Question:
In the past DLI was only one way to get access to PUMFS, is this still the case? What is the difference between the DLI and the “Access to PUMF Collection” subscription described here https://www.statcan.gc.ca/eng/help/microdata?HPA=1 ?
As well, in the past, individual researchers not affiliated with the DLI could request copies of a PUMF directly through the STC website free of charge, is this still the case? For example order links are available on public PUMF product webpages see: http://www5.statcan.gc.ca/olc-cel/olc?ObjId=89M0013X1991001&ObjType=46&lang=en
Answer:
The difference between the DLI and access to the PUMF collection comes down to the additional resources that are provided to subscribing institutions. DLI clients get access to a dedicated reference desk for DLI contacts, a forum to discuss data issues and concerns, access to a network of Canadian data librarians, as well as access to training and outreach sessions. The DLI is only open to Canadian postsecondary institutions. In contrast, the access to the PUMF collection is available to all institutions. The purpose of the PUMF collection is an amalgamated licence that gives subscribing institutions unlimited access to all PUMFs. Both programs provide access to data through the corporate Electronic File Transfer (EFT) service as well as through the web data portal, Nesstar.
Users can request PUMFs from Subject Matter areas for free, but each request has an associated licence agreement that the user must sign. While PUMFs are free, the subscription based fee supports the technical infrastructure access, licence administration fees and user support.
Monday, May 14, 2018
Thursday, May 10, 2018
2017 PCCF Questions
Question 1:
A researcher is using the 2017 PCCF, and has noted some issues:
“It appears that there are a significant number of postal codes missing. I had thought that these might be retired postal codes but there are none in the retired file for Manitoba (which is also suspicious). I had not thought about this earlier when I was just connecting postal/census areas and then doing summaries but in the last week we needed to do some work with specific postal codes and found a drop from the 2011 files.
Examples would be Churchill (R0B0E0), Stonewall/Teulon (R0C3B0), some in Portage (R1N3C4, R1N3C2), St. Laurent (R0C1P0) (and 300 others). These are currently active postal codes checking with Canada Post and existed in the last PCCF.”
Would it be possible to have someone look into this?
Answer 1:
We are aware that many postal codesOM did not go through our geocoding process in the June version, and therefore did not make it to the PCCF file. We did a lot of work on getting these postal codesOM back into the product file. We have also appended records that were coded to only the CSD/SGC to the PCCF file. As these do not currently geocode (to at least a Dissemination Area), they do not form part of the core PCCF product, but can be useful for using postal codes to get to CSD.
Also, since the postal code is intended for the distribution of mail by Canada Post the files undergo changes on a regular basis. In some instances, postal codes are retired and re-birthed with a different deliver mode type or the address information for postal codes is changed.
Question 2:
The researcher has responded:
“I only sent a small number of postal codes and they are relatively important, at least in Manitoba. I can send the whole list if necessary.
I am thinking the numbers are large enough and the areas of interest that it should be checked. If this is an issue that is a more general problem (e.g. Canada wide) not just in our little province with a few hundred postal codes.
The postal codes are not retired. I realize there are changes, but the bulk of the codes should not have moved. Many [I have not checked all} of them existed in the 2011 associated PCCF files.
Interesting comment regarding the CSD given some rural delivery, I am not sure I am interpreting what this means. Some postal codes that represent larger areas or only CSD are not included? “
Since I have not seen any other similar inquiries, I too wonder if this is just a Manitoba issue? The user has offered to send the whole list of missing postal codes. Would that be of any use, or would Subject Matter prefer to communicate directly with the user?
Answer 2:
We have received the following response from Subject Matter:
“The postal code team geocodes/links the postal codes as received from Canada Post Corporation to a census geography. They try to link at the lowest level possible, a block face. If they can’t do that, they move up to try to link to a dissemination block, if not then a dissemination area, if not then a CSD. Because the CSDs, especially in rural areas, can be very large, so if the only link between a postal code and a census geography is at the CSD level, it is not automatic that these postal codes are included in the PCCF. The team has made efforts to include more of them manually.
We are not saying that postal codes that represent rural CSDs are not included. That is a misinterpretation.
We were not able to provide specifics about why these particular postal codesOM were not found in the Dec 2017 release of the PCCF. However, we checked each against the newest internal release of the product, and they did exist in that file. We are hoping to disseminate a new version of the file sometime this summer. The postal codesOM in question should be available in that new file."
