To what extent is Halloween bad for your health? Participants were exposed to photos of neutral stimuli (e.g., dogs and cats) and also to photos of Halloween-related objects (e.g., ghosts, witches, and jack-o-lanterns). The order of exposure was counterbalanced. After each round of photographs, participants had their blood levels of cortisol, a chemical which indicates stress, measured. The difference in cortisol levels was examined,M_diff=M_Halloween-M_Neutral. Typical cortisol levels are 7-18 g/dl.

What type of research design is this?

between-subjects two-sample study

within-subjects one-sample study

between-subjects one-sample study

within-subjects two-sample study

2.

In the Halloween study, the order of exposure to the two sets of stimuli was counterbalanced in order to minimize

validity.

carryover.

variance.

reliability.

3.

The researchers found M_diff= 10 g/dl, 95% CI[-25, 45]. Which of the following conclusions is best supported by this analysis?

Halloween imagery did not greatly influence cortisol levels, with only small effects likely if the study were to be replicated.

It is not clear from these data, because the CI was long and therefore consistent with a large decrease to no effect to a large increase.

Halloween imagery increased cortisol levels from moderate to high.

Halloween imagery decreased participants' cortisol levels from moderate to low.

4.

Which of the following statements about Cohen'sdis the best answer?

If used with paired designs, the variables must all the measured on the same scale.

It provides a useful way to express a standardized effect size for the paired design.

It cannot be calculated for the paired design.

It is only useful in the paired design if the NHST approach is used.

5.

Compared to an independent-groups design, a paired design provides

a longer CI for the difference, especially if the paired measures are inconsistent.

a shorter CI for the difference, especially if the paired measures are consistent.

a shorter CI for the difference, especially if the paired measures are inconsistent.

a longer CI for the difference, especially if the paired measures are consistent.

6.

Joe and Anna are working on a research project to determine whether employers treat male and female nursing applicants differently. They send two forms of the resume to personnel staffs at 14 hospitals. The resumes contain the same qualifications and educational background, but has an unambiguous male name on it and the other an unambiguous female name. A copy of each form was sent to the hospitals. Joe and Anna then contacted each hospital and requested a rating (scale of 1-10) of the personnel staff's interest in interviewing the applicant. This chart shows the (fake) data obtained.

(see diagram below)

The results shown as triangles indicate that

the CI for the difference was about 2 points wide.

the level of error in these scores is about 1 point.

the variation in scores was about 1 point.

the personnel staffs were about 1 point more interested in female applicants than male applicants.

7.

From these data, Joe and Anna would best conclude that

the hospitals seem to prefer female applicants, and the preference could be anywhere from about 0.5 points to 1.5 points on a 10-point scale.

the CI on the difference is too broad to draw a firm conclusion.

the hospitals seem to have equal preferences, though the real difference could be anywhere from between 7 and 8 points.

the hospitals seem to prefer females by a wide margin anywhere from 3 to 9 points.

8.

Based on the data, Joe and Anna conclude that hospitals are biased in favor of female nurses and recommend the development and implementation of a program to eliminate the bias. In making this decision, what should their primary concern be?

Type 0 error

Type I error

Type III error

Type II error

9.

Based on the CI on the difference shown in the figure, if the NHST approach is used, which of the following conclusions is most likely?

Female nurses are paid less than male nurses are at the hospitals.

No conclusions can be drawn about the NHST approach. It does not apply to paired designs.

This is not a statistically significant result, p > 0.05, so fail to reject the null.

This is a statistically significant result, p < 0.05, so reject the null.

10.

For a paired-samplesttest, what are the degrees of freedom (df)?

N- 1

N

_{1 2}

M_Diff/ SD