EXAM 2 Flashcards
Quasi-experimental designs
No true IV
Looks like an experimental variable but
Can’t be randomly assigned
Can’t be interpreted as causing changes in the DV or outcome variable
ex– Studying the impact of a new traffic safety law by comparing accident rates before and after the law’s implementation
Extraneous variables
Variable environmental conditions over the course of the study
Individual differences among participants
Fluctuations in the physical/mental state of an individual participant
Confounding variables
an extraneous variable that covaries with your IV
It provides an alternative explanation for your findings
Threat to the internal validity of your study
Between-subjects design
Different participants are assigned to each group or condition.
Each group is exposed to a different level of the independent variable, allowing for comparisons between groups.
Random assignment/ Randomized designs
Comparing the test scores of two different groups of students, one that received the tutoring program and another that did not, to determine if the tutoring had a significant impact on math skills.
Within-subjects design
Same participants in all conditions
Repeated testing of individuals
Factor
independent variable
IV takes on different values called
Levels
Conditions
Groups
Treatments
Single factor
manipulating one independent variable with more than two levels or conditions to observe its impact on a dependent variable
ex– Experiment: Testing the Effect of Fertilizer Types on Plant Growth
Independent Variable (Single Factor):
Three different types of fertilizers:
Fertilizer A
Fertilizer B
Fertilizer C
Two group design
Simplest design
1 IV with two levels/groups
Multigroup design
1 IV with 3+ groups
Power to detect nonlinear relationships
Ex– caffeine consumption and speed of encoding
Randomizes
Random assignment to hopefully make groups equivalent on extraneous variables
Relies on chance– not guarantee that groups will be equivalent
Matched
Matched Design: Ensures groups are equal on matched characteristics.
Matched-Pairs: 2 conditions, 2 participants per match
Matched Multi-Group: 3+ conditions, 1 participant per match.
If no match, no study participation.
Solution for matched groups design
Recruit a sample of participants
Measure the participants on the extraneous variable you want to control
Group participants who scored the same (or close) on the matching variable
Randomly assign participants in the match-groups to each of the conditions
Matching vs. random assignment
Matching– guarantee
Random assignment– chance
The more participants you have, the more likely it is that random assignment will be successful
Within subjects design
The same participants are used in all conditions
Logic is similar to the matched-groups design
Threats to validity
Purpose is to control individual difference extraneous variables
ex– testing the same group of students’ math skills before and after a tutoring program to see if there is an improvement in their scores.
Does a pre-test lead to a post-test within subject-design
No
Pre test and post test are not levels of IV
pre test is a baseline and post test assesses the effects after the experiment
Disadvantages for within subjects design
More demanding on participants
Especially if your IV has a lot of levels
Drop-out
Participants failing to complete the study
Carryover effects
Exposure to one condition affects performance in a subsequent condition
Testing affects
Performance changes over time, unrelated to IV
Sources of carryover
Learning, contrast, habituation, dishabituation, sensitization
Learning
Participants don’t just unlearn what you’ve taught them in a previous condition
Contrast
Participants may compare treatments, affecting behavior
Habituation
Repeated exposure to a stimulus that may decrease responsiveness to that stimulus
ex– traffic noise outside of your apartment becoming very unnoticeable because you are so used to it
Dishabituation
when an organism responds to a previously familiar stimulus as if it were new again.
ex– background music: you get used to it but you would notice if it stopped or changed in a noticeable way
Sensitization
when a person or organism becomes more responsive or reactive to a stimulus after being exposed to it, often due to a previous intense or disturbing experience
Testing affects
Fatigue– Participating in repeated trials can result in poorer performance in later conditions
Exhaustion, frustration, boredom, etc
Practice– Participating in repeated trials can result in better performance in later conditions
Attrition
random dropout
Ex– people move away from area
Unrelated to the topic of study
Participant mortality
unequal rates of dropout related to topic of study
Ex– results in changes to the sample
Can result in confounding
Worse than attrition
Threats to validity
History– historical/cultural change occurs during study in between your measurements
Maturation– Development occurs between measurements
If you are already studying development this is not an issue
Counterbalancing
Equal numbers of participants experience different orders of conditions.
