Jon is a drug policy scholar at CMU. The presentation is part of an academic colloquium on the marijuana market. He did the analysis from the NSDUH and its predecessor household survey, which is obvious if a) you watch the video for 15 minutes or b) are at all familiar with the underlying data. I don't know what you mean by "primary source," unless you want me to point you to the underlying data, in which case, it's not hard to find the NSDUH public microdata. But for most other definitions, "an academic's analysis of public use data" is a fairly reliable source of information. Have a nice day.
I don't have 15 minutes to watch a video. I can read 10 pages of text in that amount of time, scanning for "wait- say that again" lines. I don't need to sit through someone else's interpretation of data when I can review the original data for myself. In its full and appropriate context, which can in turn be compared with other primary source data.
I use text for my references- they're authoritative. Unlike screenshots, they can be easily copied and pasted, and the full context can be instantly recalled with a URL.
"Primary sources provide raw information and first-hand evidence. Examples include interview transcripts, statistical data, and works of art. Primary research gives you direct access to the subject of your research.
Secondary sources provide second-hand information and commentary from other researchers. Examples include journal articles, reviews, and academic books. Thus, secondary research describes, interprets, or synthesizes primary sources.
Primary sources are more credible as evidence, but good research uses both primary and secondary sources..."
Okay. In this context, the primary source is the National Survey on Drug Use and Health, which I already explained to you. Caulkins did an analysis of those data, which is the form that most academic articles take. I actually did a similar analysis here: https://ifstudies.org/blog/pot-and-pathology. But you want a primary source so ... go analyze the NSDUH data. If you need a link, it's in the post.
ah, much better. Now there's a common basis of information I can draw on. As opposed to trying to parse out the exact details of what someone else said on a video to our mutual satisfaction.
Transcripts are where it's at. (As long as they aren't rush C-Span transcripts, which disrespect the medium of text mercilessly, and provide a convincing argument that AI is still hopelessly impaired at deciphering human speech...chess wizard or not, AI can't even call a baseball game.)
Only thing is- no year provided for the NSDUH survey: there are at least two reports under that moniker.
A link to an NSDUH study from 2020 is found only once, highlighted in the first footnote at the bottom of the page. Presumably, that's where I'll find the table in question. But I can't tell yet. If I find it, I'll supply a page number.
"1. The 2020 National Survey of Drug Use and Health cautions against comparing data collected in 2020 to prior yearsтАЩ data, because of changes in survey methodology due to the COVID-19 pandemic. I do so here both because the 2002 тАУ 2019 trend is instructive, and because the 2020 estimate deviates downward from that trend, suggesting that methodology-introduced bias underestimates the true change in heavy use.
2. For a given respondent, I estimate this as total number of past month days of use (the variable MRDAYPMO) multiplied by 12, multiplied by that respondentтАЩs weight, which in the NSDUH public use file is equivalent to how much of the target population that respondent represents. The sum of these is the total number of days of use. This is actually an underestimate, because any users who report using in the past year but not the past monthтАФabout 7.5% of respondents to the 2020 NSDUHтАФare counted by this method as not having used."
So you've done your own interpretation of (a little bit of) the primary source data. Just to be clear. Not that there's anything wrong with that; I've been known to do it myself. I try to stay out of the tall weeds of interpretations that prove difficult to check, though. The weighting game is problematic, unless the data is forthrightly amenable to being parsed and assessed simply in the reading of it.
The data Jon and I use are aggregated from the 2002 to 2019 NSDUH surveys, and Jon also uses the precursor household survey to the NSDUH. We don't get the figures out of the tables in the NSDUH *report*, which is itself not a "primary source" as you've confusingly defined it, i.e. it is an SAMHSA analysis of the raw data from a survey. To reconstruct the figures, you have to download each year's microdata and analyze them individually.
I mean, all of this is very apparent if you have a passing familiarity with how drug use is measured. So I'm not sure why you needed to be rude to me for an extended period of time to get to this point.
The analytic portion of the SAMSHA reports is the text portion that's found in various synopses of subtopics. The raw numerical survey data was compiled and submitted to SAMHSA for publication; those metrics and tables are indisputably primary source data. That includes any tables and metrics found in the text digest portion of the NSDUH report.
The new NSDUH appears to be emphasizing more incorporation of that interpretive material into the content, in comparison with the previous format. We'll see how they do. It certainly took SAMHSA and ONDCP a while to draw much importance from the findings of the early 2000s- which indicated that for a few years running, more 9th and 12th graders were obtaining their first experiences with (non-alcohol, non-tobacco) drugs from diverted prescription pills than from marijuana. They weren't even distinguishing between the types of prescription pills in their survey questions, for a while. (Fortunately, the number of opioid pill-popping teens has since gone into a steep decline- probably less than 1/3 of what it formerly was at the peak level of 10-15 years ago, although given the earlier deficiencies in survey data specifics, only rough estimates are possible. In any event, it's excellent news, even if I've never seen it as a news headline or a TV story. A drug control and drug education success story. But, oh noes, legal pot...)
oh, for crying out loud- you rickroll me to a Youtube presentation?
I don't waste my time on Youtube guided meditations. Youtube is where I go to tune into music clips.
Primary source, please. Text.
Jon is a drug policy scholar at CMU. The presentation is part of an academic colloquium on the marijuana market. He did the analysis from the NSDUH and its predecessor household survey, which is obvious if a) you watch the video for 15 minutes or b) are at all familiar with the underlying data. I don't know what you mean by "primary source," unless you want me to point you to the underlying data, in which case, it's not hard to find the NSDUH public microdata. But for most other definitions, "an academic's analysis of public use data" is a fairly reliable source of information. Have a nice day.
