Measurement and experimentation

Discussion in 'General Slingshot Discussions' started by Enoch1970, Aug 13, 2012.

  1. Enoch1970

    Enoch1970 New Member

    12
    0
    0
    Hi Everyone, <br><br>I have been a bench scientist for about 20 years now. I really enjoy seeing people who are not professionally involved with science, engage in scentific studies of their own interest. <br><br>I wanted to offer you some tips so you won't get too many arguments against your inventions, and your experimental designs. One of the most basic criticisms a scientist would make against your experiments is that your measurements may be random, because you only measure once. <br><br><ol type="1">
    <br><li>Take at least 3 measurements for each of your data points. (for instance if you want to compare theraband gold against theraband blue, you would want to take chono measurements three times for each band shooting under the same conditions each time)<br>
    </li>
    <li>Calculate a mean and standard error of your measurements for each data point. Scientists like to report data for a small number of data points as mean +/- standard error (mean +/- SE). SE will be the average error (a measure of variability) for the means for your data set. If you use excel this can be retrieved by using the descriptive statistics tool.<br>
    </li>
    <li>Use a statistical test to determine if your experimental groups are statistically different. In excell there are three tests that are handy: t-test assuming equal variance, t-test assuming unequal variance, and paired t-test. The variance as the name implies is a more crude measure of variability of your data set. The variance might be equal if you are comparing two materials of similar composition, but may be unequal if one of the materials is different than the other. The paired t-test is to compare a group against a similar group that has had some intervention. For instance if you wanted to compare the same length therabands against a thereband that you had punched some holes into. The t-test in excell will do the necessary calculations to make the comparison, if your comparison has acheived 'statistical significance' the p-value will be less than 0.05. The smaller this number is the better, and basicly this number describes the chance that this is a real comparison versus a random one.<br>
    </li>
    <li>Display your data in a graph correctly. For most things you can display your data using bar graphs, with error bars indicating the standard error that was calculated initially from your means. After some practice you will be able to tell when you have achieved statistical significance by looking at the SEs on your bar graphs for the various comparisons you wish to make. Time Courses, and dose responses, experiments which your modulating some variable incrementally are best done with line graphs, where each data point is displayed as a point,mean +/- SE<br>
    </li>
    </ol>
    <br>Anyhow I didnt want to go into great detail, but I hope may even make your experiments more fun, because you are using the methods scientists use.
     
  2. JoergS

    JoergS Administrator

    5,803
    109
    63
    I am currently enrolled in a doctorate program at the University of Duisburg-Essen, doing a study about the profitability of influenza vaccination campaigns for employees of mid size corporations. Therefore I have to deal with statistics a lot now, have forgotten so much these last 24 years since I got my master. <br><br>I agree, we need a more scientific approach. I believe that due to the nature of rubber, more than three shots are needed to get reliable data. Also I think that the tests need to be done by machines rather than human shooters, to eliminate human errors.
     

  3. onnod

    onnod Im from Holland, isnt that weird?

    6,001
    18
    38
    oh man this reminds of science class <img src="http://illiweb.com/fa/i/smiles/sleep.gif" alt="Sleep" longdesc="31">(no offence) , but anyway i think you are right
     
  4. Enoch1970

    Enoch1970 New Member

    12
    0
    0
    <table width="90%" cellspacing="1" cellpadding="0" border="0" align="center">
    <tr><td><span class="genmed"><b>JoergS wrote:</b></span></td></tr>
    <tr><td class="quote">I am currently enrolled in a doctorate program at the University of Duisburg-Essen, doing a study about the profitability of influenza vaccination campaigns for employees of mid size corporations. Therefore I have to deal with statistics a lot now, have forgotten so much these last 24 years since I got my master. <br><br>I agree, we need a more scientific approach. I believe that due to the nature of rubber, more than three shots are needed to get reliable data. Also I think that the tests need to be done by machines rather than human shooters, to eliminate human errors.</td></tr>
    </table>
    <span class="postbody"><br><br>I am also working on my doctorate in physiology, it was hard to go back to school for me, cause im about 20 years older than the other students. Anyhow I dont think it would be too hard to validate your current studies, without wrecking the fun <img src="http://illiweb.com/fa/i/smiles/icon_smile.gif" alt="Smile" longdesc="2">. I would love to help in any way I can. Also there is a Youtuber you might enjoy, since you like to build things, check out a guy mdevink, he is a retired airplane pilot from the netherlands, he makes wonderful Stirling Engines.</span>
     
  5. Bert the Welder

    Bert the Welder New Member

    2,201
    0
    0
    I was kind of thinking the same thing regarding accurate, repeatable testing. I'd think some sort of stand to hold the drawn SS would be easy to build. One end to clamp the SS in and a bar/rail going the other way to move a release along to get various, but measurable, distances representing draw length. One of Joerg's mechanical releases with an indexing pin that moves along the length of the rail. This would eliminate slight variations in hand drawn shooting as well as giving a consistent style of release. The base could be a tripod with weight hanging in the middle to give a solid platform from which to shoot. Independent X and Y axis for aiming Could even have an extension out front to mount the crony sensor so it's always in the same place. You'd want it to be built pretty hefty to eliminate flex and vibration. Lamination's of 3/4" plywood should work well. <br>Just thinking out loud.
     
  6. softgoat

    softgoat New Member

    34
    0
    0
    I saw a neat trick some months ago where the fellow was making his draw length reproducible by the simple act of attaching a string to the fork and the pouch. making his shots over the chrony far more precise. Really like these simple tricks that really improve the test. If I stumble on that video again I will post a link.