Saturday, December 12, 2009
Saturday, November 28, 2009
Tuesday, November 24, 2009
Case-Control Study Design
FYI:
1. Here is what we shall discuss (I prophesy:)
2. Then something pertaining to this
3. But ultimately, it will all be about this!
So I argue each of the FYI points in detail (and English) presently:
1. you have two-population-in-one (study population = cases + controls) and what really interests you is the difference (conceptually, statistically & literally) between the two
2. if 1,000 studies like yours (hypothesis, sample size, etc.) are done, each study will estimate the difference, in odds of exposure, between cases & controls and report this as a t-statistic. 1,000 such statistics will follow the distribution outline in the picture (either the red or the green one; we don't know!!!). Your study (which is all we know!!!) will also estimate the difference in odds of exposure between cases & controls & result in the blue vertical line in the picture :) Of course the question becomes a HAMLET style question from this point: "To Blue or To Green, That is the Question!" Red, with a log(odds) difference = 0 is obviously the null distribution. We can easily perceive that larger sample sizes make our lives easier & reduce our risks of making wrong inferences (type I or type II errors:)
3. So #1 deals with an idea (concept). #2 then translates this concept into statistics so that it can be tested in the REAL WORLD which occurs, not as ideas, but as measurable stuff! #3, however, is the biological argument & reasoning. the link i posted takes you to a nutshell LITERALLY. In this case, specifically, it takes you to the bottom left item = DNA. Feel free to click on it & see where it takes you!!! This is where you'd realize that "lagging" your exposure variable is essential.
See an example of a study in which lagging was used analytically:
Mortality from brain cancer was modestly increased among men with < 2000 hours (RR 1.61, 95% CI 0.86 to 3.01) and > 2000-10,000 hours exposure (RR 1.79, 95% CI 0.81 to 3.95), but there were no deaths from brain cancer among the most highly exposed men. A lag of five years yielded slightly increased RRs. Mortality from liver cancer was not associated with exposure to PCB insulating fluids.
1. Here is what we shall discuss (I prophesy:)
2. Then something pertaining to this
3. But ultimately, it will all be about this!
So I argue each of the FYI points in detail (and English) presently:
1. you have two-population-in-one (study population = cases + controls) and what really interests you is the difference (conceptually, statistically & literally) between the two
2. if 1,000 studies like yours (hypothesis, sample size, etc.) are done, each study will estimate the difference, in odds of exposure, between cases & controls and report this as a t-statistic. 1,000 such statistics will follow the distribution outline in the picture (either the red or the green one; we don't know!!!). Your study (which is all we know!!!) will also estimate the difference in odds of exposure between cases & controls & result in the blue vertical line in the picture :) Of course the question becomes a HAMLET style question from this point: "To Blue or To Green, That is the Question!" Red, with a log(odds) difference = 0 is obviously the null distribution. We can easily perceive that larger sample sizes make our lives easier & reduce our risks of making wrong inferences (type I or type II errors:)
3. So #1 deals with an idea (concept). #2 then translates this concept into statistics so that it can be tested in the REAL WORLD which occurs, not as ideas, but as measurable stuff! #3, however, is the biological argument & reasoning. the link i posted takes you to a nutshell LITERALLY. In this case, specifically, it takes you to the bottom left item = DNA. Feel free to click on it & see where it takes you!!! This is where you'd realize that "lagging" your exposure variable is essential.
See an example of a study in which lagging was used analytically:
Mortality from brain cancer was modestly increased among men with < 2000 hours (RR 1.61, 95% CI 0.86 to 3.01) and > 2000-10,000 hours exposure (RR 1.79, 95% CI 0.81 to 3.95), but there were no deaths from brain cancer among the most highly exposed men. A lag of five years yielded slightly increased RRs. Mortality from liver cancer was not associated with exposure to PCB insulating fluids.
Sunday, November 22, 2009
James Joyce
Integritas: "An esthetic image is presented to us either in space or in time. What is audible is presented in time, what is visible is presented in space. But, temporal or spatial, the esthetic image is first luminously apprehended as selfbounded and selfcontained upon the immeasurable background of space or time which is not it. You apprehend it as one thing. You see is as one whole. You apprehend its wholeness."
Consonantia: "Having first felt that its is one thing you feel now that it is a thing. You apprehend it as a complex, multiple, divisible, separable, made up of its parts, the result of its parts and their sum, harmonious."
Claritas: "When you have apprehended it as one thing and have then analysed it according to its form and apprehended it as a thing you make the only synthesis which is logically and esthetically permissible. You see that it is that thing which it is and no other thing. The radiance is the scholastic quidditas, the whatness of a thing."
Consonantia: "Having first felt that its is one thing you feel now that it is a thing. You apprehend it as a complex, multiple, divisible, separable, made up of its parts, the result of its parts and their sum, harmonious."
Claritas: "When you have apprehended it as one thing and have then analysed it according to its form and apprehended it as a thing you make the only synthesis which is logically and esthetically permissible. You see that it is that thing which it is and no other thing. The radiance is the scholastic quidditas, the whatness of a thing."
Friday, November 20, 2009
Subscribe to:
Posts (Atom)
