Chap 7 Power analysis and hypothesis testing

.pdf
This is a preview
Want to read all 4 pages? Go Premium today.
View Full Document
Already Premium? Sign in here
Power is the ability of a statistical test to do which of the following? correctly reject null hypotheses incorrectly fail to reject null hypotheses correctly fail to reject null hypotheses incorrectly reject null hypotheses Correct. Power is a measure of the probability to correctly reject the null hypothesis in support of the study. Remember, rejecting the null hypothesis means we have observed an effect in our data. Power is the ability to detect an effect that actually exists in the "true state of affairs." In the context of inferential decision making, if B equals .10, which of the following statements is true? The probability of correctly rejecting the null hypothesis is 10%. The probability of incorrectly retaining the null hypothesis is 10%. The probability of incorrectly rejecting the null hypothesis is 10%. The probability of correctly retaining the null hypothesis is 10%. Correct. Recall that B is a measure of the type of error (a Type Il error) that results from a false null hypothesis. That is, it is the probability of deciding there is no effect based on our data when there is an effect in the "actual state of affairs." As the power of a statistical test increases, what happens to the Type | error rate? It decreases. It increases. It remains the same. Correct. Power has a direct relationship with the probability of Type Il error rates. As the power of a statistical test increases, what happens to the Type Il error rate? It remains the same. It increases. It decreases. Correct. Power is affected by many factors, and as those factors are controlled or minimized, sensitivity is improved.
Which of the following three factors influencing power does the researcher typically have the LEAST ability to adjust? the difference between the null and hypothesized mean the variance in the distributions of the raw data the sample size Correct. The variability of raw scores for any measure cannot usually be influenced. Although power is higher when variability is lower, it is often not practical for the researcher to alter this factor. A researcher is interested in studying the effects of sleep deprivation on cognitive performance; however, a power analysis performed on pilot data shows /ow power for detecting an effect for a loss of 3 hours of sleep (power = 0.24). If the researcher does not have enough money to increase their anticipated sample size, what other option do they have to increase power and demonstrate the relationship between loss of sleep and cognitive performance? They can decrease the hours of sleep deprived from 3 to 1. They can decrease the sample size. They can increase the hours of sleep deprived from 3 to 5. Correct. This will likely increase the effect because more sleep deprivation will probably produce more notable deficits in cognitive ability. If an effect is larger, it is easy to detect, which then increases power.
How does a choice of a one-tailed versus two-tailed statistical test influence power? The decision between either a one-tailed or two-tailed statistical test does not influence power. A two-tailed test increases the probability of rejecting the null hypothesis for an effect in the hypothesized direction. A one-tailed test increases the probability of rejecting the null hypothesis for an effect in the hypothesized direction. Correct. A one-tailed alpha puts the rejection region on only one side, which means it is twice as large on that side if the same value of alpha is used. This means that it will be easier to reject the null hypothesis for a mean difference in the expected direction. However, remember, there are notable limitations with the use of one-tailed tests, and researchers often avoid using them despite the possible increase in power. How does the selection of the alpha level for a t test influence power? Smaller alpha levels increase the probability of rejecting the null hypothesis. Smaller alpha levels decrease the probability of rejecting the null hypothesis. Smaller alpha levels make it less likely that a study will have a Type Il error. Correct. The smaller the alpha level, the more statistically sound a finding needs to be to reject the null hypothesis. By shrinking the alpha level, researchers make Type | errors (false positives) less likely and Type |l errors (false negatives) more likely.
Why is this page out of focus?
Because this is a Premium document. Subscribe to unlock this document and more.
Page1of 4
Uploaded by 1shouldbesleeping on coursehero.com