as the number of grain consuming animal units on farms and in feed lots, the harvest conditions affecting grain quality, the established ratios of dockage to delivered grain and grain inspections as reported by the Canadian Grain Commission.

Although analyses are conducted on these data, the quality of the feed numbers is dependent on the quality of the other data in the S&Ds. An unusual estimate in this category may indicate a problem with another data series such as production, deliveries or stocks, rather than a change in feeding patterns.

Exports

Exports are a major component of grain disposition in the national S&Ds. This is discussed in more detail in the Trade Data section.

Trade data

Export data

Grain exports are obtained mainly from the Canadian Grain Commission and represent export clearances from licensed facilities. Unlicensed exports of non-Board grains such as truck shipments of flaxseed or oats to the United States and exports of grain products (flour and malt), supplement the licensed exports. Unlicensed exports and product and special crop exports are obtained from the International Trade Division of Statistics Canada.

The Canadian Grain Commission data are obtained during the daily weighing, grading and loading of grain at terminal and transfer elevators. Primary elevator companies also report direct exports from their facilities. Export clearances are termed "net"; however, exports of flaxseed and canola normally include some dockage.

The Canadian Grain Commission publishes export data weekly in The Grain Statistics Weekly and monthly in Exports of Canadian Grain and Wheat Flour. Final detailed crop year data are released in Canadian Grain Exports, usually in the month of November following the end of the crop year.

Unlicensed exports to the United States are the difference between the licensed data provided by the Canadian Grain Commission and the total (licensed and unlicensed) exports published by the International Trade Division of Statistics Canada. Canadian exports to the United States are provided to the International Trade Division by the US Bureau of Commerce and are based on US customs import documents. Trade data from Statistics Canada are classified according to the Harmonized System (H.S.), an international commodity classification. The HS commodity codes for grain and grain products are shown in Text Table I.

The Canadian Grain Commission export data are used for durum wheat and barley exports to all countries and for wheat (excluding durum), oats and canola exports to all countries except the United States. For wheat (excluding durum), oats and canola exports to the United States, Statistics Canada data are used. Statistics Canada data are also the source for rye and flaxseed exports, product exports such as flour and malt and special crop exports. The product exports are converted to grain equivalents using factors developed from the Miller's Monthly Report.

Import data

Import data are obtained from the International Trade Division of Statistics Canada. These data are derived from administrative records collected by the Canada Border Services Agency.

The Canadian Grain Commission compiles import data moving into the licensed system. These data are not used because a large portion of grain imports does not enter the licensed system.

Data quality

The Canadian Grain Commission reconciles their export data for the western Board grains (wheat, durum wheat and barley) with Canadian Wheat Board monthly sales reports. The Canadian Wheat Board, the sole seller of wheat and barley for human consumption or for export in western Canada maintains records of sales and shipments of their products. The Canadian Grain Commission works closely with the Board to ensure the accuracy of these export data. Regular inquiries from companies trading grain also serve as a check on the Commission's export data.

The Grain Marketing Unit does a regular review of the grain export data from the Canadian Grain Commission and from Statistics Canada's International Trade Division. Attempts are made to reconcile the two series whenever possible while still considering the differences in methodology. Timing of the receipt of documents by the two agencies may sometimes result in temporary data discrepancies.

When goods are imported into or exported from Canada, declarations must be filed with Customs giving such information as description and value of the goods, origin and port of clearance of commodities and the mode of transport. Most of this information is required for the purposes of Customs Administration. Statistics developed from administrative records of Customs are commonly referred to as Custom-based trade statistics.

Custom-based trade statistics are more accurate at measuring imports than they are at measuring exports. This is the case because Customs are typically more vigilant with respect to goods entering the country than they are with goods leaving the country.

Custom-based export statistics may understate and/or incorrectly portray the destination of exports. Export statistics are understated when the proper documentation is not filed with Customs. Exports are incorrectly portrayed when the country of final destination is inaccurately reported on the customs documentation - this occurs most frequently when goods are routed through an intermediary country before continuing on their final destination.

Statistics Canada does not have a measure of undercoverage but periodically conducts reconciliation exercises with its major trading partners – excluding the United States.

