Presentation on theme: "DLI Orientation: Concepts A Framework for Thinking about Statistical Information Chuck Humphrey Data Library University of Alberta April 2004."— Presentation transcript:
DLI Orientation: Concepts A Framework for Thinking about Statistical Information Chuck Humphrey Data Library University of Alberta April 2004
Statistical Information Two models for identifying and selecting appropriate statistical information: 1. A chart of statistical information Distinguishing statistics & data Distinguishing aggregate data & microdata
Statistical Information 2. Continuum of access Matching dissemination channels with desired products
Statistics or Data Statistics numeric facts/figures created from data, i.e, already processed presentation-ready Data numeric files created and organized for analysis requires processing not ready for display
Statistics or Data
Chart of Statistical Information
This is a typology of the categories or classes of statistical information. Remember the relationship between statistics and data, however, is causal. Statistics are created from data.
Chart of Statistical Information
An overlap occurs in this chart between Statistics: Databases and Data: Aggregate, which will be discussed below.
Chart of Statistical Information In print
In Print Rely on yearbooks, statistical abstracts, catalogues, and indexes to locate statistics in print. Examples of online indexes to print resources: Statistical Universe and Tablebase Example of an online catalogue that includes print resources: Statistics Canadas Online Catalogue
Chart of Statistical Information Online
Online Statistics Example of e-publications Statistics Canada Downloadable Publications (DSP) Example of e-tables Canadian Statistics (STC Website) Example of statistical databases CANSIM II (STC Website, E-STAT, CHASS)
E-Publications Tend to be available in PDF format Can use the Select Text Tool in the Adobe Reader and copy columns to another application
E-Tables Tend to be displayed in HTML May provide a pull-down list to view other categories in the table Some e-tables will provide an alternate format for the table that can be downloaded (e.g., the Census tables are available in comma-separated ASCII, IVT, and print-friendly formats)
Databases Often use HTML forms to define the statistics to be retrieved May offer a variety of output formats for the retrieved statistics (e.g., E-STAT provides IVT format for Beyond 20/20, graphs, charts, maps, and ASCII formats for spreadsheets and databases)
Chart of Statistical Information Aggregate Data
Aggregate Data Aggregate data consist of statistics that are organized into a data structure and stored in a database or in a data file. The data structure is based on tabulations organized by time, geography, or social content.
Aggregate Data Data Structure Time Geography Social Content Example: CANSIM II
Aggregate Data Time series data have long fueled econometric models based on macro-economic indicators. Comma-separate values (CSV) have become an important format for time series data, which is often manipulated in Excel if not analyzed in a spreadsheet.
Aggregate Data Data Structure Time Geography Social Content Example: CENSUS
Aggregate Data Increased availability of GIS software has created greater demand for Census statistics organized as aggregate data. Beyond 20/20 has become a popular tool for reshaping census statistics from 1996 and 2001 for use with GIS software. DBF is the most commonly used format to share census statistics with GIS software.
Aggregate Data A map from E-STAT of Montreal Census Tracts
Aggregate Data Small area statistics are a special category of aggregate data. These data files consist of statistics for small geographic areas usually calculated from a population or manufacturing census or an administrative database with enough cases to create accurate summaries for small areas.
Aggregate Data Data Structure Time Geography Social Content Example: Cause of Death (HID)
Aggregate Data Also known as cross-classified tables, these files tend to be made of statistics constructed from social- content variables. Examples of cross-classified tables in DLI are found in education and justice.
Chart of Statistical Information Microdata
This is raw data organized in a file where the lines in the file represent a specific unit of observation and the information on the lines are the values of variables. There are different types of microdata files, which will now be discussed.
Confidential Microdata Master files: these files contain the fullness of detail captured about each case of the unit of observation. This detail is specific enough that the identify of a case can often be disclosed easily. Therefore, these files are treated as confidential.
Confidential Microdata Share files: these are confidential files in which the participants in the survey have signed a consent form permitting Statistics Canada to allow access to their information for approved research. These files consist of a subset of the cases in the master file.
Confidential Microdata In summary, confidential microdata get grouped into two types: master files and share files.
Public Use Microdata These microdata are specially prepared to minimize the possibility of disclosing or identifying any of the cases in a file, i.e, participants in a survey. The original data from the master file are edited to create a public use microdata file.
Public Use Microdata Steps in Anonymizing Microdata Remove of all personal identification information (names, addresses, etc); Include only gross levels of geography; Collapse detailed information into a smaller number of general categories; Cap the upper range of values of variables with rare cases; Suppress the values of a variable; or Suppress entire cases.
Public Use Microdata Statistics Canada PUMFs Only available for select social surveys that undergo a review of the Data Release Committee, an internal Statistics Canada committee. No enterprise public use microdata.
Public Use Microdata Statistics Canada PUMFs Almost all PUMFs consist of cross- sectional samples, that is, samples where the data have been collected from respondents at one point in time. Longitudinal samples, where data are collected from the same individuals two or more times, are difficult to anonymize and maintain any useful information.
Synthetic Microdata These data files have been created by author divisions to assist with the analysis of confidential data files. The files provide the full variable structure of the confidential microdata but do not contain any real cases. They are intended to be used by researchers wanting to submit a file of commands in a statistical packages language for remote job submission.
Synthetic Microdata They are also being used by those with approved projects in Research Data Centres to help prepare their analysis strategies prior to working in an RDC. Synthetic files are also commonly referred to as dummy files, although a more technical use of this term does exist for this specific type of synthetic file.
Synthetic Microdata A variety of synthetic file types are being created and tested by author divisions. One type has no real data but does contain a complete set of real variables. This type is the more technical reference to a dummy file. Another type has a mix of real data but no real cases. The purpose of this type is to provide -- in the aggregate -- results that should be close to an analysis of the real microdata file.
