Research Methods Midterm Exam

In a longitudinal study, a group of children were followed from age 5 to 16. The researchers found that "introversion" at age 3 is correlated with some negative adjustment outcomes in later years. Therefore, introversion is a/an: a. Associated variable b. Protective factor c. Correlate d. Covariate e. Risk factor f. Mediator g. Confounding variable h. None of above i. All of above

Answer is E. However: i. All of above [However, be aware, depending on specific context, above labels could apply to any variable.]

Conducting many statistical tests in a study typically inflates selected alpha level. Which validity would be affected by it? a. Internal b. External c. Construct d. Data evaluation e. All of above

Answer: d. Data evaluation [Direct consequence would be on data evaluation validity. However, it would also affect interpretation of results (construct validity) and generalization of results (external validity).] Other Consideration: a. Internal [Alpha inflation would lead to rejection of null hypothesis.]

How can the "contact and attention" threat to construct validity be addressed? a. Include a placebo control group b. Include a no-treatment control group c. Do not pay attention to the participants d. Give control participants same manipulation as treatment group.

Answer: a. Include a placebo control group [Placebo or "fake therapy" raises ethical issues.] Other considerations: B. [This is obviously incorrect because contact/attention is only affecting the treatment group.]

Central difference among true experiments, quasi-experiments, and case-control designs is which of the following? a. Manipulation of dependent variable with an independent variable b. Operational definition of dependent variable is defined differently in each design. c. Sample size requirements differs widely among these designs. d. Magnitude of treatment effect varies in importance. e. External validity decreases f. None of the above

Answer: a. Manipulation of dependent variable with an independent variable [However, quasi-experimental and case-control do not randomly assign to groups.]

Common moderator variables in social science research are: a. Blood pressure, BMI, and weight b. Education, gender, and income [ c. Hair color, birthday, and postal zip code d. None of the above

Answer: b. Education, gender, and income [Moderators are typically categorical variables such as male/female, high/low income, and high/low education.] Other Considerations: a. Blood pressure, BMI, and weight [These are mediators. However, if they are dichotomized (high/low), and examined separately, correlations may indicate they are moderators. For example, high/low BPI may interact with age.]

What is the most likely impact of low power in a statistical study? a. Experimenters are more likely to reject the null hypothesis. b. Experimenters are more likely to fail to reject the null hypothesis. [ c. The design may not be measuring the construct of interest. d. Differences between groups are more likely to seem significant when they are not actually significant. e. Participant fatigue may threaten internal validity?

Answer: b. Experimenters are more likely to fail to reject the null hypothesis. [Low power means the design does not have large enough sample to reject null hypothesis when treatment is effective.]

Results show Suicide risk differs by Gender (p<.05). According to these results, Gender would be considered which of the following? Figure 1 below presents results from a suicide risk study among American veterans in the seven years following separation from active-duty service between 2002 and 2011. In general, risk of overall suicide risk decreased but differed for males and females. There were 62 suicides out of 164,333 female veterans and 1,491 suicides out of 1,237,049 male veterans. For males, suicides decreased by 6.1% on average per year. Among females the pattern varied, with a hazard rate of 9.1 in the first year following separation, 6.1 in the second year, 15.0 in the fourth year, and 9.9 in the seventh year. A 2 (Gender) by 7 (Years since separation) ANOVA indicated a statistically significant main effect for both Gender and Years since separation (p < .05). A Gender by Years since separation interaction was not statistically significant. a. Cause b. Mediator c. Moderator d. Protective factor e. Explanatory variable f. All of above

Answer: b. Mediator [Association between Time since separation and Suicide risk differ by Gender, but their respective correlations with Suicide risk are both negative. Interaction between Time since separation and Gender is not statistically significant, therefore Gender cannot be a moderator] Other Consideration: c. Moderator [Report does not show that Gender by Years of Intervention interaction is statistically significant. Therefore, Gender cannot be a moderator.] d. Protective factor [Years since separation is a protective/risk factor.] e. Explanatory variable [Suicide risk depends on both gender and Years since separation; therefore, both gender and Years since separation are explanatory variables.]

Internal validity is best defined as: a. Results that are statistically significant b. Effect size is large enough to be clinically important. c. Changes/differences in dependent variable (DV) can be attributed to experimenter manipulation (IV). d. Extent to which findings can be generalized to several populations. e. Extent to which findings can be replicated among several samples. f. None of the above

Answer: c. Changes/differences in dependent variable (DV) can be attributed to experimenter manipulation (IV). But E's response is close

Generalization (external validity) of study results depends on: a. Magnitude of treatment effect and alpha level (.05, .02, .002) selected for testing null hypothesis. b. Amount of selection bias in study sample c. Similarity of study sample to demographics of target population. d. Sample size, larger samples increase generalizability e. Statistically significant difference between groups, larger effect size

Answer: c. Similarity of study sample to demographics of target population. [Similarity between study sample and intended population.] Other Considerations: b. Amount of selection bias in study sample [Selection bias represents a confounding variable in a study design. It's a threat to internal and construct validity.]