For example, half start with A, then V, while the other half start with V, then A.
Order is assigned randomly.
Full counterbalancing ensures fairness and minimizes order effects in experiments.
of different orders= number of conditions
2 conditions = 2 X 1 = 2
3 conditions= 3 X 2 X 1 = 6
4 conditions = 4 X 3 X 2 X 1 = 24
Latin square design
When you have 3 conditions, you create 3 orders.
It alternates the order of conditions, like A before B, or B before A, and similarly for A and C, and B and C.
Helps prevent order effects in experiments with multiple conditions.
Why use different orders
Constant Order: When you always present conditions in the same sequence.
Impact on Internal Validity: Can mix up the effects of conditions with the order they’re presented in.
Balancing Effects: Helps distribute any order-related impacts evenly, making them easier to ignore unless they’re unusual (e.g., contrast effects).
Detecting Carryover: You can spot carryover effects by looking at how performance varies with different orders, making the order itself a factor in your study (included in statistical analysis).
When to use a within-subject design
Best to use when the extraneous variable is related to the independent variable
Best to use when it is difficult to find/recruit participants
You need less people
2X2 factorial design
IV #1 has two levels and IV #2 also has two levels
2X2 = 4– four different conditions
Each condition is a combination of the levels of the two different IV’s
Example–
Rehearsal, US words
Rehearsal, UK words
Rehearsal and imagery, US words
Rehearsal and imagery, UK words
3X2 factorial design
IV #1 has three levels and IV #2 has two levels
six different conditions
Example–
Rehearsal, US words
Rehearsal, UK words
Rehearsal and imagery, US words
Rehearsal and imagery, UK words
Elaboration, US words
Elaboration, UK words
2X2X2 factorial design
Adding another independent variable– 3 IV’s
IV 1 has two levels, IV 2 has two levels, and IV 3 has two levels
Eight different groups
Easiest to do two different tables
Factorial designs
Can have factorial designs for either between subjects or within subjects designs
Can have factorial designs for randomized or matched designs
Purpose of factorial designs
Manipulation two variables in the same experiment is more efficient than conducting two studies
Can look at the interaction or joint influence of the two IV’s
Why do scores vary?
Independent variable
- Quasi IV (not manipulated, preexisting difference)
- Extraneous variables
o Testing conditions, individual differences
Population and Sample
Every sample you take will have a different mean
- Population has only one mean (Population mean)
- Each sample has a mean of its worm (Sample Mean)
The difference between a sample mean and the population mean = Sampling Error
farther away the sample mean is from the population mean, the larger the error
Standard Error (of the mean)
Tells us how close or far sample means are from the population mean
“How much do sample averages differ from the true average in the entire population?”
Std dev
the more variability there is in the scores themselves, the more variability there will be in the means you obtain from different samples
understanding the spread or consistency of scores or results from your experiment
N – sample size
the smaller your sample is, the greater the likelihood that the sample mean will differ from the population mean
P value
P < .05 is good – it means we trust that observed differences weren’t just due to luck
Degrees of Freedom
Used to find the critical value (p= .05) of a statistic
Null Hypothesis (H0)
- No difference in means
- Mu = Population mean
ex– caffeine consumption and ability to code has no correlation
Alternative hypothesis (H1)
- Samples came from different populations
- Means are different
- same as research hypothesis
If the probability is less than or equal to .05
- The difference is statistically significant
- Reject the null hypothesis
If probability greater than .05
- Fail to reject the null hypothesis
- Difference is nonsignificant
We never accept the null hypothesis
We say that the groups were not significantly different, not that they’re the same
Results are nonsignificant, not insignificant
A result that’s significant at p < .01 isn’t more significant than one at p < .05
