<p>Same nature as the psychology thread (which seems to be rather successful). Ask/pose questions, answer questions, or just chime in with useful information.</p>
<p>I'll start off:</p>
<p>What conditions must be satisfied in order to perform a two-sample T-test?</p>
<p>um. unkown standard deviation
both samples independant and fairly normal</p>
<p>what's the difference between the two chi-square tests?</p>
<p>there are three: goodness of fit, independence, and homogeneity of proportions.
In GOF, single variable, single sample. In independence, two variables, single sample. Homogeneity of proportions compares samples from multiple populations to determine significance of differences.</p>
<p>For proportions, When sample p-hat is not given for finding the number needed to get a certain margin of error, what proportion do you normally use? (p-star? or something like that)</p>
<p>(I don't know if I worded that correctly or not)</p>
<p>uhh thats a z star...it depends on what confidence interval they want... (ex: 95% interval = z star of 1.96)....but why ask questions when u can have my sheet?</p>
<p>what sheet?</p>
<p>Actually .. I think this idea is pretty cool ^^ Last year, the AP Bio study thread did a similar thing and it stretched like over 10 pages long.</p>
<p>shall we continue? =P</p>
<p>You have a set of data. If you times each x-value by A and add B to each x-value, how does it affect the mean and standard deviation of the data?</p>
<p>The mean is now x multiplied by A plus B.</p>
<p>The standard deviation is multiplied by A.</p>
<p>When do you used a matched-pairs test?</p>
<p>When answering hypothesis test for a difference b/t 2 means, when we use t-test where population variances are not stated in the question, which occasion do we use df=(n1-1)+(n2-1), and which occasion do we use
df=(smaller of n1-1 and n2-1)?</p>
<p>WE NEED TO POST ALL OF THE CONDITIONS FOR EVERY TEST. If I can't get these, they will certainly be my fait accompli on the free-response section.</p>
<p>for Jillpod's earlier question, its 0.5 when they don't give you a suggested prop. to use.</p>
<p>DorisZ, I honestly don't know...
Besides using a chi-squared test for independence, what is another method you can use to determine independence in an n x n table?</p>
<p>For the t-test</p>
<p>If it's less than 10 things, it should have a normal distibution with no outliers.
If it's between 10 and 40, then it should have a normal distribution, outliers are okay
And if it's more than 40, then anything is okay</p>
<p>Is that right?</p>
<p>Doris: We use the df from the SMALLER n just to be safe...ALWAYS.</p>
<p>yeah, someone please post the conditions for all the tests</p>
<p>Conditions for all the tests are here:
<a href="http://www.mshipke.com/teachers/apstatresources.html#conditions%5B/url%5D">http://www.mshipke.com/teachers/apstatresources.html#conditions</a>
(bottom of page)</p>
<p>In the free response, say you are doing a two-sample z-test. Is it ok to write: (from calculator) z=2.@#^ P-value= .@#^<br>
Or do you have to write out the equation, like (x1-x2)/ (s2/n + s2/n) *sdfhia) <--you know, just whatever the reall equation is.</p>
<p>I would like to not have to write out the equation because it is faster and I do not allways get the right equations, even if the answer from the calculator is right.</p>
<p>I don't know if it's REQUIRED that you write the equations, but on all the student responses I've seen from previous tests, they always wrote them down. And it usually looks like this:</p>
<p>Say because (insert conditions), I will do z-test. Then write the general equation (x-u)/standard dev w/symbols, and rewrite it with the actual numbers plugged into the equation.</p>
<p>Then let your calculator do the work, and say: using the calculator, we find that the p-value is etc.</p>
<p>When do we need to use a confidence LEVEL -defined as if this sample were repeated many times, 9X% of the intervals would capture the true mean.
What sort of question would require an answer like this.</p>
<p>When its like:
What does a 95% confidence level mean?</p>
<p>Ok, my question is what is the difference in conditions for a proportion Confidence interval and Sig test.</p>