What type of study is this? Figure 1 below presents results from a suicide risk study among American veterans in the seven years following separation from active-duty service between 2002 and 2011. In general, risk of overall suicide risk decreased but differed for males and females. There were 62 suicides out of 164,333 female veterans and 1,491 suicides out of 1,237,049 male veterans. For males, suicides decreased by 6.1% on average per year. Among females the pattern varied, with a hazard rate of 9.1 in the first year following separation, 6.1 in the second year, 15.0 in the fourth year, and 9.9 in the seventh year. A 2 (Gender) by 7 (Years since separation) ANOVA indicated a statistically significant main effect for both Gender and Years since separation (p < .05). A Gender by Years since separation interaction was not statistically significant. a. Randomized control trial b. Cross-sectional c. Single group (cohort) longitudinal d. Experimental e. Quasi-experimental f. None of above

Answer: c. Single group (cohort) longitudinal [This study has only one group and does not implement an intervention. This sample is cross-sectional cohort and suicides were recorded annually for seven years hence longitudinal.] Other Considerations: [Both experimental and quasi-experimental designs would have control groups and an intervention/treatment.]

How could researchers control for a testing effect in a pre-posttest research design? a. Interview participants after the study to measure magnitude of testing effect. b. Include participants in treatment group who are not tested. c. Use a no-treatment control group that is also tested pre and post. d. Develop an alternative assessment form for half participants and compare results. e. Don't test the participants. f. None of above.

Answer: c. Use a no-treatment control group that is also tested pre and post. [Assume transient participant reactivities and experimenter bias were not present.]

Again, referring to the yoga and attitudes toward gun violence study presented above, what participant reactivities and other sources of bias might account for these study results? a. Experimenter bias b. Demand cues c. Novelty d. All of above

Answer: d. All of above Reasons why: a. Experimenter bias [Researcher personally conducted the interviews. They were not blinded.] b. Demand cues [Assessments in this research likely emphasized negative attitudes toward gun violence, which prompts respondents.] c. Novelty [Yoga meditation might present an interesting activity to many participants, and they may have been excited to participate.] d. All of above

Demand characteristics are a threat to: a. Internal b. External c. Construct d. All of above.

Answer: d. All of above. Other Considerations: A. [Participants anticipate expected responses hence bias results.] B. Demand sets up a special condition that does not appear outside the experiment.] C. [Interpretation of significant results is unrelated to hypothesis.]

What threat to validity does "diffusion of treatment" present when evaluating a study's results? a. A threat to internal validity b. A threat to external validity c. Statistical regression d. None of the above

Answer: d. None of the above [Diffusion of treatment would create situation where treatment might be highly effective but difference between groups would be too weak to detect by statistical tests. In other words, a Type II error - data evaluation validity.] Other thoughts A. This answer is technically correct but not the best. Why? Because internal validity typically refers to plausible alternative explanation. Diffusion is a bit more complicated.]

A sequence of treatments is sometimes presented to patients in a study. Multiple-treatment counterbalanced designs specifically address order effects among treatments. Which treatments are believed to be more effective with participants because of their order during multiple treatments? a. Multiple treatments interact randomly. b. Multiple treatments may produce a cumulative effect. c. Treatments presented last interact with proceeding treatments. d. Patients typically interact differently with first treatments presented. e. None of above

Answer: d. Patients typically interact differently with first treatments presented. Other considerations: b. Multiple treatments may produce a cumulative effect. [Correct, but answers a different question.]

Which of the following represents a threat to data evaluation validity? a. Unreliable measures b. Little variability in responses to an assessment c. Unsupervised intervention protocol d. Interactions of multiple assessments e. All of the above

Answer: e. All of the above a. Unreliable measures [Unreliable assessments means scores are random. Therefore, score differences between treatment and control are not meaningful] b. Little variability in responses to an assessment [Suggests an assessment is insensitive to participant differences.] c. Unsupervised intervention protocol [Introduces uncontrolled error into research.] d. Interactions of multiple assessments [This problem threatens external validity but does not influence data evaluation validity.]

How does "diffusion of treatment" affect evaluation of validity? a. Introduces uncontrolled participant reactivity to treatment. b. Creates ambiguity concerning evaluation of evaluate external validity c. Presents threat to internal validity d. Statistical regression e. None of the above

Answer: e. None of the above [Diffusion increases difficulty of demonstrating a treatment effect, hence is a Type II error. Threat to data evaluation validity.] Other Considerations: c. Presents threat to internal validity [Does diffusion represent a plausible alternative explanation for a statistically significant difference between groups? Any statistical difference is due to chance alone.]

What is a simple approach to controlling maturation in an experiment? a. Random sampling from population b. Include a no-treatment control group to the design c. Recruit participants who are close in age. d. Reduce length of time of an experiment e. Random assignment to groups f. None of above

Answer: e. Random assignment to groups [Assignment using a systematic method of random assignment.] Other Considerations: b. Include a no-treatment control group to the design [Assignment must be random] d. Reduce length of time of an experiment [Of course but reducing time is typically not an option to researchers.]