Synthetic Microdata Users of these files must be advised that none of the analytic results from these files should ever be reported. Their only purpose is to help researchers construct their statistical analysis programs to guard against syntax errors that might exist in their setup. The DLI FTP site clearly distinguishes synthetic files from real microdata files.
Summary: First Model
This first model provides a way of thinking about the types of statistical information that exist. Is the information Statistics or Data? If Statistics, is the information in print or online? If online, is it in an e-pub, e-table, or database? If Data, is the information aggregate data or microdata?
The Second Model It is one thing to know about the variety of statistical information that exists, but access to this information is a separate issue. The second model describes the various dissemination channels through which access is provided to statistical information by Statistics Canada.
Continuum of Access Statistics Canada provides access to its statistical information through a variety of services and initiatives that function as dissemination channels. Think of this variety as constituting a continuum along which levels of access are provided.
Continuum of Access There are three characteristics that make up this continuum: Cost : which runs from free to expensive; Restrictions or conditions : which run from open or no restrictions to very restricted; and Type of Information : which runs from statistics to data.
Continuum of Access ACCESS CHANNELS Open Free Statistics Restricted Expensive Data Depository Service Program Remote Job Submission Statistics Canada Website Data Liberation Initiative Custom Tabulations Research Data Centres
Statistics Canada Website Free, Open, Statistics The Daily is an important source of publicly-released official statistics. It has been available on the Website for several years and was the primary source for free statistics in the early years of the Statistics Canada website.
Statistics Canada Website Free, Open, Statistics With the introduction of Community Profiles from the 1996 Census in 2000 and more recent offerings from the Health Statistics Division, this dissemination channel has had a big increase in the amount of statistics available at the national, provincial, CMA, CSD, and Health Region levels.
Depository Service Program Free, Open, Statistics The Depository Service Program (DSP) has provided public access to government information for over 75 years. Through a network of public, special, and academic libraries, the Treasury Board has paid Federal Departments to release publications to the public through the DSP.
Depository Service Program Free, Open, Statistics Statistics Canada has a large series of publications that it makes available through the DSP. Many of these titles are available online in PDF format and are part of the Statistics Canada Downloadable Publication series. While these statistical publications are free, the public is required to go to a DSP library to access them.
Data Liberation Initiative Fee, Licenced, Conditional Access, Data and Statistics DLI provides a wider range of statistical information than the Statistics Canada Website or the DSP, but access in no longer free and rules apply those who are eligible to use these materials. This is a move away from free-&- open to fees-&-conditional access.
Data Liberation Initiative Fee, Licenced, Conditional Access, Data and Statistics DLI provides member institutions in the post-secondary educational sector with access to all standard data products, which consists of the statistical databases, public use microdata files, and geography files listed for sale in the Statistics Canada Online Catalogue.
Data Liberation Initiative Fee, Licenced, Conditional Access, Data and Statistics Patrons of this service must hold a current affiliation with a member institution and are restricted in their use of these materials for teaching, scholarly research, or institutional planning. Furthermore, secondary redistribution of DLI materials is not allowed.
Customized Tabulations Pay-per-view Access A long-term dissemination channel within Statistics Canada has been custom tabulation services. This is a contract service with Statistics Canada to produce tables from surveys or the Census that have not been produced for public release. Each customized product comes with its own licence.
Remote Job Submission AKA, Remote Data Access (RDA) This is a relatively new service for a select number of surveys. The terms of access vary among the author divisions offering this service. Some charge a fee (e.g., access to YITS and PISA is $75 a run), while other divisions do not charge. The Health Statistics Division requires a proposal to access the surveys for which it provides remote job submission.
Remote Job Submission AKA, Remote Data Access (RDA) Synthetic files have been created to assist with the preparation of the statistical command files that are submitted for remote processing. An analysis is prepared in the command language of a statistical package supported by the author division (SAS or SPSS, e.g.) and submitted via to the division.
Remote Job Submission AKA, Remote Data Access (RDA) All results are screened by the author division for disclosure issues prior to the output being sent to the researcher who submitted the job. This dissemination channel provides a means of producing analysis from confidential data files with conditional approval and in some instances for a fee.
Research Data Centres Restricted Access to Confidential Data Research Data Centres house select confidential data files in a controlled Statistics Canada office environment. Access is provided on a project-by- project basis. A SSHRC-administered application process is used to evaluate the proposed use of the confidential data.
Research Data Centres Restricted Access to Confidential Data Furthermore, a security clearance with Statistics Canada must be passed. With approval from both the SSHRC peer review and the security clearance, the members of a research project must undergo an orientation to the RDC, swear an oath to the Statistics Act, and sign a contract with Statistics Canada.
Research Data Centres Restricted Access to Confidential Data The advantage of RDC access over Remote Job Submission is that researchers get to work directly with the confidential data source.
CANSIM II and Trade Analyzer Services available for selected titles. Remote job submission is the most developed for NPHS. Applications can now be submitted through the SSHRC Web site. ACCESS Open Free Statistics Restricted Expensive Data
Using the Two Models Combining these two models should assist you in identifying and selecting appropriate statistical information. The types of statistical information should help you identify an appropriate product, while the continuum of access should help you locate the channel or channels through which the statistical information is disseminated.
Using the Two Models Hopefully, you will find this framework useful in your data reference interviews, which is a separate topic in this orientation, and in navigating the DLI FTP site for various statistical information.
Warning Remember that while Statistics Canada is an important source of statistical information in our country, it is not the only source. Other important sources include other federal government and provincial departments, data libraries and archives, non- & inter-governmental agencies, and commercial vendors.