Experimental Design Basics
Must all experiments contain a control or comparison group? Explain.
No. A control group is ideal, but some experiments (e.g., within-subjects designs) compare the same group to itself at different times, acting as its own control.
When would you use a quasi-experimental design instead of an experimental design?
when you cannot randomly assign participants to groups (e.g., studying the effect of gender, a natural disaster, or a new school curriculum)
What is the difference between the prevalence and the incidence of a disease or disorder?
prevalence is the total number of existing cases in a population, while incidence is the number of new cases over a specific time period.
Distinguish between the mean, median and mode.
the mean is the arithmetic average, the median is the middle score, and the mode is the most frequent score
What is the difference between a correlation coefficient and a coefficient of determination?
the correlation (r) measures the strength/direction of a linear relationship, while
the coefficient of determination (r²) tells us the proportion of variance shared between the two variables
Using an example, identify an independent and a dependent variable in a study.
In a study testing if a new drug (IV) reduces blood pressure (DV), the drug is manipulated and blood pressure is measured. (Answers will vary)
Why is regression-to-the-mean a threat in quasi-experimental designs
because extreme scores on a pretest will naturally move closer to the average on a posttest, which can be misinterpreted as a treatment effect when no true effect exists
What types of measurement tools are used in descriptive research?
surveys/questionnaires, observational checklists, and existing records or archives
Draw a negatively skewed distribution. What does it tell us?
a distribution with a long tail to the left (low scores)
It tells us most scores are clustered at the high end (e.g., an easy exam where most students scored high)
Distinguish between a Type I and a Type II error.
Type I is a false positive (rejecting a true null), and
Type II is a false negative (failing to reject a false null)
Why must researchers ensure that their experimental groups are roughly equivalent before manipulating the IV? How do they ensure this?
To prevent confounding variables. They ensure equivalence through random assignment to groups.
Give an example of a good, interrupted time-series design.
measuring traffic accidents every month for 2 years before and 2 years after implementing a new drunk driving law
The correlation between self-esteem and shyness is -.50. Interpret this correlation.
a moderate, negative relationship: as self-esteem increases, shyness tends to decrease (and vice versa)
What does it indicate if a participant has a z-score of 2.5? What about -0.80? What about 0.00?
2.5 is 2.5 SDs above the mean (rare/high), -0.80 is 0.8 SDs below the mean (slightly low), and 0.00 is exactly at the mean
If the obtained (or calculated) value of t is less than the critical value, do you reject or fail to reject the null hypothesis?
Fail to reject the null hypothesis (the result is not statistically significant)
Discuss the trade-off between internal and external validity. Which is more important?
Internal validity (causal conclusions) often requires artificial lab conditions, hurting external validity (generalization). Internal is typically more important in explanatory experiments, but the 'best' depends on the research goal.
What are some factors that contribute to error variance in a set of data
individual differences among participants, measurement error, and situational noise (e.g., temperature, time of day, researcher fatigue)
Why do researchers often examine scatterplots of their data when doing correlational research?
To check for non-linear relationships, outliers, and restricted range, which can misrepresent the true correlation coefficient.
Why do so few studies in psychology use random samples?
because random sampling (e.g., from a full national population) is logistically difficult, expensive, and time-consuming; most researchers use convenience samples from their university
How many IVs are involved in a 3 x 3 factorial design? How many levels of each factor? How many experimental conditions?
2 IVs, each with 3 levels, for a total of 9 experimental conditions
What kind of variance does a researcher want to maximize? Reduce? Eliminate?
Maximize systematic variance (effect of IV),
reduce error variance (random noise), and
eliminate confounding variance (alternative explanations)
What is effect size? What are the typical effect sizes in psychological studies?
a standardized measure of the magnitude of a relationship or effect (e.g., Cohen's d)? Typical effects are small (d = 0.2), medium (d = 0.5), and large (d = 0.8) in psychology.
Why can’t we infer causality from correlation?
Because of the directionality problem (X could cause Y, or Y could cause X) and the third-variable problem (Z could cause both X and Y)
What are the two primary considerations when determining how large one’s sample size should be?
the desired statistical power (ability to detect an effect) and the expected effect size of the phenomenon (smaller effects need larger samples)
If you wanted to have 20 participants in each experimental condition, how many participants would you need for a 2 x 3 x 3 completely randomized factorial design?
360 participants.
(2 x 3 x 3 = 18 conditions, 18 conditions x 20 participants = 360).