Notices
Results 1 to 4 of 4

Thread: Statistical Analysis of Lancet Article

  1. #1 Statistical Analysis of Lancet Article 
    Forum Freshman
    Join Date
    Oct 2004
    Location
    Texas
    Posts
    31
    This is statistics more than mathematics but hopefully it will fit in this category. Let me preface this by saying I have no interest in discussing the politics of this or the morality of war - there are other forums for those discussions. :wink:

    I am interested in the analysis of this report (you can read the entire thing here). There has been a lot of support as well as criticism of it's conslusions but my ability to analyze the statistics behind it hamper any attempt to support or contradict those conclusions.

    It has reported confidence interval of (95% CI 8000–194 000). Is that too broad a range? What does this really mean? I hear a lot of talk about the gaussian curve regading this and probablitly of accuracy. :?

    I wonder if the sample size is appropriate to draw the conclusions it draws and if cluster sampling skewed those results.

    In short, I wonder a lot about statistics. I have a good engineering and science background but it's been a long time since those college statistics classes. Since the exit polling debacle this week and how often we see statistics in the news, I think it would be great to figure out a way to seperate the wheat from the chaff in these types of reports so I can understand what's really going on. I picked the lancet article since it's recent and so widely reported.

    Any statisticians here willing to help me look into this and help me answer my questions?


    Reply With Quote  
     

  2.  
     

  3. #2  
    Forum Isotope (In)Sanity's Avatar
    Join Date
    Oct 2004
    Location
    Mesa AZ
    Posts
    2,699
    Sorry Gary, I just don't think anyone will ever post to this thread. It's kind of well...morbid :wink:


    Reply With Quote  
     

  4. #3  
    Forum Sophomore buffstuff's Avatar
    Join Date
    Nov 2004
    Location
    Somewhere over the rainbow...
    Posts
    157
    Its in math so count me out! :P
    Progress isn't made by early risers. It's made by lazy men trying to find easier ways to do something. -Robert Heinlein
    Reply With Quote  
     

  5. #4 Hi Gary I think this would be a useful reply read it. 
    Forum Freshman sara8's Avatar
    Join Date
    Jul 2005
    Location
    Cairo
    Posts
    14
    hi Gary,
    I read your thread in the math forum I think I can help you with some statistical concepts here . The thing is ppl are sometimes are unable to make decisions so they rely on statistics to help them ....what are the information that Statistical Analysis can give to help ppl make decisions ?Somtimes you're given a very big population that you're unable to calculate it's mean or it's variance .Why are mean and variance are important? population mean represents the population Observations it's a measure of central tendency while the population Variance is a measure of measure of dispersion (how much the mean really represents the population observation ) why is the mean and variance important ? because for example we would all want to know the mean life time of human beings the mean capacity an elevator ...all these stuff are important for us .

    How about if we're talking now about the mean life time of human beings ?is it easy to calculate the mean of this population? ofcourse it's very difficult and very expensive ?knowing the mean life time could result in various decisions so the mean is very important here to calculate the Mean of this population we should rely on a very important statistical concept called "SAMPLING DISTRIBUTION" meaning from this given population you choose samples of size n that are not biased (for example you choose a sample of size 5 and choose them in a random fashion (meaning your sample should contain old age median age ,......))then you calculate the sample mean of this same say this sample is sample # 1 --->mean 1 and then you choose another sample you calculate then mean2 and then another sample then you calculate mean3 ...etc by then you'll have mean 1 ,mean2 ,mean3 ,...... isn't now the mean of these samples the random variable we're looking for? by taking the mean of all these means meaning(mean1+mean2+mean3,....)/#of samples we gat the means of these means or the grand mean .This grand mean tends to the population mean if your samples are chosen right . What is the use of all that?


    Well Sometimes you won't be able to collect all these samples for economical reasons and for saving time . Note: as the sample size increase ofcourse the grand mean --->to the population mean because the sample tends to represent the population as the sample size increase
    but at the same time you won't be able to increase the sample size to 100000000 for example because it's costly . Okay now what did statisticians do ? they said we'll choose only one sample and let this sample represents our population for example if we don't know the population mean and we found out that the sample mean we got is 5.5 then we will conclude that the population mean is nearly 5.5(this is not true but I am just explaining) why nearly ??? because maybe the population mean isn't 5.5 for real we only concluded that from the sample what else? we only relied on 1 sample ....that's the use of the so called "Confidence Interval " okay what is the confidence interval ??? let me explain first what the level of Confidence the person who wants to know the mean life time of human beings tell the statistican ....I want you to tell me the population mean or give me an interval in which the population mean will most probably with level of confidence 95% meaing you must be sure 95% that the population mean will lie in this interval "Confidence interval"....Confidence level--->" YOU CAN CHOOSE 1 SAMPLE AND CONCLUDE THE POPULATION MEAN BUT MAKE SURE THAT THIS SHOULD BE 95% TRUE" meaning from alllllll the samples we can take your conclusion should be true with all samples but only 5% of these samples tells that your conclusion is false . alpha=0.05 or the probability that your conclusion is wrong 1-alpha= 0.95 the probability that the population mean will lie in the interval you calculated(interval estimate) .


    A confidence interval is :is the interval that you calculate by knowing the sample mean and the sample size and the level of confidence .You'll be sure after calculating the interval 1-alpha*100% that population mean will lie in this interval.

    Which is ofcourse so important .

    As the level of confidence decrease the Confidence interval "width" decrease (which is more accurate) and As the level of confidence increase the Confidence interval Increase(which is not more accurate) the sample size also plays an important role since as you choose your sample size with size n where n is big it tends more to represent your population so your interval estimation will always be kind of accurate.

    I hope this helps... I suggest that you search google also it'd be very helpful there are many sites about statistics out there

    Sara,
    The things that hurt instruct
    Reply With Quote  
     

Bookmarks
Bookmarks
Posting Permissions
  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •