Guest guest Posted June 11, 2001 Report Share Posted June 11, 2001 Aloha Moses, This will be my last reply to you, barring any additional data of significance being provided by you. If you give the birthdata of all the 52 astrologers you used, we can look at the probabilities more logically. In statistics, the sample set is selected very carefully. It is not just the number of samples that matters. The composition of the sample set is also important. Were your 52 samples ideally scattered in the sample space with respect to all the key degrees of freedom? If not, it will introduce a bias in the probability distribution. This is particularly an issue when the sample set is small. If you give all the birthdata, I may be able to elaborate. > Aloha Narahimsa, It is Narasimha. > > people. If we are expecting 35-40% at each degree (among all people) > > and get 49% at one degree and 10% at another degree, it means nothing. > > Again, you're quoting the wrong data. I am not "quoting" anything. That was just an example. > be a bit higher than this, but not much) i only came up with one degree > around 49.0% of the time. The next closest degree was more than 10 percentage > points less, and none of the other degrees around the 49.0% degree came up > that high. That's enough to make me inquire more, which for some reason you > seem to have a problem with. I have no problem with it and constantly said "good luck and keep me informed". However, I only pointed out that there is nothing statistically significant so far. May Jupiter's light shine on us, Narasimha Quote Link to comment Share on other sites More sharing options...
Guest guest Posted June 12, 2001 Report Share Posted June 12, 2001 Dear Moses and friends, If you don't mind I want to step into this descussion. Narasimha might have been worng in some calculations (say in "quoting" 46%). But, he was trying to make a valid point which was not taken correctly by you. In epidimiology (a statistical study of reasons for cetain decease or effect of a medicine etc), we generally take a control sample set and a sample for test. Both the samples should come from similiar population. For example, if a medice is to be tested for its effectiveness for a particular decease, 2 sets of people from similar background (social back ground and similiar ages) are selected. One set is given the medicine. The other set is not given medicine. Then looking at the proportion in both the sets who got effected by decease, the effectivity of medicine is judged. When Narasimha was saying if 42 of them were born in this period .., he was basically having control set in mind (I guess, he may not have it exactly like that, but the idea which was troubling him essentially this). If you consider 52 astrologer, take 52 non-astrologers data (non-astrologer 52 data must be similiar to 52 astrologer data. i.e. If you have a astrologer born at 1-Jan-1900, take one non-astrologer around the that date). Now what ever you do you can compare on the two sets. Hope you got my point. Here this control set is mainly to eliminate extrenious factors like time from where the data is coming and other environmental factors. Please, do not take these comments as trying to put down your work. We are only trying to give you better idea on how to do statistical research. If Narasimha, has given the wrong figures, he might have mis-calculated. Take the point he is trying to make. If it has any valid point, take it. When Narasimha says your work does not have any statistical significance, it is not end of it. It either means that he did not understood your experiment properly or there is some problem with the experiment. Do not get annoyed by the numbers. Try to understand what he is saying. Hope I could clarify on control set. If I consider people born in 1900, 1912, 1924, 1948, 1960, 1984, 1996, 2008, 2020, 2032; all of them have Jupiter in one sign (mostly), but, I considered people from 1900-2032! (Again I am not saying you have take data like this. But about Jupiter we now very much. There may be some such problem with in the data). It is very defficult to calculate correct probabilities in this situation. So, taking a control set will ease it to some extent. Please do not get upset. We are not trying to show-down your work. Your effort is excellent. I would really appreciate it. Please keep us informed of any studies you do. It is really exciting. Best regards, Vijay. pvr (AT) mediaone (DOT) net wrote: Aloha Moses, This will be my last reply to you, barring any additional data of significance being provided by you. If you give the birthdata of all the 52 astrologers you used, we can look at the probabilities more logically. In statistics, the sample set is selected very carefully. It is not just the number of samples that matters. The composition of the sample set is also important. Were your 52 samples ideally scattered in the sample space with respect to all the key degrees of freedom? If not, it will introduce a bias in the probability distribution. This is particularly an issue when the sample set is small. If you give all the birthdata, I may be able to elaborate. > Aloha Narahimsa, It is Narasimha. > > people. If we are expecting 35-40% at each degree (among all people) > > and get 49% at one degree and 10% at another degree, it means nothing. > > Again, you're quoting the wrong data. I am not "quoting" anything. That was just an example. > be a bit higher than this, but not much) i only came up with one degree > around 49.0% of the time. The next closest degree was more than 10 percentage > points less, and none of the other degrees around the 49.0% degree came up > that high. That's enough to make me inquire more, which for some reason you > seem to have a problem with. I have no problem with it and constantly said "good luck and keep me informed". However, I only pointed out that there is nothing statistically significant so far. May Jupiter's light shine on us, Narasimha Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You are posting as a guest. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.