A researcher is using the 2017 PCCF, and has noted some issues:
“It appears that there are a significant number of postal codes missing. I had thought that these might be retired postal codes but there are none in the retired file for Manitoba (which is also suspicious). I had not thought about this earlier when I was just connecting postal/census areas and then doing summaries but in the last week we needed to do some work with specific postal codes and found a drop from the 2011 files.
Examples would be Churchill (R0B0E0), Stonewall/Teulon (R0C3B0), some in Portage (R1N3C4, R1N3C2), St. Laurent (R0C1P0) (and 300 others). These are currently active postal codes checking with Canada Post and existed in the last PCCF.”
Would it be possible to have someone look into this?
Answer 1:
We are aware that many postal codesOM did not go through our geocoding process in the June version, and therefore did not make it to the PCCF file. We did a lot of work on getting these postal codesOM back into the product file. We have also appended records that were coded to only the CSD/SGC to the PCCF file. As these do not currently geocode (to at least a Dissemination Area), they do not form part of the core PCCF product, but can be useful for using postal codes to get to CSD.
Also, since the postal code is intended for the distribution of mail by Canada Post the files undergo changes on a regular basis. In some instances, postal codes are retired and re-birthed with a different deliver mode type or the address information for postal codes is changed.
Question 2:
The researcher has responded:
“I only sent a small number of postal codes and they are relatively important, at least in Manitoba. I can send the whole list if necessary.
I am thinking the numbers are large enough and the areas of interest that it should be checked. If this is an issue that is a more general problem (e.g. Canada wide) not just in our little province with a few hundred postal codes.
The postal codes are not retired. I realize there are changes, but the bulk of the codes should not have moved. Many [I have not checked all} of them existed in the 2011 associated PCCF files.
Interesting comment regarding the CSD given some rural delivery, I am not sure I am interpreting what this means. Some postal codes that represent larger areas or only CSD are not included? “
Since I have not seen any other similar inquiries, I too wonder if this is just a Manitoba issue? The user has offered to send the whole list of missing postal codes. Would that be of any use, or would Subject Matter prefer to communicate directly with the user?
Answer 2:
We have received the following response from Subject Matter:
“The postal code team geocodes/links the postal codes as received from Canada Post Corporation to a census geography. They try to link at the lowest level possible, a block face. If they can’t do that, they move up to try to link to a dissemination block, if not then a dissemination area, if not then a CSD. Because the CSDs, especially in rural areas, can be very large, so if the only link between a postal code and a census geography is at the CSD level, it is not automatic that these postal codes are included in the PCCF. The team has made efforts to include more of them manually.
We are not saying that postal codes that represent rural CSDs are not included. That is a misinterpretation.
We were not able to provide specifics about why these particular postal codesOM were not found in the Dec 2017 release of the PCCF. However, we checked each against the newest internal release of the product, and they did exist in that file. We are hoping to disseminate a new version of the file sometime this summer. The postal codesOM in question should be available in that new file."
Tuesday, May 8, 2018
Recoding occupations into social classes
Question
Basic sociology question. I have a student asking me how to recode professions (NOC-S) into "social classes" (upper, middle, working class). He wanted to do it with those 10 categories. I told him there is no way you can do that with such broad domains of activity since the lawyer is in the same category as the legal assistant and an administrator is in the same group as the secretary. So he asked me what about those 30 categories. It still looks problematic to me. You have more distinctions between professional and technical occupations but how do you define what fits into "middle class"? So to anyone who is familiar with this, my question is are there "official norms" to divide the National Occupational Classification into "social classes" and what level of detail is necessary to do so? I found this article of Boyd in the Canadian Review of Sociology (appendix B) that defines the Boyd-NP scores for the 2001 census. Are those « officially » recognized and can they be adapted to the 2011 NHS?
Basic sociology question. I have a student asking me how to recode professions (NOC-S) into "social classes" (upper, middle, working class). He wanted to do it with those 10 categories. I told him there is no way you can do that with such broad domains of activity since the lawyer is in the same category as the legal assistant and an administrator is in the same group as the secretary. So he asked me what about those 30 categories. It still looks problematic to me. You have more distinctions between professional and technical occupations but how do you define what fits into "middle class"? So to anyone who is familiar with this, my question is are there "official norms" to divide the National Occupational Classification into "social classes" and what level of detail is necessary to do so? I found this article of Boyd in the Canadian Review of Sociology (appendix B) that defines the Boyd-NP scores for the 2001 census. Are those « officially » recognized and can they be adapted to the 2011 NHS?
