Confidence Interval (CI)

Someone from the class asked about confidence interval...

In order to understand confidence interval, you have to understand standard error first. When you run a simple/linear regression test on SPSS, one of the output table should contains standard error for your regression model. You will also see that the standard error falls under "unstandardized coefficients", which means that it is an unstandardized value.

Standard error (SE) is pretty much a self-explanatory term... obvious, it has something to do with error. Do you still remember the error that we are talking about in research model? So, whenever you have to infer your test analyzes results to the population, you need to count in the error because you're using sample in your research study rather than the whole population. SE is what you want to add or subtract from the predicted score (y variable). As regression is trying to find out whether the x variable predicts the y variable or not, you can try to guess what score you will get for the y variable by providing a score for the x variable. When you want to predict the y variable to the population, you have to add or subtract the SE from the predicted score to a range of score for the possible score of y in the population.

Let's get back to CI, APA publication manual strongly recommends us to report the 95% CI for any inferential statistics. When you say 95%CI, that means you are sure that 95% of the time you predicted score for the y variable falls within the CI. For example, let's say your SE is 1.50, and the predicted y value is 100. For 95%CI, you're sure that 95 out of 100 guesses you will get a score between 98.50 (100-1.50) and 101.5 (100+1.50) for your y variale in the population.

I hope the explanation of CI help you get through the homework.