Psychology 270- Research Methods
Cluster
Clusters of individuals are identified and then a subset of clusters is randomly selected to sample from Adv: makes it easier to choose members randomly from small clusters to better represent the population Dis: Ignore segments of people
Quasi-Independent variables
Causal relationship is not always clear as for a truly independent variable -> Same as subject variable
IACUC
Ethics for animal research
• History Effects
History events affects posttest in pretest-posttest study I.e 9/11 events- economy change etc
Coercion
Forcing participants to participants without their consent
Descriptive v. Causal Research Question
Descriptive asks questions about the presence of behavior while causal asks what causes a specific behavior.
Simple Random Sample (Say Adv and dis)
Each member of the population has an equal chance of being selected using random sampling Adv: Reduce Sample Error Dis: Difficult to ensure each member of a large pop can be chosen in a sample
Measures of Central Tendency
Mean, Median and Mode
Inter-rater Reliability
Measure of the degree to which different observers rate behaviors in similar ways; typically meant for natural observation
Volunteer
Members of population are chosen based on convenience and on who volunteer Adv: easy to obtain dis: not representative
Quota
Members of population are selected such that the proportion of a group in the sample is equal to the proportion of the group population Adv: easier to obtain, allows for better representation dis: not represent population because random sampling is not used
Type 2 Error
Null Hypothesis is false and one failed to reject the null
Type 1 Error
Null hypothesis is true but one rejected the null hypothesis
• Testing effects
Occur when participants are tested more than one and it affects later testing
• Social Desirability Bias
Participants responded according popular social beliefs
Convenience/ Purpose Sample
Probability of choosing an individual in the population is not known.
Variability
Range, SD, Df
• Nonresponse Bias:
Individuals chosen for the sample do not respond, biasing the sample
Ratio scale
True zero point, can say 2x as much
Deductive Reasoning
Using general information to make specific predictions (theory-driven hyp.)
Regression Toward the Mean
When a participant scores an extreme scores, either really high or low and the next time they take it score it close to the mean, test is not unreliable
• Internet samples
maybe more diverse and less biases, fewer constraints based on date and locations
Cronback's Alpha
measures internal consistency: average correlation between all scores on all paired questions on a survey
• Demand characteristics
bias that participants change their behavior based on their perception of the study and its purpose
Degrees of Freedom
n-1: number of scores that can vary in the calculation of statistics
Maturation
natural changes that occur to the participants during the course of the study that can result in a bias
Construct Validity
needs a good operational definition, indicates that the study measures what it intended to measure
Nominal
no ordered, categories, qualitative data
Qualitative Data
nonnumerical data
Within Subject Design
each participant experiences all levels of variable
Between Subject Design
each participant experiences only one level of the variable
Quantitative Data
numerical data
p-value > is alpha=
cannot reject null
Counterbalance
a control used in within subject experiment where different participants are assigned in equal numbers to the different orders of the conditions (within subject)
Ordinal
a scale of data measurement that involves ordered categorical responses, not equally spaced, ex: very anxious to not at all anxious
Factorial Design
experiment or quasi experiment that includes more than one independent variable
Empiricism
gained knowledge through systematic observations of the world. Observing behaviors gives research gives a more accurate understanding of the cause of behaviors than other methods of gaining knowledge. Ex: watch, survey, make people complete a task
Content Analysis
an archival data collection technique that involves analysis of the content of an individuals spoken or written record
• Field experiment
an experiment conducted in the participants natural environment
small n-design
an experiment conducted with one or a few participants to better understand the behavior of this individuals.
Confounding Variable
an extraneous factor present in a study that may affect the results
Multivalent
an independent variable that has 3 or more levels
Bivalent Independent Variable
an independent variable with two levels
Standard Deviation
average distance of the scores and the mean in a given distribution, squaring that value and adding them up dividing by degrees of freedom
Single-Blind vs. Double-Blind Design
hide their group assignment vs. participants and researchers are unaware of group assignment
• Open-ended
participants can put whatever they want to put as their answers
• Attrition/mortality- What kind of threat?
participants choose not to compete the study- they do not come back to finish ie. drop out Internal and External Validity threat
• Two- tailed hypothesis
considers both directions, (memory changes over time); verses one-tail is when it is directional (older people have lower memory scores than young...)
System Observation
data collection techniques where control is exerted over the conditions under with the behavior is observed
Operation definitions
definitions of abstract concepts
IRB vs. IACUC
human vs. nonhuman ethical research committee
Data-Driven Hypothesis
hyp. for a study based upon the results of previous related research
Theory-Driven Hypothesis
hyp. for a study that is based on a theory about the behavior of interest.
