Mistakenly rejecting the null hypothesis is a type 1 error. These errors are not avoidable and are part of statistical testing, but we can lessen the occurrence by setting the significance at a lower level. However, by setting the significance level lower; let us say .001, we then increase the chance of type 2 errors. Failing to correctly reject the null hypothesis creates a type 2 error, this is because; according to Aron (2009) “with an extreme significance level like .001, even if the research hypothesis is true, the results have to be quite strong for you to reject the null hypothesis.” To avoid type 2 errors, it is a good idea to increase the size of the experiment. The sample size effects type 2 errors by determining the amount of sampling error. With small samples; effects are harder to detect, by increasing sample size we boost the statistical power of the test. The selection of the alpha affects the type 2 test by causing the researcher to fail when rejecting the H0 if the alpha selection is incorrect.

If the researcher is worried about the possibility of rejecting H0 when it is true, then a smaller alpha level like .01 or even .001 should be used. That would minimize the change that researcher would incorrectly reject H0. Explain a factor that the researcher can control to change the type II error. A researcher wants to find out how many times the red light will flash on an electronic Simon game. He begins an experiment and plays the game 4000 times and counts the number of red light flashes. The null hypothesis is that the game is not set to a counting rotation, and the probability that the red light will flash is one out of four times. For this test the researcher will use a two tailed test because he has not indicated if he wants the true probability to be above or below 1/4. The researcher will then decide the significance level of the test, (5% will be used in this scenario).

In setting the significance level at 5% it shown that the result of the experiment is significantly greater than 1000 red flashes out of 4000 flashes; with this level we know the probability of this occurring is less than 5%. Once the researcher is done with the experiment, he will then be able to say if the game is set to a counting rotation and is biased, or he will be to state the opposite. If the game is fair the researcher has made a Type 1 error (A type 1 error is when we wrongly reject the null hypothesis). In this scenario the researcher set the significance level at 5% so if the researcher reported that the experiment does not provide evidence that the game is biased; but it is, the has then made a type 2 error. The researcher can change the type 2 error by increasing the size of the experiment by playing the game 40000 times instead of 4000.

Reference

Aron (2009). Statistics for Psychology [5] (VitalSource Bookshelf), Retrieved January 11, 2013 from: http://digitalbookshelf.southuniversity.edu/books/0558403867/id/ch01

Courtney from Study Moose

Hi there, would you like to get such a paper? How about receiving a customized one? Check it out https://goo.gl/3TYhaX