1. The definition of hypothesis testing
It is used to determine sample to sample, and the sample overall difference is caused by sampling error or statistical inference method is essentially caused by the difference.
2. significant test principle
It is to make certain assumptions about the general characteristics of the study sample and then through statistical inference, this hypothesis should be rejected or accepted to make inferences.
3. The basic idea of hypothesis testing
Reductio ad absurdum and small probability principle
Reductio ad absurdum is to test the hypothesis put forward, then the appropriate statistical methods, the use of small probability principle, to determine whether the hypothesis was established, small probability principle refers to a small probability event once again experiment basically does not occur.
4. hypothesis testing two types of errors
Type I error: null hypothesis is correct, but wrong to refuse it, that "true refused" error, the probability of its occurrence probability of error is first class.
Type II error: the original assumption is incorrect, but the error did not refuse it, namely, "The Pseudo" error, the probability of its occurrence as a Type II error probability.
5. hypothesis testing reasons make two types of errors.
Small probability time is not impossible, but the probability of its occurrence is very small, we can not completely exclude the possibility of its occurrence.
Step 6. hypothesis testing
Determine the appropriate null hypothesis and alternative hypothesis
Select the test statistic, and calculate its value with the selected sample to obtain the test statistic observations.
Calculating test statistic probability of occurrence of observed values, namely p-value.
Given the significance level a, and make decisions. If p <a, the null hypothesis is rejected, on the contrary, there is no reason to reject the null hypothesis.
7. Mean process concept.
One useful method for describing and analyzing scale variables, you can get the required analysis of many variables and statistical indicators of central tendency discrete trends, and he can compare different groups or cross-category.
8. Process Mean effect.
The process may calculate one or more independent variables category due subgroup averages and related univariate statistical variables, you can also get one-way ANOVA and correlation test limit line from the process.
9. The process can be divided from the mean of each category of each grouping variable selection of many statistics subgroup.
Together, the number of cases, mean, median, the group median, standard error of the mean, minimum, maximum, range, a first category variable value of the variable packet, the last packet of a category variable value of the variable , standard deviation, variance, kurtosis, standard error of kurtosis, the skewness, standard error of skewness, percentage, the percentage of the total number of percentage, and the percentage of the sum of the number, geometric mean, harmonic mean.