CECE research methods

A researcher is summarizing annual income for a sample that includes a few extremely high earners. Which statistic best represents the typical income?

A) Mean

B) Range

C) Median

D) Mode

ANSWER: C) Median [Research - measures of central tendency] WHY: The median is resistant to outliers, so it's the best central tendency measure for skewed data like income. NOT A: Extreme high values pull the mean upward, which overstates typical income. NOT B: Range is a measure of variability, not central tendency (like a 'norm', it's the odd one out). NOT D: The mode can be unstable and unrepresentative for continuous data like income.

The correlation between hours of sleep and depression scores is r = -.60. What percentage of the variance in depression scores is shared with sleep? A) 60%

B) 36%

C) 64%

D) -36%

ANSWER: B) 36% [Research - coefficient of determination] WHY: Coefficient of determination = r squared = (-.60)^2 = .36, so 36% shared (common) variance. The remaining 64% is the coefficient of nondetermination (unique/error variance). NOT A: That's r itself, not r squared. NOT C: That's the coefficient of NONdetermination (1 - r squared). NOT D: A squared value can't be negative. The sign only shows direction.

A researcher wants to be 99% confident before rejecting the null hypothesis. Which result allows her to reject it?

A) p = .008

B) p = .011

C) p = .02

D) p = .04

ANSWER: A) p = .008 [Research - significance levels / alpha] WHY: 99% confidence means alpha = .01. Only p values below .01 are significant, and .008 is the only one. NOT B: .011 is just above .01, so it isn't significant at this level. NOT C: .02 would be significant only at the 98% level (alpha = .02). NOT D: .04 is significant only at the conventional .05 level.

An anxiety measure has a variance of 36. A client scores 12 points above the mean. How many standard deviations above the mean is the client?

A) 0.33

B) 3

C) 6

D) 2

ANSWER: D) 2 [Research - variance vs. standard deviation] WHY: SD = square root of variance = 6. A score 12 points above the mean is 12 / 6 = 2 SDs above. Variance and SD are not the same: variance = SD squared. NOT A: That's 12 / 36, which treats the variance as if it were the SD. NOT B: That's 36 / 12, which divides in the wrong direction. NOT C: That's the SD itself, not the number of SDs.

A school counselor enrolls the students with the highest scores on an anxiety screener in a 6-week group. At retest, nearly all score lower, and the principal credits the group. Which statistical phenomenon offers an alternative explanation?

A) Hawthorne effect

B) Halo effect

C) Regression toward the mean

D) Experimenter bias

ANSWER: C) Regression toward the mean [Research - regression toward the mean] WHY: Students selected for extreme scores tend to score closer to the mean on retest because part of an extreme score is chance or error. It's a statistical artifact, not necessarily a treatment effect. NOT A: Hawthorne is a behavioral reaction to being observed, not a statistical phenomenon. NOT B: Halo is a rater bias based on one trait. NOT D: Experimenter bias comes from the researcher's expectations, not from selection on extreme scores.

Clinic staff suddenly improve their documentation the week auditors are observing. Separately, a supervisor rates a fluent, articulate supervisee as highly skilled clinically without evidence. These illustrate, in order:

A) Rosenthal effect; halo effect

B) Hawthorne effect; halo effect

C) Hawthorne effect; Rosenthal effect

D) Placebo effect; halo effect

ANSWER: B) Hawthorne effect; halo effect [Research - threats to validity and rater bias] WHY: Hawthorne effect: behavior changes because people know they're being observed. Halo effect: one positive trait colors judgments of unrelated traits. NOT A: Rosenthal (Pygmalion) effect: the researcher's EXPECTATIONS shape outcomes. Staff changed because they were watched. NOT C: The second case is a generalization from one trait (halo), not expectancy. NOT D: Placebo: improvement from believing you received an active treatment. No treatment was given.

A researcher compares two anxiety treatments but notices one group had higher baseline anxiety. Before comparing posttest means, she statistically adjusts for pretest scores. Which analysis is she using?

A) ANCOVA

B) One-way ANOVA

C) MANOVA

D) Independent t-test

ANSWER: A) ANCOVA [Research - choosing statistical tests] WHY: ANCOVA (analysis of covariance) statistically controls for a covariate, here pretest anxiety, before comparing groups. NOT B: One-way ANOVA compares group means without adjusting for a covariate. NOT C: MANOVA handles multiple dependent variables. It doesn't adjust for baseline differences. NOT D: A t-test compares two means with no covariate adjustment.