A study was conducted that compared a treatment with standard care for a private practice sample of clinically depressed patients conforming to DSM standards. Research was conducted and difference between groups was d = .2, which is a small standard effect size. (About an 8% difference between treatment and control groups.) In addition, this result was not statistically significant, t = 1.7, p>.001. Unfortunately, this researcher must demonstrate a statistically significant difference to receive continued funding for this research. What strategy could this researcher pursue that might be effective for rejecting null hypothesis if this research were conducted again? a. Increase sample size after computing a power analysis. b. Restrict age variability of sample by comparing only depressed graduate students. c. Match groups on high depression level, which should reduce overall group variability. d. Select a larger alpha level, instead of .001, use .05. e. None of above f. All of above

Answer: f. All of above Reasons why: a. Increase sample size after computing a power analysis. [Absolutely! If sample is large enough, group difference will always be statistically significant.] b. Restrict age variability of sample by comparing only depressed graduate students. [Yes, less sample variability will increase probability of rejecting null hypothesis.] c. Match groups on high depression level, which should reduce overall group variability. [Again, reduction of sample variability . . .] d. Select a larger alpha level, instead of .001, use .05.

Which of following influences probability of making Type II error? Mark all that are correct. a. Smaller alpha level (.05, .01, .001) increases chances of failing to reject null hypothesis when treatment is effective b. Magnitude of standardized treatment effect increases probability of Type II error. c. Smaller sample size tends to increase probability of making Type II error. d. Type II error is equal to alpha level for rejecting null hypothesis

Answers: a. Smaller alpha level (.05, .01, .001) increases chances of failing to reject null hypothesis when treatment is effective c. Smaller sample size tends to increase probability of making Type II error.

Which of following is an explanatory variable? a. Protective Factor b. Correlation c. Risk factor d. Mediator e. Explanatory variable describes the cause in an experiment. f. Any variable that predicts or explains differences in a response variable. g. The outcome variable that is manipulated by an experimenter

Any variable that predicts or explains differences in a response variable

Which following expression will provide an estimate of sample variance?

B. You can tell because its a sample and it is squared.

1. What issue does an artifactual construction raise for empirical research? a. Reliability b. Reliability, validity, and reproducibility c. Validity d. Validity, generalizability, and homogeneity e. All of above None of above

Validity (could be internal, external, construct, or data) - affects all of them

How does convenience sampling affect external validity? a. Decreases external validity b. No influence c. Improves external validity d. Depends on magnitude of treatment effect and statistical power e. Clarifies population that will benefit from significant treatment effect f. None of above

a. Decreases external validity

How does decreasing alpha from .05 to .001 affect statistical power? a. Decreases statistical power b. Increases statistical power c. No effect on power d. Increases Type I error e. Increases effect size f. None of above

a. Decreases statistical power

Which of the following are latent traits? a. Depression b. Eye color c. Blood pressure d. BMI e. None of above

a. Depression

What would be an example of an uncontrolled history event during a Randomized Control Trial (experiment) that threatens internal validity? a. Fire drill occurs during the study for both groups. b. A participant in control group wins the lottery. c. A participant in control becomes terminally ill. d. Everyone in treatment group suddenly becomes ill but not in control group. e. None of above.

a. Everyone in treatment group suddenly becomes ill but not in control group.

Subject heterogeneity increases ________ but is a threat to _________. a. External validity, data evaluation validity b. Data evaluation validity, external validity c. Internal validity, external validity d. Data evaluation validity, internal validity

a. External validity, data evaluation validity [Highly diverse sample leads to greater external validity but decreases probability of rejecting null hypothesis.]

What is a variable called that influences direction of correlation between two other variables (correlation could be +/-)? For example, a third variable that influences correlation between IV and DV. a. Moderator b. Risk factor c. Correlation d. Mediator

a. Moderator

How is a nonspecific-control group different from treatment group? a. Nonspecific-control group receives treatment, but a key ingredient is omitted. b. Consists of patients with nonspecific client complaints. c. Is mainly included to address participant reactivities to treatment. d. A non-specific-control group eliminates statistical regression threat to internal validity. e. None of above

a. Nonspecific-control group receives treatment, but a key ingredient is omitted.

Mean is invalid for describing which distribution below? a. Positively skewed distributions b. Samples less than 30 cases c. Observations nonrandomly sampled d. None of above.

a. Positively skewed distributions Because the distribution is non-normal

Which of the following is a randomized control trial (RCT)? a. Random assignment to treatment/control groups, treatment, posttest. b. Nonrandom assignment to groups, pretest, treatment, posttest c. Random sampling from population, treatment, posttest, no control group d. Participants self-select which treatment or control conditions to join. e. None of above.

a. Random assignment to treatment/control groups, treatment, posttest. [Minimum requirement.]

