Questions of Research
For several weeks I’ve been reaping texts on research design methodologies and separating the chaff. I now have mountains of intellectual grist in my brain and I feel like I had better actually do something productive before it begins to rot. I’ve decided to write down some research ideas.
The Impact of Zipf
I recently listened to a lecture about bibliometrics and learned about the wonders of Zipf’s law and its scholastic brethren: Lotka’s Law (authors), and Bradford’s Law (Journals). I wonder, with the rise of specialty journals and increased publishing pressure, if Zipf’s law has begun to come off the rails. Do Lotka’s and Bradford’s Law still apply? How do they relate to ISI’s impact factors? If Zipf’s law still does hold, could it be used to discern impending fragmentation of a particular discipline?
Socio-Economic Book Clusters
I’ve been using a library for as long as I can remember. I depend on my public library for three particular genres: science fiction, self-help financial, and pop-guru management (I wouldn’t actually buy these things!). Is there any way that someone could predict my book consuming habits? During my training as a librarian I became aware of the “fiction problem” and learned that book consumption habits can’t be predicted… or can they. It would be interesting to cluster book similarities based on patron borrowing patterns. In addition, if the patrons used branch libraries the clusters could be referenced against the socio-economic conditions of the surrounding neighbourhood. Social profiling may be odious but it would make reader’s advisory much simpler!
Text Obsolescence
Some texts never seem to expire. Others—the Koran or the Book of Mormon, for example—seem to be getting ever more popular. How long do texts last? When is a text obsolete? Citation analysis could yield some very interesting patterns for particular disciplines or fields of study. Implementing a form of content analysis could further enrich the data source. What type of text lasts longer based on citation half lives: monograph or journal article? What type of monograph lasts longer: a description for laymen, an erudite and witty explanation, or a detailed tome.
Information Theory and Technical Indicators
I am a veteran of the dot bomb era. During those years I tracked my stock religiously and believed that my stock options would make me rich. I was even involved in day trading and the use of “technical indicators” and other kabbalistic devices to guide my stock picking decisions. Did these techniques really work? I’m not sure but I’d like to know. One possible analysis involves the use of Information Theory to determine signal/noise ratios. For my particular experiment, I could use a 10-day indicator like the MACD and compile noise calculations for a 10-year period of time that covers the dot bomb era, and the recent wars. The calculations would be based on the probability that the indicator correctly signalled the purchase over a one-month period of time as compared to a moving baseline covering a six-month period. I hypothesize that speculative booms increase the channel noise. Industries will behave differently and certain indicators will be better than others. If I’m investing in high tech, how can I know how noisy the price signal is? Which indicators should I use: Bollinger bands, stochastics, MACD, etc.? Although I think this issue is pretty cool, it also has applications in the LIS world. A similar process could be used to determine the efficiency of n-grams when querying genetic databases. From a media studies perspective, a similar process could be used to discuss Hayek’s assertions regarding price structures and the communication (or lack of communication) of complex information.
On to Content Analysis…
It was interesting to read about content analysis. I feel that content analysis is an important asset for the researcher’s toolchest. I am, however, concerned about the subjective nature of the coding decisions made by the researcher and the difficulties of constructing an “exhaustive and exclusive” classification scheme… blah, blah, blah.
For further discussion of content analysis limitations—and the limitations of research in general—please consult the archives of my postings. I’m getting the itch to move and actually start doing some research rather than criticizing and commenting on the validity of the various techniques.
For several weeks I’ve been reaping texts on research design methodologies and separating the chaff. I now have mountains of intellectual grist in my brain and I feel like I had better actually do something productive before it begins to rot. I’ve decided to write down some research ideas.
The Impact of Zipf
I recently listened to a lecture about bibliometrics and learned about the wonders of Zipf’s law and its scholastic brethren: Lotka’s Law (authors), and Bradford’s Law (Journals). I wonder, with the rise of specialty journals and increased publishing pressure, if Zipf’s law has begun to come off the rails. Do Lotka’s and Bradford’s Law still apply? How do they relate to ISI’s impact factors? If Zipf’s law still does hold, could it be used to discern impending fragmentation of a particular discipline?
Socio-Economic Book Clusters
I’ve been using a library for as long as I can remember. I depend on my public library for three particular genres: science fiction, self-help financial, and pop-guru management (I wouldn’t actually buy these things!). Is there any way that someone could predict my book consuming habits? During my training as a librarian I became aware of the “fiction problem” and learned that book consumption habits can’t be predicted… or can they. It would be interesting to cluster book similarities based on patron borrowing patterns. In addition, if the patrons used branch libraries the clusters could be referenced against the socio-economic conditions of the surrounding neighbourhood. Social profiling may be odious but it would make reader’s advisory much simpler!
Text Obsolescence
Some texts never seem to expire. Others—the Koran or the Book of Mormon, for example—seem to be getting ever more popular. How long do texts last? When is a text obsolete? Citation analysis could yield some very interesting patterns for particular disciplines or fields of study. Implementing a form of content analysis could further enrich the data source. What type of text lasts longer based on citation half lives: monograph or journal article? What type of monograph lasts longer: a description for laymen, an erudite and witty explanation, or a detailed tome.
Information Theory and Technical Indicators
I am a veteran of the dot bomb era. During those years I tracked my stock religiously and believed that my stock options would make me rich. I was even involved in day trading and the use of “technical indicators” and other kabbalistic devices to guide my stock picking decisions. Did these techniques really work? I’m not sure but I’d like to know. One possible analysis involves the use of Information Theory to determine signal/noise ratios. For my particular experiment, I could use a 10-day indicator like the MACD and compile noise calculations for a 10-year period of time that covers the dot bomb era, and the recent wars. The calculations would be based on the probability that the indicator correctly signalled the purchase over a one-month period of time as compared to a moving baseline covering a six-month period. I hypothesize that speculative booms increase the channel noise. Industries will behave differently and certain indicators will be better than others. If I’m investing in high tech, how can I know how noisy the price signal is? Which indicators should I use: Bollinger bands, stochastics, MACD, etc.? Although I think this issue is pretty cool, it also has applications in the LIS world. A similar process could be used to determine the efficiency of n-grams when querying genetic databases. From a media studies perspective, a similar process could be used to discuss Hayek’s assertions regarding price structures and the communication (or lack of communication) of complex information.
On to Content Analysis…
It was interesting to read about content analysis. I feel that content analysis is an important asset for the researcher’s toolchest. I am, however, concerned about the subjective nature of the coding decisions made by the researcher and the difficulties of constructing an “exhaustive and exclusive” classification scheme… blah, blah, blah.
For further discussion of content analysis limitations—and the limitations of research in general—please consult the archives of my postings. I’m getting the itch to move and actually start doing some research rather than criticizing and commenting on the validity of the various techniques.
0 Comments:
Post a Comment
Subscribe to Post Comments [Atom]
<< Home