You may want to take a look at this blog http://blogs.msdn.com/msnsearch/archive/2006/01/20 /515606.aspx to read about Microsoft's reluctant handing over of the search records. User information was scrubbed out, leaving a list of queries with a count of how often they occurred as well as a random set of indexed pages. What this means is that if you searched for "my name is jim smith and I live at XYZ and would like to see child pornography" that maybe you could be concerned. If you queried for "sex with children", your query will be counted as +1 with the 128 other people who searched for that term on day Z. If you searched for your social security number, telephone number etc... the government may have a record of that now...oohhhh scary.
Dr. Kanazawa sent me an email that stated the control variables accounted for 18.48 of the variance for boys and 15.24% of the variance for the girls. This leaves occupation of parent accounting for approximately 1% of the variance in either of the two models, suggesting to me that not much can be made from the project. I was also informed that only one parent's occupation was known for the study, so I am not sure if the children may have been counted twice or not.
I have a few problems with the article currently. First off, I do not understand how they took into account one or both parents with job type X. If they treated each parent as a separate entity and put both into the data set, they have violated the basic assumption of independance of observation in OLS regression (i.e. they count the same result (kid) twice, once for mom and once for dad). Second, the R^2 values for male and female offspring were around.2 which is reasonably low. This was true INCLUDING all the control variables... there may be a significant (but insamely small) relationship here, but because they failed to report the results in a block-regression manner, we just have a bunch of significant varaibles thrown into on R^2 package. I have written the authors for this info and will update if they reply...
I don't see why they shouldn't be open source. As a scientist, I believe that the research I do, which is for the most part, subsdised by the government, should be free to others to access. Peer review ensures that what is published escapes the vortex of pseudoscientific drivel that is out there (I recall a freind showing me a book on "personal magnetism" published in the 1920's that claimed you could send an 'ion burst' from your forehead to affect those who opposed you, stupefying them into submission." Written by a Ph.D. which made it legitimate, of course). Recently I was trying to access some work as a reference to my dissertation, and I was told by the website that I could pay $75 to access the article for 48 hours! Along with that insult came the realization that once done with my degree, my dissertation automatically goes to ScienceDirect, who will then sell copies for $35 (which I get $0). In order to graduate, I MUST release my work to them. This sugests that the current system is less than fair to all, and I believe making it open source would rectify this problem to some degree. The price that institutions pay for journals is rediculous, and really anymore it doesn't matter what journal things are published in. Here is my theory on how it used to be:
Once upon a time, people only read the best journals because they didn't have access to all the journals out there, and given the limited amount of search time and search resources, they stuck to the "goodies". Nowadays, however, with advanced search engines etc... you no longer have this limited access/time/resource problem, which makes the more lenient journals as accessable as the more elite. Surely those poor dolts toiling away to get tenure have to aspire to be published in the A-list journals, but the layman (read: grad student) doesn't really give a flip if it appears in the journal of elite learning or Uncle Bob's Skool Tymes as long as it is on topic, makes sense, and is well written. The peer revies process is a safety net (not perfect of course) for the worst papers, but overall helps to keep the stuff that is published honest. Open source is a great way to share all this important knowledge. Professors should relish it because it means they get cited more often. Students should love it because it's easy to find what they need. The only one who loses is whoever is making bank from the exorbant subscription costs that are currently being charged. Somebody is making money here..and it isn't the typesetters, the statisticians, or the copy-editors. If a journal costs $7000 a year per institution, the first 100 institutions to subscribe more than cover it's yearly costs...
Slashdot Mirror
You may want to take a look at this blog http://blogs.msdn.com/msnsearch/archive/2006/01/20 /515606.aspx
to read about Microsoft's reluctant handing over of the search records. User information was scrubbed out, leaving a list of queries with a count of how often they occurred as well as a random set of indexed pages. What this means is that if you searched for "my name is jim smith and I live at XYZ and would like to see child pornography" that maybe you could be concerned. If you queried for "sex with children", your query will be counted as +1 with the 128 other people who searched for that term on day Z. If you searched for your social security number, telephone number etc... the government may have a record of that now...oohhhh scary.
Dr. Kanazawa sent me an email that stated the control variables accounted for 18.48 of the variance for boys and 15.24% of the variance for the girls. This leaves occupation of parent accounting for approximately 1% of the variance in either of the two models, suggesting to me that not much can be made from the project. I was also informed that only one parent's occupation was known for the study, so I am not sure if the children may have been counted twice or not.
I have a few problems with the article currently. First off, I do not understand how they took into account one or both parents with job type X. If they treated each parent as a separate entity and put both into the data set, they have violated the basic assumption of independance of observation in OLS regression (i.e. they count the same result (kid) twice, once for mom and once for dad). Second, the R^2 values for male and female offspring were around .2 which is reasonably low. This was true INCLUDING all the control variables... there may be a significant (but insamely small) relationship here, but because they failed to report the results in a block-regression manner, we just have a bunch of significant varaibles thrown into on R^2 package. I have written the authors for this info and will update if they reply...
I don't see why they shouldn't be open source. As a scientist, I believe that the research I do, which is for the most part, subsdised by the government, should be free to others to access. Peer review ensures that what is published escapes the vortex of pseudoscientific drivel that is out there (I recall a freind showing me a book on "personal magnetism" published in the 1920's that claimed you could send an 'ion burst' from your forehead to affect those who opposed you, stupefying them into submission." Written by a Ph.D. which made it legitimate, of course). Recently I was trying to access some work as a reference to my dissertation, and I was told by the website that I could pay $75 to access the article for 48 hours! Along with that insult came the realization that once done with my degree, my dissertation automatically goes to ScienceDirect, who will then sell copies for $35 (which I get $0). In order to graduate, I MUST release my work to them. This sugests that the current system is less than fair to all, and I believe making it open source would rectify this problem to some degree. The price that institutions pay for journals is rediculous, and really anymore it doesn't matter what journal things are published in. Here is my theory on how it used to be: Once upon a time, people only read the best journals because they didn't have access to all the journals out there, and given the limited amount of search time and search resources, they stuck to the "goodies". Nowadays, however, with advanced search engines etc... you no longer have this limited access/time/resource problem, which makes the more lenient journals as accessable as the more elite. Surely those poor dolts toiling away to get tenure have to aspire to be published in the A-list journals, but the layman (read: grad student) doesn't really give a flip if it appears in the journal of elite learning or Uncle Bob's Skool Tymes as long as it is on topic, makes sense, and is well written. The peer revies process is a safety net (not perfect of course) for the worst papers, but overall helps to keep the stuff that is published honest. Open source is a great way to share all this important knowledge. Professors should relish it because it means they get cited more often. Students should love it because it's easy to find what they need. The only one who loses is whoever is making bank from the exorbant subscription costs that are currently being charged. Somebody is making money here..and it isn't the typesetters, the statisticians, or the copy-editors. If a journal costs $7000 a year per institution, the first 100 institutions to subscribe more than cover it's yearly costs...