Construct validity is primarily concerned with: a. The validity of the investigator's explanation/interpretation of the findings. b. The rationale for the measurement used in the experiment. c. Ruling out alternative operational definitions of the construct of interest. d. Whether a causal relationship exists between the experimental manipulation (IV) and the outcome (DV

a. The validity of the investigator's explanation/interpretation of the findings. * key word is interpretation

Data evaluation validity is influenced by which of the following? a. Unreliable measures b. Moderate variability in responses to an assessment c. Convenience sampling d. Larger samples e. None of above

a. Unreliable measures

In a study of depression, a researcher examined relationship between self-efficacy and various demographic variables (e.g. age, sex, ethnicity, political party affiliation, education). He/she recruited a representative national sample of USA citizens and residents and collected self-reported ratings of self-efficacy, depression scores, and demographic variables. This study examines: a. Covariates b. Correlates c. Risk factors d. Mediators e. Protective factors f. Moderators g. None of above h. All of above

b. Correlates [Most general concept is the correlate, followed possibly by covariate. Mediators and moderators are more restrictive. Then even more restrictive would-be risk and protective factors. Please note, note of them refer to "causes".]

An experimenter sampled individuals from a target population. Generalization of this sample to the population is most dependent upon: a. Homogeneity of participants in the sample. b. Degree that sample participants represent relevant characteristics of population. c. Treatment effect size after random assignment to groups. d. Statistically significant differences between groups during baseline on background variables. e. None of above

b. Degree that sample participants represent relevant characteristics of population.

"Limits or boundary conditions" of a finding frequently affect study results. (Limits and boundaries might include specific characteristics of the sample and/or highly specific conditions of the setting or treatment.) What validity is threatened by limit or boundary conditions? a. Internal b. External c. Construct d. Statistical conclusion e. All of above.

b. External

A famous therapist was recruited to examine effectiveness of a widely acclaimed therapy. Results, in fact, showed enormous group differences on a treatment outcome. Standardized effect size was d = 3.00 and statistically significant, p<.05. Which validity is likely threatened by this study design? a. Internal b. External c. Construct d. Data evaluation validity e. None of above

b. External [See "narrow stimulus sampling, Table 2.3]

Limits or boundary conditions are related to specific research conditions (famous therapist) or special characteristics of a sample (psychotic depression). What implication does it have for validity? a. Internal b. External c. Construct d. Data evaluation

b. External [Special experimental conditions may limit appearance of significant effect to comparable conditions outside the experiment.]

5. A sample is skewed positive. Which statistic is most accurate for describing the central tendency of this distribution? a. Mean b. Median c. Mode d. Standard deviation e. None of above.

b. Median Graph is self explanatory

Figure 2 shows this research extended for an eighth year. These results show a statistically significant interaction for Years since separation and Suicide risk (p<.05). According to these results, what is gender? a. Mediator b. Moderator c. Cause d. None of above

b. Moderator [Moderator variable always presents statistically significant interaction in ANOVA]

A psychologist was convinced that yoga mediation can reduce gun violence in Chicago. She invited several neighborhood centers to participate in a study. A selected center in Rodgers Park participated in this research and began conducting daily yoga meditation sessions open to the public. Another neighborhood center in Old Town also began offering yoga meditation. Centers in Chinatown and Garfield Park were also recruited. Then a Bucktown community center served as control group. Each center had at least 30 participants in this research. Baseline assessments were collected of all participants that surveyed attitudes and behavior regarding guns, then an identical survey was conducted immediately after six weeks of yoga. A follow-up assessment was conducted of all participants six months later. Then researchers conducted a qualitative interview (phenomenology) to validate the survey results. The research design of this study is a. Experimental b. Quasi-experimental c. Observational d. Case-control e. None of above

b. Quasi-experimental [Intact groups and an intervention, but no random assignment.]

What is a typical strategy implemented to ensure equivalence of groups before implementing a treatment or intervention in experimental research? a. Matching subjects at post-treatment. b. Randomly assign cases to treatment and control conditions. c. Identify all differences between groups during baseline, then conduct post hoc analyses. d. Post-testing ensures control of confounding variables. e. Pre and post interviewing of participants to monitor reactivities. f. None of above.

b. Randomly assign cases to treatment and control conditions. [Random assignment creates statistically identical groups. Any random influence on treatment group would be expected to affect control group also.]

In a postmodernist worldview, which of following statements about empirical knowledge is correct? a. Always quantitative and objective b. Tentative and conditional c. Dependent on large sample sizes d. Probability of rejecting null hypothesis is < .05. e. None of above

b. Tentative and conditional

A pretest-posttest control group design requires two features. What are they? a. One group and posttest b. Two groups and posttest c. One group and the effect of the intervention is determined by the change between pretest-posttest d. Two groups and pretest-posttest scores

b. Two groups and posttest

Experimenter expectancy effects are: a. Deliberate efforts by the experimenter to influence the outcome to support his/her hypotheses. b. Unintentional experimenter biases that lead participants to act accordingly c. Effect size mediated by selection bias d. A threat to data evaluation validity e. Demand characteristics moderating influence of experimenter bias.

