# T- Test

T-test are used when you have to test the hypotheses of an unknown population mean, it  is used as a substitute.

a t distribution is computed through using sample variance . this is due to t- test using variance instead of error and the fact that t is used to test the difference between two means  .t distribution are often bell-shaped and   have a definite mean of zero . the shape if the t distribution changes due tho the degrees of freedom: the larger the degrees f freedom the closer to  z the t distribution looks even though  they are known to have more variability, be flatter and more spread out  than z – distributions which hs a more central peak. the flatness of the t is caused by the variability of  the scores in the distribution.

although t-test are exactly as great as they sound – they have their complications to for example testing  variable and finding causation.so how useful are they, a study that helps ys e answer this very question  as it shows that t-test help us to see whether the null hypothesis is true  rest and thus has  any given  effect, it also dictates the variability af a sample and the effect of  sample size has on this.  this study plainly highlights the strength of such test in cases where there are a small amount if variables pitted against each othe as : t-test become more complex  as more variables are added to the mix.

http://beheco.oxfordjournals.org/content/14/3/446.short

# blog comments for wednesday 14th

raw2392.wordpress.com

ryan1392.wordpress.com

sarric2392.wordpress.com

psuc9c.wordpress.com

# CHi squared

Chi- squared is also know as the goodness of fit, this means that it tests the shape and proportion of the data in relation to the null. the goodness of fit if tested by comparing the observed frequencies with the distribution set in  the null hypothesis.

chi squared in my eyes is the most understandable of all the  statistical tests as there is no need to calculate sample means you just use calculation of individuals in each category . this is a result of the observed frequency ; the number of individuals in each category.

I believe tha chi- squared is one of  the most helpful statistical tests as it is able to test the relationship between to variables thus enabling causality. this is done by evaluating the between the two variables.

David schoenfeild conducted an experiment using this statistical test. these  tests were based both expected and observed frequencies and whether these covariets fall within the hypothetical L. in this study similarly to clinical trails there had the be partitions within the observable data in order to tell significance . here chi squared helped as it tested the individual components of the data in order for a correlation between Cox’s proposed model and the for the proportional hazards regression model .

http://biomet.oxfordjournals.org/content/67/1/145.short