International data

Data used in the International S&Ds and in the International Oilseeds Data table of this publication originate with Statistics Canada, Agriculture and Agri-Food Canada, the Foreign Agricultural Service of the United States Department of Agriculture and OIL WORLD. Countries around the world have different marketing years (i.e. August-July, October-September, etc.) due mainly to climatic differences. Since marketing years are not consistent between countries, care should be taken in adding stocks for these differing periods.

Price data

The price data used in this publication are obtained from a wide variety of sources. Canadian cash grain prices are obtained from The Canadian Wheat Board, ICE Futures Canada, The Ontario Wheat Producers' Marketing Board, the Ontario Corn Producers' Association, the Ontario Soybean Growers, Provincial departments of agriculture and the United Nations. Canola oil and meal prices are obtained from a survey of industrial firms done by the Agriculture Division of Statistics Canada. Cash grain prices for the United States are obtained mainly from United States Department of Agriculture regional offices. Board prices are obtained from the Canadian Wheat Board. The Canadian Wheat Board also provides the Pool Return Outlook and Estimated Pool Return data. Future prices, volumes and open interest data are those published by ICE Futures Canada and CME Group (formerly The Chicago Board of Trade), The Kansas City Board of Trade and The Minneapolis Grain Exchange in the United States. Exchange rates are obtained from the Bank of Canada. Special crop prices are daily bids obtained directly from some of the major firms purchasing these crops. The monthly and crop year average prices are simple, non-weighted averages.

Special crops data

Special crop production and farm stock data are obtained in the same manner as the major grains (See Production and Farm Stocks). Commercial stock data are obtained from a survey of special crop companies (see Survey of Commercial Stocks of the Major Special Crops). Import and export data are obtained from Statistics Canada's International Trade Division and are obtained from Customs documents. Delivery data are estimated from the Grains and Specialty Crop Survey, levy data obtained from provincial associations, input from special crop processors and industry analysts and from supply-disposition analyses.

Survey estimates

Millers surveys

Survey description and frame

The Miller's Monthly Report is a monthly census of companies milling over 500 metric tonnes of grain in Canada for human consumption. Companies milling very small quantities are surveyed annually on the Miller's Annual Report. These surveys began in 1925.

The mailing list for the survey is updated annually from the Annual Survey of Manufactures. Updates are also regularly obtained from trade sources and from the survey itself.

The data collected comprise part of the domestic disappearance of grains for human uses contained in the national supply-disposition tables for the major grains. The data are also used by governments, grain millers, farmers and other businesses for the purpose of market research and consultation.

In addition, the annual survey is used in the determination of expansion factors for the monthly data and to determine the final crop-year totals for the previous year.

Instrument design

The respondent completes a paper questionnaire. The last redesign of the questionnaire was in 2000. The collection is done by mail with telephone follow-up.

The questionnaire asks for the amount of grain milled, products produced by grade, stocks in mill bins (unlicensed positions), operating days and plant capacity. Some of these data are used for internal purposes such as S&D analyses and are not published.

Error detection, imputation and data quality

Questionnaires are checked prior to data entry to ensure that the total quantity of flour and millfeeds produced is comparable to the quantity of wheat milled. Summarized data are reviewed for extreme variability from month-to month and compared to estimates of other plants producing similar products.

Current month data are imputed when necessary based on same plant, previous month. Data are revised on or after the annual survey or when actual data are received. Occasionally, data are also available from industry sources. The impact of imputation is considered small since the data are fairly stable on a month-to-month basis. The survey data are not benchmarked.

Total milling estimates from this survey are compared to grain milled reported by the Canadian Grain Commission licensees and obtained from their 'Weekly Report of Grains at Process Elevators'. Average extraction rates and supply-disposition analyses also aid in data validation.

Data accuracy

While considerable effort is made to ensure high standards throughout all stages of collection and processing, the resulting estimates are inevitably subject to a certain degree of non-sampling error. Examples of non-sampling error are coverage error, data response error, non-response error and processing error. The major source of non-sampling error for this survey is considered to be data response error.

Coverage error can result from incomplete listing and inadequate coverage of the population of mills. However, given the infrastructure and the supplies of grain needed for a mill, it is unlikely that a new plant could start-up undetected and that any coverage error would be temporary and would have only a minimal effect on the resulting estimates. The estimates also include data from the small mills reporting to the Millers Annual Report.

Data response error may be due to questionnaire design, the characteristics of a question, inability or unwillingness 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. This survey has been in place for many years and most respondents are well versed in the survey concepts.