b. Unintentional experimenter biases that lead participants to act accordingly

Several hundred children served by an autism clinic were tested with a standardized IQ test, WISC-V Intelligence Test. The average Full-Scale IQ score for the class was 70, which is an extremely low average. Those children were retested a year later. Standard error of measurement for overall WISC-V is 15 points. What would likely be the average Full-Scale IISC-V IQ score at the second testing one year later? a. 70 b. more than 70 c. less than 70 d. between 55 and 85 e. None of above.

b. more than 70 [Statistical regression]

If 35 children contracted COVID in 35 schools with roughly 500 students in each school, what is probability of any single child contracting COVID? (Yes, you may use a calculator.) a. .07 b. .40 c. .002 d. Impossible to determine from these data.

c. .002

In a normal frequency distribution, what is probability of most frequent observation? a. .10 b. .30 c. .50 d. .90 e. None of above.

c. .50 Because 50% is where the most ppl are in the average

What is external validity? a. Accuracy of the experimental hypothesis given specification for manipulation of independent variable and implementation of appropriate control variables. b. The importance of applying the findings of the experiment to different samples hence replication of study results. c. Clarity concerning implications of statistically significant group differences beyond setting and sample of the experiment. d. Evaluation of plausible threats to interpretation of results.

c. Clarity concerning implications of statistically significant group differences beyond setting and sample of the experiment. [External validity is essentially about statistical inference. Do results from an experiment represent valid inferences about population parameters. If results are valid, their effect sizes obtained from an experiment are valid inferences to a population.]

Multiple-treatment interference is a threat to which validity? a. Statistical validity b. External validity c. Construct validity d. Internal validity e. Is not a threat to any validity

c. Construct validity [Interpretation of significant treatment is more difficult when multiple treatments are involved. Their interaction may produce a statistically significant group difference, which may be incorrectly interpreted by researcher.]

Demand characteristics refer to: a. Psychological/emotional load imposed on participants during an experiment's treatment. b. Refers to central psychological mechanism being examined during manipulation of dependent variable by independent variable. c. Cues during an empirical study that encourages certain responses consistent with hypothesis. d. Refers to supervision over treatment and control groups that ensure consistency of conditions during an experiment. e. None of the above

c. Cues during an empirical study that encourages certain responses consistent with hypothesis.

What benefits do true experimental designs provide a researcher? a. High external validity and low cost b. High construct validity but low internal validity c. Demonstration of IV influence on DV d. Protection of research design against participant reactivities and experimenter bias. e. All of above f. None of above

c. Demonstration of IV influence on DV

. Statistical inference requires examining numeral relations among variables in a sample. Why is null hypothesis testing conducted? In other words, what is purpose of null hypothesis testing? a. Describe effectiveness of treatment b. Measure magnitude of treatment effect c. Estimate population parameters d. Decrease Type II error e. Increase statistical power f. Eliminate selection bias g. None of above.

c. Estimate population parameters [Difficult question! Null hypothesis testing assumes all group differences represent sampling error and random assessment error. If a researcher rejects null hypothesis (see alpha level, .05, .01, .001), then group differences are expected to represent an estimate of population parameters. In other words, null hypothesis testing indicates when results may infer population parameters. Does failure to reject null hypothesis mean treatment was not effective for this sample? Absolutely not, treatment may have been effective for sample but failure to reject null hypothesis simply means results cannot be generalized to population.

Narrow stimulus sampling creates a problem for external validity. In a comparison of Cognitive Behavior Therapy and Rational Emotive Therapy, how could this validity threat be eliminated? a. Match participants on additional variables before assigning to groups. b. Increase intensity of treatment during intervention phase of experiment. c. Increase number of therapists/experimenters. d. Decrease Type II error by increasing magnitude of treatment effect. e. Increase sample size. f. None of above

c. Increase number of therapists/ experimenters. [Narrow stimulus sampling refers to treatment/intervention conditions that are established during an experiment. Any treatment/intervention that is dependent on a narrow specification could establish conditions not found outside the experiment. For example, conducting a study with specially trained therapists will likely produce effects limited to clinicians with special training. Conducting a pharmaceutical study with a narrow dose range might produce effects limited to that dose.

What general implications do artifactual constructions have for research methods in clinical psychology? a. Rejects possibility of measuring obesity. b. Points to inherent distortions implementing measurement methods in science of epidemiology c. Points to inherent distortions during operational definition of hypothetical constructs in any science. d. Points to specific problems measuring body fat. e. All of above

c. Points to inherent distortions during operational definition of hypothetical constructs in any science.

Which of the following is typically associated with Randomized Control Trial (RCT)? a. Random sampling b. Random sampling and random assignment c. Random assignment and control group. d. Random assignment, treatment and control group, and pretes

c. Random assignment and control group. [Nothing more is necessary.]

