6, Foundations: Methods and Approaches Flashcards
experiment
- investigation looking for cause and effect
- changes a variable (the cause) and measures how it will affect another variable (the effect)
- only empirical way to prove causation
independent variable
“cause” variable. gets manipulated in an experiment, it affects the dependent variable
dependent variable
“effect” variable. Changes based on the value of the independent variable
control variable
a constant variable in an experiment, should not change even if the independent does
population
group of interest for study
representative sample
a sample from a population to represent the population. Used because populations are often too big to study
representativeness
degree to which the sample reflects the characteristics of the population it is drawn from
experimental group
the group receiving/reacting to the independent variable
control group
the group that doesn’t receive the independent variable but is to be kept identical to the experimental group in all other respects
random sampling
sampling random people from a population
randomly assigned
randomly dividing the representative sample into the experimental and control group
biases
factors that skew the representativeness of a given sample
hindsight bias
I-knew-it-all-along bias
bias of selection
when random sampling is done based on a specific physical space
if you wanted to conduct a survey based on all college students on a campus, you wouldn’t stand in the quad and interview people that walk by because you wouldn’t be representing the students that don’t have classes right now, therefore not random
self-selection bias
when people being studied have control over whether or not they participate in the experiment
highly opinionated people will be more likely to fill out a survey, skewing results
pre-screening/advertising bias
- common in medical research
how volunteers are screened or the experiment is advertised may skew the sample
if a researcher is advertising a treatment that helps to quit smoking, they will receive volunteers who may well have quit smoking on their own anyway
healthy user bias
when the sample is in better shape than most of the rest of the population of study
single-blind design
only subjects do not know whether they are in the control or experimental group
double-blind design
neither subject nor researcher knows who is in the experimental or the control group
- generates more pure results because the researcher cannot inadvertently reveal to the groups which group they are in
placebo
seemingly therapeutic object/procedure that causes the control group to believe that they are in the experimental group
correlational research
determining degree of association between two or more variables that can occur naturally
confounding/third/extraneous variable
an unknown factor playing a role in the experiment
surveys
a tool of correlational studies
either interviews or questionnaires, used to accumulate vast amounts of data to study variable relationships
often used for voter characteristics, criminal behavior, teen alcohol and drug use
longitudinal studies
study over long period of time with few subjects
cross-sectional studies
study to test wide array of subjects from different backgrounds so the study is more general
clinical research
medical research on human subjects
case studies
- common form of clinical research
- intensive psychological studies of single individuals
- intended to draw a generalized conclusion so that the results can be applied to others in similar situations
generalizable
applicability to similar circumstances
conceptual definition
theory/issue being studied
operational definition
the way the theory/issue of study will be observed or measured in the study
internal validity
degree of certainty that results from an experiment can be attributed to the manipulation of the independent variable, not from a confounding variable
external validity
extent to which the results of a study can be applied to the real world
reliability
whether or not the study generates the same results every time from the same inputs
inter-rater ability
the degree to which different raters agree on observations of data from the same study
naturalistic observation
observing and studying real-world behavior outside of the lab
a difficulty is that there are many confounding variables in the real world
descriptive statistics
statistics that summarize data
inferential statistics
statistics that allow researchers to test hypotheses about data and determine how confident they will be in inferences about said data
central tendency
the typical value in a set of data
mean
mathematical average of a set of numbers
mean average of 15 and 30 is 22.5
mode
most frequently occurring value in a dataset
(if there are two, the dataset is considered bimodal)
median
the number that falls exactly in the middle of a distribution of numbers
median of [3, 4, 4, 5, 6] is 4
normal curve
a graph of the averages of statistics, picture a bell curve
range
largest number minus the smallest number
variability
how much the numbers in dataset differ from one another
standard deviation
- average distribution of numbers around the mean
the mean average of 10 and 30 is 20; the same is true for 19 and 21. because 10 and 30 are so much farther away than the mean, there is a greater standard deviation
percentile
how far in a percentage a given number (or statistic) is; if you’re in the 60th percentile of something, you have scored higher than 60 percent of the others
positive skew
most values are on the lower x end of the curve, the largest y values are there also
negative skew
most values are on the higher x end of the curve, the largest y values are there also
correlation coefficient
statistical technique to describe relationships between attributes of study
ranges from 1.00 to -1.00, positive meaning that there is positive correlation and negative indicating negative, higher value of number = higher degree of correlation
Pearson correlation coefficient
type of correlation coefficient that describes how close to linear the relationship of two attributes is
ranges from 1 to 0 to -1, 1 means as x increases so does y, -1 means as x increases y decreases etc
0 means the attributes aren’t related at all
positive correlation
as one attribute increases, so does the other
negative correlation
as one attribute increases, the other decreases
sample size
number of observations or individuals measured
normally denoted by N
null hypothesis
statement that a treatment had no effect in an experiment
alternative hypothesis
statement that the treatment had an effect in the experiment
alpha
the probability that the result of an experiment can be attributed to chance rather than manipulation of the independent variable
lower alpha = less chance in the experiment
Type I error
the incorrect conclusion that a difference between hypothesis and result exists when it does not
Type II error
the incorrect conclusion that there is no difference between hypothesis and result when there is
p-value
the probability of making a Type I error
0.05 p-value means there is a 5% chance of making a Type I error
deception
deceiving participants in an experiment, could be justified by saying that it allows for less bias in the experiment
Stanley Milgram
the guy who made some participants ‘shock’ other participants in the name of teaching, to test the limits of human obedience
confederates
people who know the true nature of the experiment but pretend to be participants
Institutional Review Boards (IRBs)
boards that assess ethics of an experiment before it is allowed to run
informed consent
agreement to participate in an experiment only after adequate information is given about it
debriefing
the time after an experiment where participants are told the purpose of the experiment they just partook in, as well as an explanation for any deception
confidentiality
a requirement that a participants information is not publicized, at least not with their name attached