A Socioeconomic Scale for Canada: Measuring Occupational Status from the Census
Monica Boyd, University of Toronto
Canadian Review of Sociology
https://doi.org/10.1111/j.1755-618X.2008.00003.x
Answer 1:
There used to be two scales - Pineo-Porter classifications, and the Blishen scale – each SOC code was assigned a value which mapped to “class”. If there is a crosswalk between NOC and SOC, it might be able to update Pineo-Porter classifications or the Blishen codes to the NOC … or someone might even have already done so!
Take a look at https://www.jstor.org/stable/20460616?seq=1#page_scan_tab_contents –
A Scale of Occupational Prestige in Canada, Based on NOC Major Groups
John Goyder and Kristyn Frank
The Canadian Journal of Sociology / Cahiers canadiens de sociologie
Vol. 32, No. 1 (Winter, 2007), pp. 63-83
Published by: Canadian Journal of Sociology
DOI: 10.2307/20460616
Stable URL: http://www.jstor.org/stable/20460616
Page Count: 21
Answer 2:
Subject Matter does not use the term ‘social classes’ when classifying occupations so there is no ‘official norm’ at Statistics Canada.
For the ‘Canadian Review of Sociology’ article we couldn’t open the full article and not sure what is mentioned in appendix B.
But from the text in the ‘Abstract’ it looks like the article mentioned ‘Nam-powers-Boyd’ method was used for Census of Occupation and not Census of Population.”
Answer 1:
There used to be two scales - Pineo-Porter classifications, and the Blishen scale – each SOC code was assigned a value which mapped to “class”. If there is a crosswalk between NOC and SOC, it might be able to update Pineo-Porter classifications or the Blishen codes to the NOC … or someone might even have already done so!
Take a look at https://www.jstor.org/stable/20460616?seq=1#page_scan_tab_contents –
A Scale of Occupational Prestige in Canada, Based on NOC Major Groups
John Goyder and Kristyn Frank
The Canadian Journal of Sociology / Cahiers canadiens de sociologie
Vol. 32, No. 1 (Winter, 2007), pp. 63-83
Published by: Canadian Journal of Sociology
DOI: 10.2307/20460616
Stable URL: http://www.jstor.org/stable/20460616
Page Count: 21
Answer 2:
Subject Matter does not use the term ‘social classes’ when classifying occupations so there is no ‘official norm’ at Statistics Canada.
For the ‘Canadian Review of Sociology’ article we couldn’t open the full article and not sure what is mentioned in appendix B.
But from the text in the ‘Abstract’ it looks like the article mentioned ‘Nam-powers-Boyd’ method was used for Census of Occupation and not Census of Population.”
2015 Household and the Environment Survey Weights
Question:
I have a student working with the 2015 HES survey and asked about the weights available for the PUMF. The user guide states: "The sample design used for the HES 2015 was not self-weighting. When producing simple estimates including the production of ordinary statistical tables, users must apply the proper survey weights."
The only weight variable available in the PUMF is the WTHP: PUMF, survey weight of household. Am I correct that this is what the user guide refers to as the “proper survey weights” and that there are no other weights variables for this PUMF?
Answer:
The client interpretation is correct.
Follow-up Question:
The student has asked why there is a difference between the HES CANSIM tables and the HES PUMF with the weights applied. Can Subject Matter provide any information on this?
Follow-up Answer:
“The Households and the Environment Survey reports on many values in terms of “all households”, as in “x% of Canadian households that had a thermostat indicated it was programmable”. In general, this is calculated as follows:
In other words, all of the households that are in-scope to be asked a question appear in the denominator (in this example households that had a thermostat). That a household might have said “don’t know” (DK) or refused to answer the question (RF) doesn’t change the fact that they were in-scope for the question and thus should be in the denominator.
“All households” is a special case because it includes all households regardless of whether a respondent might have answered “don’t know”, refused to provide an answer to the question or was a “not stated” (was not administered the question for some reason other than flow), so the calculation is more like this:
However, it appears that calculates percentages slightly differently, more along these lines:
The households that were “don’t know”, refused to provide an answer to the question are not taken into consideration in the denominator.
This can lead to different estimates, which is what we’re seeing here. For the specific estimate that the researcher is asking about, here’s what I found:
I hope this sheds some light on the situation.”
I have a student working with the 2015 HES survey and asked about the weights available for the PUMF. The user guide states: "The sample design used for the HES 2015 was not self-weighting. When producing simple estimates including the production of ordinary statistical tables, users must apply the proper survey weights."
The only weight variable available in the PUMF is the WTHP: PUMF, survey weight of household. Am I correct that this is what the user guide refers to as the “proper survey weights” and that there are no other weights variables for this PUMF?
