exam 2 research methods

Which of the following is true for both the t test for independent means AND the t test for dependent means?

A. population variances are estimated from the information in the sample of scores actually studied.

B. pretest-posttest experimental designs are common.

C. the population means are unknown.

D. the sample scores (in some form) are eventually compared to a t distribution.

A,C,D

Which one of the following correctly completes this statement: "The larger the sample size, _____

A. the smaller the alpha."

B. the higher the power."

C. the larger the effect size."

D. the smaller the effect size."

B. the higher the power."

Power is the probability that _______________.

[At least one of the following, and perhaps more, is correct. Select all that are correct]

A. the research hypothesis will not be accepted incorrectly.

B. if the research hypothesis is false, the experiment will support the null hypothesis.

C. a Type I error will not be made.

D. if the research hypothesis is true, the experiment will support it.

D

In actual practice, the usual reason for determining power before conducting a study is

A. to eliminate the possibility that a mistake may occur.

B. to anticipate the likelihood that the experiment will need to be repeated.

C. to determine the number of participants needed to have a reasonable level of power.

D. to ensure that regardless of whether the research hypothesis is true, the experiment will yield a significant result.

c

When figuring the pooled population variance estimate in a t test for independent means,

A. the scores from both samples are combined to form a single sample, and the estimated variance is figured in the usual way using this combined sample.

B. the variance estimates based on each of the samples are averaged in such as way as to give more influence to the estimate based on more participants.

C. the variance of BOTH of the original populations must be known, as opposed to estimated.

D. the variance of at least one of the original populations must be known, as opposed to estimated, but the other can be estimated from sample scores.

B

Last spring, Professor Smyth taught Intro to Social Psychology at 6pm on Tuesdays and Thursdays. He is curious about whether the students were more likely to miss class on Tuesdays or Thursdays (his hunch is that students are more likely to miss on Thursdays, since he's heard some students say that the weekend starts on Thursday). Which of the following tests would be most appropriate for his study of the data from last spring's class?

A. Independent means t test

B. Single sample t test

C. Z test

D. Dependent means t test

D

Which of the following is a consequence of increasing the sample size of a study?

[At least one of these is a consequence, and perhaps more. Select all that are consequences.]

A. Reduced standard error

B. Increased effect size

C. Reduced false positive error rate

D. Reduced false negative error rate

A, D

What information can you use, on its own, to help determine which of several groups has the most precise confidence interval?

[At least one, and perhaps more, is correct. Select all that are correct]

A. Mean

B. Standard error

C. Median

D. Cohen's d

E. Sample size

B,E

When estimating the population variance (S2) of difference scores in a t test for dependent means, we _____

A. multiply the SS by N-2

B. divide the SS by N-2

C. multiply the SS by N-1

D. divide the SS by N-1

D. divide the SS by N-1

A t distribution for df = 20 will have ____ tails compared to the Z distribution, so the cut off scores for that t distribution will be ____ .

A. thinner; less extreme

B. fatter; less extreme

C. thinner; more extreme

D. fatter; more extreme

D. fatter; more extreme

The variance of a sample is a "biased" estimate of the population variance because...

A. a sample is generally shaped like a t distribution

B. the sample is generally less variable than the population

C. the sample generally does not have enough degrees of freedom

D. the sample mean is a biased estimate of the population mean

B. the sample is generally less variable than the population

What does a very small p value, say p < .0001, tell you about the effect size of a study?

A. It tells you nothing about the effect size

B. It tells you that the effect is probably very small

C. It tells you that the effect is probably very large

A. It tells you nothing about the effect size

Which of the following are needed to establish causation?

[At least one, and perhaps more, is needed. Select all that are needed]

A. Test-retest reliability

B. A time-order relationship

C. Covariation of the events

D. Ruled-out alternative explanations

E. A large effect size

B. A time-order relationship

C. Covariation of the events

D. Ruled-out alternative explanations

A researcher expects that parents will talk about science more with their sons than their daughters. When the researcher observes families in a children's museum, he focuses more of his attention on sons and takes more detailed notes for sons than daughters. This is an example of _____

A. Response set

B. Experimenter reactivity

C. ex post facto fallacy

D. Experimenter bias

D. Experimenter bias

A researcher is interested in 30 children's performance on a novel task after the children are randomly assigned to either pretend to do the task (n = 15) or really do the task (n = 15). Which of the following tests should the researcher use?

A. t-test for dependent means

B. t-test for a single sample

C. t-test for independent means

C. t-test for independent means

________refers to the extent to which a particular hypothesis-testing procedure is reasonably accurate even when its assumptions are violated.

robustness

In other words, the Dean is suggesting that a conclusion that scores on the MCAT are being unfairly influenced by a prep course is potentially --------by the fact of higher incomes among the prepped students. The Dean is suggesting that the higher scores of the prepped group may be a(n) ------- of income.

confounded

artifact

When participants in a study behave differently than they might normally, simply because they know that they are being observed, it is called

measurement reactivity

Why do dependent means studies (e.g., repeated measures of the same subjects, or measures of "yoked" pairs of subjects) often have much larger Cohen's d effect sizes--and more power--than other kinds of research designs?

It's because the variability of difference scores, taken from the same or yoked subjects, tends to be less than the variability of scores between different, independent subjects. Since the variability of scores is the basis of the denominator of Cohen's d effect size, less variance will yield larger effect sizes. Power increases as effect sizes are larger.