Interval scale
involve numerical responses, quantitative data, numerical categories that are equally spaced, Likert scale.
Case Studies
involves intensive studies on a particular subjects that is often exploratory when the researchers in learning about a behavior when little is known about it.
Content analysis
is a specific type of archival data observation technique that involves analysis of what someone has said or written- a coding scheme must be designed.
Predictor variable- Scatter Plot
predict the score of another variable
• P-value
probability level associated with an inferential test that indicates the likelihood of obtaining the data in the study when the null hypothesis is true
p-value is ≤ alpha
reject null; significant accept alt.
• Internal Consistency
reliability tests relationships between scores on different items of a survey.
Authority
relying on the knowledgeable person or group as a means of knowing about the world
• Convergence bias:
sampling error that occurs when the sample chosen to complete the survey does not provide a good representation of the population
Split half reliability
scores are similar in different sets of questions on a survey that address similar topics
Nuremberg Code
set of ethical guidelines developed for research with human participants based on information gained during the Nuremberg trails after WWII; ~first ethical code
Main Effect
test of the difference between all means for each level of an independent variable in an ANOVA.
Interaction Effect
tests the effects of one independent variable at each level of another independent variable in an ANOVA
Testability
the assumption that explanations of behavior can be tested and falsified through observation.
Determinism
the assumption that phenomena have identifiable causes. Ex: why do people get depressed? Starts with a prediction and the information gathered from the research is either for or against that prediction.
External Validity
the degree to which the results of a study apply to individuals and realistic behaviors
Reliability
the degree to which the results of study can be repeated under similar conditions
• Sampling Error
the difference between observation in the sample and the population, sample value is unlikely to hit the population value
Third- Variable Problem- threat?
the presence of extraneous factors in a study that affect the dependent variable can decrease the internal validity of the study
• Alpha level
the probability level used to indicate a cutoff probability that will allow rejection of null hyp.
• Test-Retest Reliability:
the scores were consistent over time
Parsimony
think reductionism- the assumption that the simplest explanation of a phenomenon is most likely to be correct- this is called "Occam's Razor". This is especially true if complex explanations do not have any support. Start simple and then expand to complex if it cannot be explained simply.
Nonverbal response scale
those that show pictures rather than words for children or those with low verbal skills ex: how do you feel? :(, :/, :I, :}, :)
• Experimental Bias
unintentionally the researcher treats one group differently from another
Deduction
using logical reasoning and current knowledge as a means of knowing about the world
Inductive Reasoning- What driven hyp?
using specific information to make more general prediction (data driven hyp.)
Outcome variable-Scatter Plot
variable being predicted by the predictor variable
Dependent variables
variables that are measured
Amount Variable
variables that include levels with a different amount of the treatment changing from variable to variable
• Hawthorne Effect- What threat is it?
when participants change their behaviors due to the fact that they are being observed External Validity threat
Internal Validity
will provide a good test of causal relationship by removing alternative explanations of the data
External Validity
• The degree to which the results of the study apply to individuals and realistic behaviors outside the study
Response Rate
the percentage of people out of the total number available who responds to a survey
• Closed-ended Response
Likert scales, too little response= cause bias, too many= confusing although increases validity, make sure there are equal number
Order Effects
Order of conditions in a within-subjects design can affect data collection in different condition
Pretest-Posttest Design
a type of research design where behavior is measured before and after a treatment or condition is implemented
Presence/Absence Variable
a variable that involves a manipulation with a level that involves that treatment a level that does not involve the treatment
Face Validity
on the surface, a study or scale appears to be intuitively valid
Common Ethic Problems in Reporting Research
-Plagiarism - Falsification of Data - Giving proper credit
Stratified Random Sample Adv/dis
Proportion of group in the sample is equal to the proportion of the group in the population using random sampling Adv: Reduces bias to an identified characteristic of the population by equating proportion in the sample and the population for that characteristic to better represent the population Dis: can be difficult to ensure equal probability of being chosen from a large population
Intuition
Relying on common sense as a means of knowing the world
Observation
Relying on what one observes as means of knowing the world
Probability Sample
Sample chosen such that individuals are chosen with a specific probability (basically all types of sampling techniques)
Matched design
a between-subject experiment that involves sets of participants matched on a specific characteristic with each member of the set randomly assigned to a different level of the independent variable (between subject)
• Criterion-Related Validity
determining the validity of the scores of a survey by comparing it to scores and established measures
Range:
difference between the highest and the lowest score, crude because it ignores everything in the middle of the distribution
Quasi-Experiment
type of research design where there is an experiment but random assignment is not possible i.e age/gender/smoke-notsmoke