Non-response error is related to respondents that may refuse to answer, are unable to respond or are too late in reporting. In these cases, data are imputed. The extent on any imputation error decreases with increases in the response rate and attempts are therefore made to obtain as high a response rate as possible. Final response for this survey is about 90% on a monthly basis and 100% annually. Analysts keep in contact with the mills and the related industry associations to maintain the high response rate.

Processing error may occur at various stages of processing such as data entry, editing and tabulation. Measures have been taken to minimize these errors. A few trained staff work on this survey. Data entry and edit are performed simultaneously due to the spreadsheet design which allows errors to be quickly seen. Historical ratios also aid in eliminating outliers created by data entry. Tabulation is automated to eliminate human error.

Report on Crushing Operations

Survey description

The Report on Crushing Operations is a monthly census of oilseed crushing plants in Canada. The data are part of supply-disposition statistics of major grains and allow the calculation of the domestic disappearance component. They are also required to verify grain production and farm stocks.

Reference period

This survey is based on the calendar month. This survey became active in August 1971.

Survey frame

The universe consists of eleven crushing plants. Updates are rare and are obtained from trade sources.

Instrument design

Data are collected direct from respondents on paper questionnaires with facsimile follow-up. The questionnaire was last revised in 1999.

The information requested includes seed crushed and oil and meal produced and the related month-end stocks; receipts of canola from producers by province; and imports for crushing.

Stocks of canola include those owned by the reporting companies and held at crushing plants, in transit and at export positions. Stocks of soybeans are those held at crushing plants only. No data are available on soybeans in-transit or at export positions; however, the volume of soybeans in such positions is considered to be small.

Canola meal produced may include some additives such as water or kaolin. This may occasionally result in a larger amount of canola oil and meal produced than seed actually crushed.

Error detection

Data quality is maintained by standard editing techniques that are particularly rigorous with this survey because it is small. Questionnaires are scanned before data entry. After data entry, each manufactured product is checked to ensure that recovery rates fall within established edit limits. Other edits include ensuring plants report the commodities and the approximate volume expected; comparing current data to data from previous months and by comparing trends between plants.

Compiled data are reconciled with Canadian Grain Commission crush data as obtained from the 'Weekly Report of Grains atProcess Elevators'. The Canadian Grain Commission does not obtain data from unlicensed crushers. Published survey data are monitored closely by the Canadian Oilseed Processors Association and by Canadian crushing companies. Due to the numerous check mechanisms on the survey estimates, the overall data quality is considered to be very good. Revisions to this series are usually minimal. No benchmarking is necessary.

Confidentiality

While considerable effort is made to ensure high quality standards throughout all stages of collection and processing, the resulting estimates are inevitably subject to a certain degree of non-sampling error. Examples of non-sampling error are coverage error, data response error, non-response error and processing error. The major sources of non-sampling error are response errors, such as reporting seed with dockage instead of clean seed. Estimation is kept at a minimum because of follow-up procedures and good industry cooperation.

Coverage error can result from incomplete listing and inadequate coverage of the population of crushing plants. However, given the infrastructure and the oilseed supplies needed for a crushing plant, it is unlikely that a new operation could start-up undetected. The Canadian Oilseed Processors Association also aids in list maintenance.

Non-response error is related to respondents that may refuse to answer, are unable to respond or are too late in reporting. In these cases, discussions are held with the respondents. Data are never imputed. Analysts keep in contact with the crushing plants and the Canadian Oilseed Processors Association to maintain the 100% response rate.

Processing error may occur at various stages of processing such as data entry, editing and tabulation. Measures have been taken to minimize these errors. A few trained staff work on this survey. Data entry and edit are performed simultaneously due to the spreadsheet design. Tabulation is automated to eliminate human error.

Survey of commercial stocks of corn and soybeans

Survey description

The survey collects data on stocks of corn and soybeans in unlicensed, commercial elevators and on industrial use of corn for food, distilling or ethanol. (Industrial users normally operate elevators that accumulate grain prior to processing). The mail survey is conducted three times a year: December 31, March 31 and August 31 (the end of the crop year for corn and soybeans) to coincide with surveys collecting data on farm stocks.

Survey frame

The survey frame is updated once a year from administrative records maintained by the Ontario Ministry of Agriculture, Food and Rural Affairs. This list of grain elevators is exhaustive and regularly updated. Each fall, after ensuring that the frame is current, a sample of more than 100 elevator firms is randomly selected that stratifies by elevator capacity and type of elevator (country, feed or industrial use). The sample remains the same for the three surveys conducted within that crop year.