Answer:
The client interpretation is correct.
Follow-up Question:
The student has asked why there is a difference between the HES CANSIM tables and the HES PUMF with the weights applied. Can Subject Matter provide any information on this?
Follow-up Answer:
“The Households and the Environment Survey reports on many values in terms of “all households”, as in “x% of Canadian households that had a thermostat indicated it was programmable”. In general, this is calculated as follows:
In other words, all of the households that are in-scope to be asked a question appear in the denominator (in this example households that had a thermostat). That a household might have said “don’t know” (DK) or refused to answer the question (RF) doesn’t change the fact that they were in-scope for the question and thus should be in the denominator.
“All households” is a special case because it includes all households regardless of whether a respondent might have answered “don’t know”, refused to provide an answer to the question or was a “not stated” (was not administered the question for some reason other than flow), so the calculation is more like this:
However, it appears that
The households that were “don’t know”, refused to provide an answer to the question are not taken into consideration in the denominator.
This can lead to different estimates, which is what we’re seeing here. For the specific estimate that the researcher is asking about, here’s what I found:
Weighted (WTHP)
|
Reported a forced air furnace as a primary heating system (%)
| |
Reported a forced air furnace (numerator)
|
4,083,754
| |
All households (i.e. all HH in-scope for the question about primary heating systems)
|
5,280,396
|
77%
|
5,144,177
|
79%
|
Thursday, May 3, 2018
Data on post-secondary students
Question:
I have a request for data aggregated to show where post secondary students live by small area geographies in southern Ontario. The largest area that would be useful is census subdivision, but census tracts would be ideal. Does anyone know if this data is available? Would it be a custom tabulation? I think the data may be collected in the Post-secondary Student Information System.
If it is possible to get this data we would be interested in having it broken down by institution type, Classification of Instructional Programs (Primary Grouping) and Student Status (international/ Canadian Students).
I have a request for data aggregated to show where post secondary students live by small area geographies in southern Ontario. The largest area that would be useful is census subdivision, but census tracts would be ideal. Does anyone know if this data is available? Would it be a custom tabulation? I think the data may be collected in the Post-secondary Student Information System.
If it is possible to get this data we would be interested in having it broken down by institution type, Classification of Instructional Programs (Primary Grouping) and Student Status (international/ Canadian Students).
Answer:
Data at the level of provinces is the smallest geographic area that we can provide from the Postsecondary Student Information System. We can provide data about students whose permanent address upon admission is Ontario.
Data at the level of provinces is the smallest geographic area that we can provide from the Postsecondary Student Information System. We can provide data about students whose permanent address upon admission is Ontario.
Wednesday, May 2, 2018
WDS server
Question:With difficulty, I found the link to the WDS server yesterday, but have not managed to repeat the feat today.
It is described in the Survival Guide:
Why is this so hard to find? (I may be the only one with this problem, but I hope not.)
Answer:
It's behind the "Restricted access ..." link. The way I remembered this is that I have to have the link proxied to add it to my own web pages. It's not an open product.
It is described in the Survival Guide:
Why is this so hard to find? (I may be the only one with this problem, but I hope not.)
Answer:
It's behind the "Restricted access ..." link. The way I remembered this is that I have to have the link proxied to add it to my own web pages. It's not an open product.
Tuesday, May 1, 2018
Tuition and Living Accommodation Costs 2017-18
Question:
An administrator is looking for the 2017-18 iteration of these files. I checked on Odesi and on the Statistics Canada B202/20 server and could not see this version. In theory the data has been release in September 2017 (http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=3123) . When can we expect the files?
Answer:
We located the files. The Tuition and Living Accommodation Costs (TLAC) for 2017-18 (ver09) have been available since October 5th 2017 on the DLI EFT site.
Here is the file location: /MAD_DLI_IDD_DAM/root/other_autres/3123_TLAC_FSS/
The files will also be uploaded to WDS shortly.
An administrator is looking for the 2017-18 iteration of these files. I checked on Odesi and on the Statistics Canada B202/20 server and could not see this version. In theory the data has been release in September 2017 (http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=3123) . When can we expect the files?
Answer:
We located the files. The Tuition and Living Accommodation Costs (TLAC) for 2017-18 (ver09) have been available since October 5th 2017 on the DLI EFT site.
Here is the file location: /MAD_DLI_IDD_DAM/root/other_autres/3123_TLAC_FSS/
The files will also be uploaded to WDS shortly.
Subscribe to:
Posts (Atom)