It isn't a primary source.
I don't have 15 minutes to watch a video. I can read 10 pages of text in that amount of time, scanning for "wait- say that again" lines. I don't need to sit through someone else's interpretation of data when I can review the original data for myself. In its full and appropriate context, which can in turn be compared with other primary source data.
I use text for my references- they're authoritative. Unlike screenshots, they can be easily copied and pasted, and the full context can be instantly recalled with a URL.
can you just quickly define "primary source"
Yes.
"Primary sources provide raw information and first-hand evidence. Examples include interview transcripts, statistical data, and works of art. Primary research gives you direct access to the subject of your research.
Secondary sources provide second-hand information and commentary from other researchers. Examples include journal articles, reviews, and academic books. Thus, secondary research describes, interprets, or synthesizes primary sources.
Primary sources are more credible as evidence, but good research uses both primary and secondary sources..."
https://www.scribbr.com/working-with-sources/primary-and-secondary-sources/
That's one site of many, but the sources are agreed on the definition. There's even a Wiki page! https://en.wikipedia.org/wiki/Primary_source
Pro tip: https://duckduckgo.com/?q=%22primary+source%22&t=newext&atb=v336-1&ia=web
Found the answer in under 60 seconds. That's how I use the Internet for research. Not fumbling around in the Twitter sandbox.
I could have supplied the definition myself- some of us learned it in school- but I wanted outside support.
Okay. In this context, the primary source is the National Survey on Drug Use and Health, which I already explained to you. Caulkins did an analysis of those data, which is the form that most academic articles take. I actually did a similar analysis here: https://ifstudies.org/blog/pot-and-pathology. But you want a primary source so ... go analyze the NSDUH data. If you need a link, it's in the post.
ah, much better. Now there's a common basis of information I can draw on. As opposed to trying to parse out the exact details of what someone else said on a video to our mutual satisfaction.
Transcripts are where it's at. (As long as they aren't rush C-Span transcripts, which disrespect the medium of text mercilessly, and provide a convincing argument that AI is still hopelessly impaired at deciphering human speech...chess wizard or not, AI can't even call a baseball game.)
Only thing is- no year provided for the NSDUH survey: there are at least two reports under that moniker.
A link to an NSDUH study from 2020 is found only once, highlighted in the first footnote at the bottom of the page. Presumably, that's where I'll find the table in question. But I can't tell yet. If I find it, I'll supply a page number.
Interesting footnotes on that linked page, too:
https://ifstudies.org/blog/pot-and-pathology
"1. The 2020 National Survey of Drug Use and Health cautions against comparing data collected in 2020 to prior yearsтАЩ data, because of changes in survey methodology due to the COVID-19 pandemic. I do so here both because the 2002 тАУ 2019 trend is instructive, and because the 2020 estimate deviates downward from that trend, suggesting that methodology-introduced bias underestimates the true change in heavy use.
2. For a given respondent, I estimate this as total number of past month days of use (the variable MRDAYPMO) multiplied by 12, multiplied by that respondentтАЩs weight, which in the NSDUH public use file is equivalent to how much of the target population that respondent represents. The sum of these is the total number of days of use. This is actually an underestimate, because any users who report using in the past year but not the past monthтАФabout 7.5% of respondents to the 2020 NSDUHтАФare counted by this method as not having used."
So you've done your own interpretation of (a little bit of) the primary source data. Just to be clear. Not that there's anything wrong with that; I've been known to do it myself. I try to stay out of the tall weeds of interpretations that prove difficult to check, though. The weighting game is problematic, unless the data is forthrightly amenable to being parsed and assessed simply in the reading of it.
The data Jon and I use are aggregated from the 2002 to 2019 NSDUH surveys, and Jon also uses the precursor household survey to the NSDUH. We don't get the figures out of the tables in the NSDUH *report*, which is itself not a "primary source" as you've confusingly defined it, i.e. it is an SAMHSA analysis of the raw data from a survey. To reconstruct the figures, you have to download each year's microdata and analyze them individually.
The public use files can be found here: https://www.datafiles.samhsa.gov/data-sources
Yes. "That's a start", as we intrepid data researchers say.
I mean, all of this is very apparent if you have a passing familiarity with how drug use is measured. So I'm not sure why you needed to be rude to me for an extended period of time to get to this point.
The analytic portion of the SAMSHA reports is the text portion that's found in various synopses of subtopics. The raw numerical survey data was compiled and submitted to SAMHSA for publication; those metrics and tables are indisputably primary source data. That includes any tables and metrics found in the text digest portion of the NSDUH report.
The new NSDUH appears to be emphasizing more incorporation of that interpretive material into the content, in comparison with the previous format. We'll see how they do. It certainly took SAMHSA and ONDCP a while to draw much importance from the findings of the early 2000s- which indicated that for a few years running, more 9th and 12th graders were obtaining their first experiences with (non-alcohol, non-tobacco) drugs from diverted prescription pills than from marijuana. They weren't even distinguishing between the types of prescription pills in their survey questions, for a while. (Fortunately, the number of opioid pill-popping teens has since gone into a steep decline- probably less than 1/3 of what it formerly was at the peak level of 10-15 years ago, although given the earlier deficiencies in survey data specifics, only rough estimates are possible. In any event, it's excellent news, even if I've never seen it as a news headline or a TV story. A drug control and drug education success story. But, oh noes, legal pot...)