Edit and imputation

After telephone follow-up, completed questionnaires are received from more than 90% of the sample. Data are compared historically, when possible, against the corresponding data from a year earlier and also relative to the elevator's registered capacity. Non-responses (including partial non-responses, no contacts and refusals) are concentrated in those elevators with smaller capacities. Partials are imputed from last year's historical figures, when possible. Otherwise, non-responses are imputed using the nearest-neighbour approach within the stratum. On rare occasions, when a missing questionnaire appears to be for a unique elevator, the initial sample weights are adjusted.

Instrument design

This originated as a mail survey with telephone follow-up. It was converted to an Internet survey. A few respondents are unable to report electronically. Therefore, a combination of Internet and paper reporting is being used. There are two questionnaires – one for industrial users and one for country elevators.

Both questionnaires request the volume of company–owned stocks and of stocks held for others. In addition, industrial users are asked how much corn they used for processing purposes in to-date in the crop year which runs from September 1 to August 31.

Data accuracy

The stocks and industrial use estimates are generated at the provincial level and are used in S&D analysis. The survey's stocks estimates are added to licensed stocks, published by the Canadian Grain Commission, to obtain total commercial stocks.

While considerable effort is made to ensure high standards throughout all stages of collection and processing, the resulting estimates are inevitably subject to a certain degree of non-sampling error. Non-sampling error is not related to sampling and may occur for many reasons. For example, non-response or incomplete responses are important sources of non-sampling error. Population coverage, differences in the interpretations of questions and mistakes in recording and processing data are other examples of non-sampling errors.

Non-sampling errors are controlled through a careful design of the questionnaire, the use of a minimal number of simple concepts, consistency checks and automated edits. Measures such as response rates are used as indicators of the possible extent of non-sampling errors. The response rate for this survey ranges from 90-95%. The non-sampling error associated with the provincial estimates is believed to be negligible due to (a) the few variables collected, (b) the limited size of the sample, and (c) the manual editing done by a subject-matter person prior to data capture.

Sampling error can be measured by the standard error (or standard deviation) of the estimate. The coefficient of variation (CV) is the estimated standard error percentage of the survey estimate. Estimates with smaller CVs are more reliable than estimates with larger CVs. The CVs for the total stocks are in the range of 3-4%. Generally any estimate with a C.V. value under 5% is considered to be of excellent quality.

Survey of commercial stocks of the major special crops

Survey description

The survey collects data on national, commercial stocks of peas, lentils, mustard, canary seed, sunflower seed and chickpeas.

The survey is conducted three times a year: December 31, March 31 and July 31 to coincide with surveys collecting data on farm stocks.

Reference period

This survey is conducted with reference dates of December 31, March 31 and July 31. The first survey took place in August 2000 and data were collected for the periods July 1999 and July 2000.

Survey frame

All companies known to store special crops, whether licensed or unlicensed, are included. Special crop dealers with no storage facilities or companies that maintain stocks in US positions only are excluded.

The original mailing list was provided by the Canadian Special Crops Association. Additions were made from the licensee list of the Canadian Grain Commission and from lists maintained by Saskatchewan Agriculture and Alberta Agriculture and Rural Development. As well, updates continue to be made from trade sources.

Instrument design

Data are obtained for canary seed, chickpeas, dry field peas, lentils, mustard seed and sunflower seed. The questionnaire requests the volume of company-owned stocks and of stocks held for others. There are two types of stocks held for others: held for farmers and held for companies.

Quantities held for farmers comprise part of the commercial stocks since these stocks have physically left the farm and would not be included in the farm stock data. The identification of quantities held for companies permits the removal of duplicate reporting. In the special crops business, some companies sell their crops to others for export while still maintaining the stocks in their own facilities. While the seller would report these stocks as held for others, the purchaser would report these stocks as company owned.

Error detection

There is little imputation as the response rate is high and editing procedures are rigorous. Data are verified by comparison to previous reports and by comparing trends between companies. Commercial stock estimates are combined with the farm stock estimates to produce total stock estimates. These data are then analyzed in national supply-demand tables and are compared to industry forecasts.