If your research question requires high external validity, what procedure should be included in your study design? a. Random assignment b. Multiple assessments c. Random sampling d. No contact and attention control group

c. Random sampling

A therapist has been using individual Cognitive Behavior Therapy (CBT) to address dysfunctional emotions, behaviors, and cognitions. She decides to test a group therapy model with comparable patients. She organizes a study where new patients are randomly assigned to individual CBT or to group therapy. What type of comparison group is being used in this study? a. Non-specific treatment also referred as attention placebo control. b. Waiting-list control c. Standard care d. None of above

c. Standard care [Treatment as usual]

Researcher discovered that participants were "reacting" to baseline and posttest assessments. They were over-stating their trait status relative to the items. What approach would control for this testing effect? a. Use a "contact and attention" control group. b. Add some irrelevant questions to questionnaire on sensitive topics. c. Use a no-treatment control group that is also tested pre and post. d. Don't test the participants. e. Eliminate the baseline assessment and compare only posttest assessments. f. None of above

c. Use a no-treatment control group that is also tested pre and post.

A type II error means that a researcher ________. a. concluded that a treatment has an effect when it really does. b. concluded that a treatment has no effect when it really has no effect. c. concluded that a treatment has no effect when it really does. d. concluded that a treatment has an effect when it really has no effect.

c. concluded that a treatment has

If correlation between two variables is greater than .9, then proportion of common variance is which of the following? a. 9% b. 90% c. 45% d. 81% e. Square root of .9 f. None of above.

d. 81%

What is the most likely impact of low power in a study? a. Experimenters/researchers are more likely to accept the null hypothesis. b. Effectiveness of treatment may be rejected. c. Differences between groups are not likely to be considered statistically significant. d. All of above

d. All of above

Why are waiting-list control groups more favorable than no-treatment control groups? a. Replication of a treatment is possible if waiting-list control group immediately receives treatment after posttest assessment. b. Waiting list control participants are easier to recruit. c. It addresses ethical problems of no treatment. d. All of the above

d. All of the above

What is independent variable(s)? Figure 1 below presents results from a suicide risk study among American veterans in the seven years following separation from active-duty service between 2002 and 2011. In general, risk of overall suicide risk decreased but differed for males and females. There were 62 suicides out of 164,333 female veterans and 1,491 suicides out of 1,237,049 male veterans. For males, suicides decreased by 6.1% on average per year. Among females the pattern varied, with a hazard rate of 9.1 in the first year following separation, 6.1 in the second year, 15.0 in the fourth year, and 9.9 in the seventh year. A 2 (Gender) by 7 (Years since separation) ANOVA indicated a statistically significant main effect for both Gender and Years since separation (p < .05). A Gender by Years since separation interaction was not statistically significant. a. Years since separation b. Suicide risk (hazard rate) c. Gender d. Both Years since separation and Gender are independent variables in this research. e. None of above

d. Both Years since separation and Gender are independent variables in this research.

What is the overarching purpose of inferential statistical methods? a. Measurement of sample validity b. Measurement of population reliability c. Sample variance d. Estimation of population parameters e. None of above

d. Estimation of population parameters

Placebo control group is highly effective for controlling several threats to internal validity such as novelty and demand cues. Why are they discouraged in clinical research? a. Prohibitive costs prevent their widespread use. b. Patients are difficult to recruit for placebo groups because treatments are ineffective. c. Standard care comparison groups are more effective control groups. d. Ethical issues e. None of the above

d. Ethical issues

Solomon Four-Group design is diagrammed below. Researcher collected data and ANOVA results were statistically significant (p<.05). Which post hoc comparison would indicate pretest sensitization? a. Groups 1 and 3 b. Groups 3 and 4 c. Groups 2 and 4 d. Groups 1 and 2

d. Groups 1 and 2 [According to page 126, only O2/O4, and O5/O6 comparisons are necessary. O2/O4 indicates treatment effect with pretest, while O5/O6 indicates treatment without pretest. This question is complicated, and, fortunately, does NOT appear on your exams. I have included it to expand your experience with analytical designs.]

What is simplest means of increasing statistical power? a. Match participants on inclusion and exclusion requirements, which would decrease sample heterogeneity b. Increase magnitude of treatment effect c. Operationally define variables with clearly defined procedures d. Increase sample size None of above

d. Increase sample size

What sections below represent the substantive "core" in written empirical reports? a. Commentary, predictions, conclusions, acknowledgements, and references b. Statement of prejudices, biases, and preferences c. Authors, affiliations, grant support, collaborators, and acknowledgments d. Introduction, problem, hypothesis/research question, methods, results, and discussion e. None of above

d. Introduction, problem, hypothesis/research question, methods, results, and discussion

Construct validity is about: a. Rationale for the study questions b. Operational definition of dependent variable (DV) c. Purported causal relations between experimental manipulation (IV) and outcome (DV). d. Investigator's explanation/interpretation of findings.

d. Investigator's explanation/interpretation of findings. [How can you "prove" that you understand why an experiment was successful?]

How does a moderator variable affect correlation? a. It increases overall strength of correlation between two variables. b. It is a nuisance variable that distorts computation of variance components. c. It decreases overall strength of correlation between two variables. d. It may change direction of correlation for certain groups. e. None of the above

d. It may change direction of correlation for certain groups.

