Friday, February 25, 2005

Trade by Industry

Question

A UBC doctoral student is doing research on industries and wants summary information on Canadian bilateral trade with different countries by industry classification, not by the usual commodity classification.

The Statistics Canada-supplied data on Strategis' Trade Data Online has this for recent years using the NAICS classification and back to 1990 with the SIC classification. She needs further statistics for the period 1980-1989 to match some other work she is doing. She recognizes that she may have to do some manual work of matching commodities with industries. A note on the Trade Data Online site notes that the matching represents "a crude estimation of industry trade and is susceptible to errors." It also states the following:

"In an effort to quantify industry trade, Statistics Canada associates the exported commodity's Harmonized System (HS) code with a NAICS code using internal concordances."

CANSIM and every other source I have checked present import and export data by commodity, not by industry.

Since the data are originally from Statistics Canada I volunteered to put the question to the source. Is there anywhere to find this industry-related trade data, or to get access to the internal concordances noted above?

Answer

Trade data is available from 1980 to 1989 using the SIC classification system. However, it is only available as a custom request.

The US versions of the NAICS-HS concordances, for exports and for imports, can be downloaded from:
http://www.census.gov/foreign-trade/reference/codes/index.html#concordance

Wednesday, February 16, 2005

Using Beyond 20/20 to Convert to SPSS

Question

I'm looking for highest level of education by CSD for Newfoundland. It needs to be in raw data format (It's for an SPSS based stats course). Is this data, from either from 1996 or 2001 census, available as a PUMF? I can only find B2020 tables.

Answers

1) Your initial question sounded like you want CSD to be the unit of analysis: "I'm looking for highest level of education by CSD for Newfoundland." If this is the case, you were very close to solving your own question using the 2001 Census table, 95F0491XCB01002.IVT.

Here is an example of what a raw data format of this table for just CSD's from Newfoundland would look like using only the
first eight CSD's and first four Census characteristics:

Portugal Cove South (1001105) T 00000 60 20 35 0
Biscay Bay (1001109) T 00000 0 0 0 0
Trepassey (1001113) T 00000 135 25105 0
St. Shott's (1001120) T 00000 40 0 35 0
Division No. 1, Subd. U (1001124) SUN 190 40145 0
Cape Broyle (1001126) T 00000 70 60 10 0
Renews (1001131) T 00000 80 65 20 10
Fermeuse (1001136) T 00000 50 10 25 15

The columns of numbers represent, respectively, Total population 15 to 24 years by school attendance, Not attending school,
Attending school full time, and Attending school part time. These are just the first four Census characteristics of 42 contained in 95F0491XCB01002.IVT.

How to get this into SPSS quickly? Use B2020 to create the data structure by reorganizing the table so that Geography is
in the rows, Census characteristics are in the columns, and you've selected just CSD's in Newfoundland. Then do a File / Save As and output the data using the Excel Worksheet format.

Open this file in Excel and remove the first four rows in the spreadsheet, which contain stub headings from the B2020 table,
and re-save this file.

Next, start an SPSS session and use File / Open / Data and select Excel as the file type to open. In the "Opening Excel Data Source" dialogue box, uncheck the box for "Read variable names from the first row of data."

2) If I understand what you are looking for correctly, you are looking for the ascii version of table 95f0418xcb01006.ivt, yes?
That's the BST for highest degree, certificate or diploma, by age groups and sex, for census subdivisions.

You will find it in ascii format on the DLI ftp site, in
census/2001/Ascii/Topic-Based-Tabulations/education-in-canada-attendance-ascii/

There are no SPSS commands, but they could be created from the accompanying documentation. You will need to write a data list statement for a number of records per case (ie CSD).

The alternative to using the ascii file would be to use the Beyond 20/20 file to structure the rows the way you want them, then write the file out as a .csv or .xls file, and read that in to SPSS. It would be rather easier than working with the ascii file, at least while we can still read .ivt files.

Market Trends in Deli and Processed Meat Industry

Question

I have a student looking for information on market trends in the deli and processed meat industry but I am coming up relatively empty handed. Any suggestions?

Answer

There are a couple of CANSIM tables that come close. CANSIM Table 080-0010 - Retail commodity survey, quarterly (Dollars) is in E-STAT and provides quarterly figures from 1998 to 2003. The geographic coverage is only for all of Canada, though. This table includes an entry for "Deli (excluding cheese), salad bars and prepared foods for take-out."

Table 80-0013 provides annual sales instead of quarterly from 1999-2001 using the same retail commodity classification
as 80-0010.

The NAICS code for delicatessens is 44521. And while I could find three more related tables, they didn't provide stats at this level of detail provided in 80-0010. Table 80-0014, however, groups 44512, 44521, 44522, 44523, 44529 together into a Convenience and Specialty Food Stores category. This table provides monthly sales data from January 1991 to April 2004 (in E-STAT). The geographic coverage includes provinces.

CANSIM Table 080-0014 - Retail trade, sales by trade group based on the North American Industry Classification System (NAICS), monthly (147 series).

Table 080-0015 provides the coverage of 80-0014 but quarterly rather than monthly.

Another possibility is Table 80-0011 - Annual retail store survey, financial estimates by store type and trade group based on the North American Industry Classification System (NAICS), which covers 1999 to 2001 (in E-STAT) again using the same NAICS groupings as in 080-0014.

I didn't have any luck with Strategis.

PMB (Print Measurement Bureau) is a good source but when I searched for deli or delicatessen, I only found entries for cheese and not meats. The two 'cheese' entries were under the Groceries Section where a number of chains were identified. There are also entries for types of meat consumed.

Tuesday, February 15, 2005

Schizophrenia and Prison Inmates

Question

A researcher at McMaster is working on a project about the "Economic cost of Schizophrenia in Canada". Based on the literature review about 5% of inmates have a diagnosis of schizophrenia. The researcher needs data on the total inmate days for all the prisoners and average cost per inmate day for the fiscal year of 2003/4 (or most current year data is available) to calculate the cost of incarceration for persons with schizophrenia in Canada. Does anyone know where to find this data?

Answer

From what I am aware of we don't collect inmate data. You will have to contact the research branch at Correctional Services Canada, they should be able to give you the data for federal inmates. If your researcher is also interested in provincial data you
will have to contact each provinical ministry for that data.
http://www.csc-scc.gc.ca

Tuesday, February 8, 2005

1986 EA Boundary File

Question

Does anyone happen to have EA boundary file for the 1986 Census, which they could provide to me? Or am I missing this on the ftp site?

Answer

Turns out you can't find them because they don't exist! Census Tract is as low as they go for 1986 (even for the division).

Thursday, February 3, 2005

Finding concordance between 2001 DAs and 1996 EAs

Question

We compiled the total population in 2001 by provincial electoral district in working with DAs (dissemination areas). In order to calculate population increase by provincial electoral district between 1996 and 2001, I need to compile population by provincial electoral district in 1996. Therefore, I would need the concordance between DAs in 2001 and EAs (enumeration areas) in 1996, because DAs didn't exist in 1996. Does such a tool exist? Or has anybody an idea about how I can calculate in another manner the population increase between 1996 and 2001 by provincial electoral district?

Answer

The version of GeoSuite on the DLI FTP site contains the EA to DA correspondence table. After you install GeoSuite, you'll find in its directory two files: ea-sd_corr.mdb (which is an MS Access database) and ea-sd_corr.txt (which is an ASCII text file that is actually in CSV format). The heading of the fields in these files are: 1996EAuid, DAuid, PartFlag, BLOCKuid, EApop1996, EApop2001.