Revisions, if required, will be made during the analysis of subsequent stock surveys. There is no confidentiality for the survey estimates since forms have been provided by all of the major firms which permit publication of the survey totals.

Data accuracy

While considerable effort is made to ensure high standards throughout all stages of collection and processing, the resulting estimates are inevitably subject to a certain degree of non-sampling error. Examples of non-sampling error are coverage error, data response error, non-response error and processing error.

Coverage error can result from incomplete listing and inadequate coverage of the population. This industry is relatively new and there have been many changes to the universe. However, a wide variety of sources are used to regularly update the universe including member lists from the Canadian Special Crop Association, licensees of the Canadian Grain Commission and lists of special crop exporters and marketers compiled by provincial governments and others. Press clippings are also monitored daily. Since relatively few companies make up the majority of the stocks, it is generally believed that any under coverage would be small.

Data response error may be due to questionnaire design, the characteristics of a question, inability or unwillingness 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. However, the respondents have been changing quickly due to company amalgamations and bankruptcies and some lack the background to ensure consistency. Therefore, the survey analysts are conscious of the need to monitor reporting and to discuss any anomalies with the companies in question.

Non-response error is related to respondents that may refuse to answer, are unable to respond or are too late in reporting. In these cases, data are imputed. This is considered to be the most likely source of any error for this survey. The extent of any imputation error decreases with increases in the response rate and attempts are therefore made to obtain as high a response rate as possible while minimizing the response burden. Analysts liaise with the companies and the related industry associations to maintain the high response rate of approximately 95% for the whole survey. The response may be less for individual commodities.

Processing error may occur at various stages of processing such as in data entry and tabulation. Measures have been taken to minimize these errors. Only a few trained staff work on this survey. Edits in the electronic reporting system prevent the entry of outliers by respondents. The spreadsheets used for data entry and tabulation of both the electronically reported data and the data reported on paper permit the analysts to quickly detect apparent anomalies. It is considered that processing errors are minimal.

Feed grain purchases survey

Survey description

The survey addresses a gap in the Western Provinces regarding the value of feed grains. The survey data are used to update the quantities and values of grain deliveries. The quantities are used to improve the estimates of unlicensed grain deliveries in farm supply-disposition tables that in turn improve the estimates of crop production and farm stocks. The values are subsequently used to improve the published farm cash receipts and by the Canadian System of National Accounts (CSNA) to calculate the Gross Domestic Product (GDP) and related variables.

Collection period

Twice a year, at the end of the crop year (July 31st) and the calendar year (December 31st).

Target population

Sixteen companies, which report for all of their subsidiary locations across the Western Provinces. These firms are feed mills that buy grain directly from farmers or from grain dealers. The list of mills was obtained from industry discussions and from the Animal Nutrition Association of Canada. The list is maintained from trade sources and from the survey itself. Feed lots are excluded.

Instrument design

The questionnaire was designed in consultation with internal and external specialists, as well as some respondents, before the start of the survey in 2003.

Sampling

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

Data are collected for all units of the target population, therefore no sampling is done.

Data sources

Responding to this survey is mandatory. Data are collected directly from survey respondents and extracted from administrative files.

The first collection period was in July 2003. The collection is done by mail with mail/facsimile and telephone follow-up.

The questionnaire asks for the crop year to-date quantities of feed grains purchased from farmers and grain dealers by grain. The data are requested for grains originating from individual provinces in the west, for the total east, for other countries and in total.

Error detection

Questionnaires are checked prior to data entry to ensure that the geographic distributions add to the total. In July, data are compared to the December reports to ensure the July data are at least equal to the crop year to-date data reported in December. Reported data are also compared to previous reports to ensure that the commodities reported are the same. Any changes are reviewed for reasonableness given the current availability of feed and the number of livestock.

Data are not generally imputed. Data are revised for non-response or for incorrect reporting when revisions are received. The survey data are not benchmarked.

Data accuracy

Since this is not a sample survey, there is no sampling error.

While considerable effort is made to ensure high standards throughout all stages of collection and processing, the resulting estimates are inevitably subject to a certain degree of non-sampling error. Examples of non-sampling error are coverage error, data response error, non-response error and processing error. The major sources of non-sampling error for this survey are considered to be coverage error and non-response error.

Coverage error can result from incomplete listing and inadequate coverage of the population of feed mills. While coverage is considered to be very good in Manitoba and Saskatchewan, there have been difficulties identifying the mills in Alberta. There are also many feed mills that purchase only small quantities occasionally. The effect of not including every small mill in the survey is considered to be negligible.