Mediators are explanatory variables that: a. Represent plausible rival hypotheses b. Increase statistical power of an analysis c. May be controlled with a no treatment control group d. Link independent variables and dependent variables e. None of above

d. Link independent variables and dependent variables

Many methods could be implemented to measure a construct such as depression. Which statement below is incorrect? a. All empirical measures are artifacts of a methodological construction. b. All measuring methods are artificial constructions. c. In general, measurement is a creative byproduct of contemplation. d. Measurement requires a natural unit and nonarbitrary zero. e. None of the above. f. All of above.

d. Measurement requires a natural unit and nonarbitrary zero.

Suppose a psychologist wanted to test effectiveness of high intensity interval training (HIIT) for treatment of narcissistic personality disorder (NPD). She randomly selected a sample of clients from a narcissistic personality disorder clinic and randomly assigned them to a treatment group and a waiting list control group. In addition, she asked each client to give her the name of one friend who does not have NPD and is like the client in age, gender, SES and other demographic variables. The psychologist then recruited those persons from the community and confirmed with a standard assessment that they do not have NPD. The psychologist then enrolled them in a no treatment comparison group. This group was used to help clarify if HIIT reduced NPD behavior in the experimental group. What type of comparison group is this community sample? a. Routine care group b. No treatment control c. No contact control d. Non-equivalent control

d. Non-equivalent control

An experimental study was designed to compare therapy across several groups. This study was conducted three times with a different alpha level, .05, .01, and .001, respectively. If all three studies showed statistically significant differences across groups, which alpha level indicates strongest treatment effect? a. .05 b. .01 c. .001 d. None of above.

d. None of above.

What can we know about "things-in-themselves"? This statement refers to Kant's insight about differences between perception and reality. a. Tentative hypotheses are tested and those results that replicate reveal empirical reality. b. Social conventions define reality, which establishes validity for "things in themselves." c. Empirical replication eventually reveals "truth". d. Nothing.

d. Nothing.

What does rejection of null hypothesis mean in statistical analysis? a. Probability of mean group difference is less than beta when p<.05. b. Probability of mean group difference is less than zero. c. Probability of mean group difference occurred by chance. d. Probability of mean group difference is less than alpha. e. None of above.

d. Probability of mean group difference is less than alpha.

Pretest sensitization sometimes contaminates participants' reaction to treatment, as well as participant responses to posttest assessment. Which design protects against pretest sensitization? a. Randomized control trial b. Quasi-experimental design c. Case study observational d. Solomon four group design e. Retrospective cross-over design

d. Solomon four group design

Which statement below is correct definition of "standard error"? a. Probability of an observation randomly selected from a population. b. Mean effect size divided by variance of sample distribution. c. Standard deviation of sample mean d. Standard deviation of sampling distribution of sample means. e. None of the above.

d. Standard deviation of sampling distribution of sample means.

When are patched-up (i.e. nonequivalent) control groups likely to be used? a. To compensate for nonrandom assignment during experimental designs b. Increase external validity when comparing intact groups. c. To increase statistical power when large groups of patients drop out of a study. d. To increase control over sample characteristics during quasi-experimental designs. e. To decrease probability of Type I error. f. To increase treatment effect size when results are not statistically significant. g. None of above

d. To increase control over sample characteristics during quasi-experimental designs.

In study above concerning yoga and gun violence in Chicago, statistical analysis was conducted of group differences. Two factor ANOVA was conducted, five levels of neighborhoods by pre and posttest attitude assessments showed statistically significant group differences, F = 5.6, p <.05. Magnitude of average group difference was huge, standardized effect size d = 1.8. Treatment group average showed 100% reduction in support for gun violence. Correlation between survey responses and coded qualitative results (positive/negative interview results) was high, >.80. What confounding factors (uncontrolled) might account for these positive results? a. Groups were not randomly assigned, therefore selection bias such as ethnic differences among neighborhoods could account for these differences. b. Age was not controlled. If neighborhoods differed in age, then age could account for attitude differences. c. Income differences d. Educational differences e. All of above

e. All of above

Quantitative and qualitative inquiry are best described by which statement below? a. Their target phenomena are identical. b. Their results are complementary. c. Both require empirical observations. d. Validity and reliability are defined differently. e. All of above

e. All of above

Which of following affect validity of ANOVA results? a. Violation of homogeneity of variance b. Bimodal group distributions c. Non-independent observations d. None above e. All of above

e. All of above

"Contact and attention" is a threat to which of the following? a. Internal validity b. External validity c. Construct validity d. None of above e. All of above.

e. All of above. Other consideration: A. [Participants are responding to experimenter attention and not treatment/manipulation. In other words, significant effect is result of attention and not treatment.] B. [Contact or attention may not appear outside experiment.] C. [Interpretation of significant results is unrelated to participants' responses.]