Data response error may be due to questionnaire design, the characteristics of a question, inability or unwillingness 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. However, this survey is relatively new and some respondents not be well versed in the survey concepts which introduces some non-sampling error.

Non-response error is related to respondents that may refuse to answer, are unable to respond or are too late in reporting. In these cases, data are generally not imputed. Attempts are therefore made to obtain as high a response rate as possible. Final response for this survey is about 90% annually.

Survey of Grain Used for Industrial Purposes

Description

This is a survey of plants using grain mainly to produce ethanol or biodiesel in western Canada. There are few plants, but their numbers are increasing and they use significant volumes of grain.

The use data are split into purchases from licensed grain companies and from farmers.

The data are required by the Agriculture Division of Statistics Canada for calculations of grain deliveries and supply-demand. These data are then used to determine grain stocks and production and farm cash receipts.

Collection period

The survey is conducted three times per year to coincide with calculations of farm stocks at December 31, March 31 and July 31.

The target population

The target population consists of plants using grain mainly to produce ethanol or biodiesel in western Canada. Plants using grain for other industrial purposes may also be surveyed.

Instrument design

The questionnaire was designed in consultation with internal and external specialists, as well as some respondents, before the start of the survey in 2008.

Sampling

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

Data are collected for all units of the target population, therefore no sampling is done.

Data sources

Responding to this survey is mandatory. Data are collected directly from survey respondents.

The first collection period was in July 2008. The collection is done by mail with mail/facsimile and telephone follow-up.

The questionnaire asks for the crop year to-date quantities of grain used, primarily wheat and corn, and on the stocks of grain at the plants.

Error detection

Data are compared to the previous reports to ensure the data are at least equal to the crop year to-date data reported earlier.

Reported data are also compared to previous reports to ensure that the commodities reported are the same.

Any changes are reviewed for reasonableness given grain pricing and plant capacity.

Imputation

Data are not generally imputed.

Quality evaluation

This is a census and the data quality is maintained by standard editing techniques which are rigorous. Apparent data discrepancies are either scrutinized by professional staff or the company involved is contacted. Supply and disposition trends, used by government and industry stakeholders, help to confirm the results of the survey.

Revisions and seasonal adjustment

Data are revised for non-response or for incorrect reporting when revisions are received. The survey data are not benchmarked.

Data accuracy

Since this is not a sample survey, there is no sampling error.

Coverage error can result from incomplete listing and inadequate coverage of the population of biofuel plants.

Data response error may be due to questionnaire design, the characteristics of a question, inability or unwillingness 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. However, this survey is relatively new and some respondents may not be well versed in the survey concepts which may introduce some non-sampling error.

Non-response error is related to respondents that may refuse to answer, are unable to respond or are too late in reporting. In these cases, data are generally not imputed.

Attempts are therefore made to obtain as high a response rate as possible. Final response for this survey is expected to be 90%.

Revision policy/Data comparability

As a general policy, revisions are included in the next issue of the publication.

Every attempt is made to keep the methods comparable over time. When this is not possible, an attempt is made to revise the methods used and the accompanying data back to the last Census of Agriculture year. The Census is conducted every five years.

The S&Ds are updated about five times per year. Major updates include new production or stock data, annual revisions to import and export data by the International Trade Division of Statistics Canada and the release of final crop year data on commercial stocks, deliveries, exports, and domestic disappearance by the Canadian Grain Commission.

Production and farm stock data, produced by Statistics Canada, are subject to revision for two years after first being published. Further revisions may be made during intercensal revisions. Revisions are a result of benchmarking to the last Census of Agriculture.

Canadian Grain Commission data on deliveries, commercial stocks and exports are subject to constant revisions throughout the current crop year. Following the end of the crop year, revisions are made at about week six of the new crop year and in the following spring. No further revisions are made.

Trade data produced by the International Trade Division of Statistics Canada are revised monthly during the current calendar year for the major grains and quarterly for the special crops. At the end of the calendar year, the previous four years are open for revisions.

Revisions to oilseed crushing survey data, although rare, are made when received from the crushers and are released monthly.

Revisions to milling data are made when received from the millers and are published monthly. At the end of the crop year, data from the annual survey (smaller mills) results in the largest adjustment of the year.