Which following property corresponds to a linear scale? a. Ordinality b. Monotonicity c. Additivity d. Magnitude e. All the above

e. All the above

Why is multiple-treatment interference a problem? a. Creates Type I error b. Creates Type II error c. Decreases statistical power d. Increases response variability e. Decreases external validity f. All of above g. None of above

e. Decreases external validity [Multiple treatment creates conditions where a treatment may not be effective in isolation. It may only generalize if multiple treatments are presented.]

What problem does selection bias present to internal validity? a. Diffusion of treatment becomes an issue. b. Heterogeneity of sample is decreased. c. Random assignment to groups is invalid. d. Larger control group is needed to accommodate group bias. e. Intact groups are nonequivalent f. None of above

e. Intact groups are nonequivalent [Selection bias means some aspect of compared groups have not been controlled. For example, random assignment may have failed, therefore groups are nonequivalent. Therefore, statistical analysis may not detect effective treatment difference, Type II error.]

Posttest-only control group designs: a. Do not require equivalent groups b. Consist of only an experimental group c. Do not control external validity threats d. Test sensitization problems e. Is conducted without a pretest f. Present higher demand cues to participants g. None of above

e. Is conducted without a pretest

In a study of personality correlates of obsessive-compulsive disorder (OCD) in children, a group of OCD diagnosed children from a clinic were compared with a control group of children randomly selected from the community. Examination of background variables showed overall consistency between groups. Baseline personality assessment, however, indicated statistically large differences between OCD treatment and control groups. A non-specific treatment emphasizing common or non-specific factors of psychotherapy was presented to the control group. After six months of treatment, personality differences between groups were not statistically significant. What conclusion can the investigator draw? a. Treatment was effective for changing children's OCD personality characteristics. b. Personality of children with OCD does not differ from children without OCD. c. Personality of children referred to a clinic with OCD do not have stable personality characteristics. d. Personality characteristics of OCD children are risk factors to the development of OCD. e. No valid conclusions can be drawn from this study.

e. No valid conclusions can be drawn from this study. [Did treatment group improve or did control group decline? Impossible to determine from this design. In general, unmatched, nonequivalent groups are problematic.]

Which following claim to knowledge is empirical? a. Belief b. Opinion c. Authority d. Inference e. None of above

e. None of above

Why are no-treatment control groups problematic? a. They do not account for individual differences in outcomes. b. They are difficult to organize and do not address statistical regression. c. They frequently show higher attrition than alternative control groups. d. They are more effective with intact groups. e. None of above

e. None of above

2. Which following pattern corresponds to a Latin square? a. A B C A b. B C C B c. B A D A d. A B A C e. None of above

e. None of above [Every treatment only appears once in every row (and column).]

Randomized controlled clinical trial should have which following characteristic. a. Self-selection to treatment groups to eliminate selection bias. b. Subsets of participants are assigned to different treatment conditions. c. Random assignment to treatment only group. d. Participants are matched and assigned to control group, then remaining participants are randomly assigned to treatment. e. Participants meeting inclusion/exclusion requirements are randomly assigned to treatment or control groups f. Participants are first assigned to a control group and then randomly assigned to different treatment conditions. [This procedure would create selection bias.]

e. Participants meeting inclusion/exclusion requirements are randomly assigned to treatment or control groups Other Consideration: Participants are first assigned to a control group and then randomly assigned to different treatment conditions. [This procedure would create selection bias.]

A randomized control trial (RCT) is widely recognized as the "Gold Standard" in health sciences research. Which following statement about RCT is correct? a. Intact patient groups are matched on key variables then treatment effectiveness is compared. b. Patients with a defined condition self-select into treatment groups. c. Patients are randomly assigned to treatment and non-specific treatment groups. d. Intact patient groups receive either intervention or standard care. e. RCT typically compares a proposed new treatment against an existing standard of care. f. None of above

e. RCT typically compares a proposed new treatment against an existing standard of care.

Non-Equivalent Groups Design is probably the most frequently used design in social research. It is structured (organized) like a pretest-posttest randomized experiment, but it lacks a key feature- random assignment to treatment and control groups (see image below, NR = No random assignment). Comparing highly different groups leads to what validity threat? a. History b. Statistical regression to mean c. Novelty d. Contact and attention e. Selection bias f. Experimenter bias g. Demand cues h. Diffusion i. None of above j. All of above.

e. Selection bias

Construct validity could be affected by which of the following? a. Participant reactivities b. Test sensitivity c. Statistical power d. Experimenter bias e. None of above f. All of above.

f. All of above. [If any of the above represents major influence on participants' responses, then researcher's purported explanation is invalid (false, incorrect, wrong.]

A study was conducted of ethnic bias during admission to clinical psychology graduate programs in USA. Graduate Record Exam (GRE) scores and demographic data were compared among all graduate programs across forty years (1980 - 2020). Unfortunately, GRE was revised several times during the study interval. Consequently, programs were compared with scores from different GRE test forms. Which threat to validity does this issue raise? a. Testing b. Construct validity c. Testing reactivity d. Response bias e. Testing sensitization f. Instrumentation g. None of the above

f. Instrumentation