Ch.2: research methods Flashcards
empiricism
The belief that accurate knowledge can be acquired through observation
theories
hypothetical explanations of natural phenomena; ideas about how something works
hypothesis
a falsifiable prediction based on a theory
what are the methods (2) used to study human behaviour?
- methods of observation (determine what people do)
2. methods of explanation (why people do what they do)
operational defintions
a description of a property/variable in measurable terms
ie: for example, we might operationally define happiness as “a person’s self-assessment” or “the amount of dopamine in a person’s brain” or “the number of times a person smiles in an hour.”
construct validity
the extent to which one adequately characterizes the property/variable; how well a study conceptualizes it variables
ie: measuring smiles per hour to measure emotions per day over measuring their IQ
power
the ability to detect the presence of differences or changes in the magnitude of a property
empirical method
set of rules and techniques for observation
the scientific method
a procedure for using empirical evidence to establish facts.
naturalistic observation
gathering data by observing behaviour in an ordinary setting, without researcher interference
what are a case study/method and some of its disadvantages
- the deep study of an individual or small group in hopes of revealing universal principles
- research usually on unusual phenomena
- DESCRIBES behaviour (experimentation explains it) what cant always be studied in a lab
disadvantage: large time commitment, the small number group is NOT representative of the population as a whole (experimentation can be used to apply what learned to the larger population(
correlational research
requires at least two variables to be measured so that one variable can be used to predict the other
- NO variable manipulation by researcher
disadvantages of experimentation
some variables cant be manipulated without violating ethical standards: - informed consent safety - privacy of data benefit to society - a benefit to participants
demand characteristics
settings that cause subjects to behave as they think someone else (eg. experimenter) wants or expects
how to avoid demand characteristics?
- naturalistic observation
- if you cant do naturalistic observation:
privacy and control
- gathering info privately or anonymously
- measuring behaviour that people are unable or unlikely to control (such as dilation of pupils to tell whether or not you’re lying)
unawareness
- make sure ppl are unaware of the true purpose of the observation
how can you avoid observer/experimenter bias
double-blind design: neither the researcher not the participant knows how the participants are expected to behave
- avoids the placebo effect
normal distribution
where the most measurements of a frequency distribution are concentrated in the middle
mean = median =mode
what are the descriptive stats of central tendency?
mode: the value of the most frequently observed measurement.
mean: the avg value of all the measurements.
median: the value in the middle
descriptive stats:
s calculated from a distribution of scores, indicating the central tendency (avg) and the variability; brief summary statements aBOUT ESSENTIAL INFORMATION FROM A frequency DISTRIBUTION
- indicated whether or not experimental treatment changed performance
third-variable (confounding variable)?
correlation between two variables cannot be taken as evidence of a causal relationship between them because a third variable might be causing them both
the third-variable (confounding variable) problem?
correlation between two variables cannot be taken as evidence of a causal relationship between them because a third variable might be causing them both
***why correlation DOES NOT = causation
what’s a solution to the third-variable problem?
- experimentation: eliminating or controlling other possible causes, thus creating a a set of conditions that differ in ONLY ONE WAY. Experimentation finds causal relationships
- via manipulation: a technique for determining the causal power of a variable by actively changing its value - randomization
a good detector has ____ and _____
power and reliability
If a person smiles a bit more often on Tuesday than on Wednesday, a powerful smile-detector will detect different amounts of smiling on those two days. If a person smiles exactly as much on Wednesday as she did on Tuesday, then a reliable smile-detector will detect identical amounts of smiling on those two days. A good detector detects differences or changes in the magnitude of a property when they do exist (power), but not when they don’t (reliability).
negatively skewed data leans to the….
right
positively skewed data leans to the
left
standard deviation
how each of the measurements in a frequency distribution differs from the mean; involves finding the distance between each individual score and the mean, and then computing the average of these distances.
- indicate how many exceptions there are.
- when the standard deviation gets larger = the more the two groups overlap and the smaller (weaker) the effect size and vice versa (smaller SD = less overlap = stronger effect size)
if the distribution has extreme scores that pull the mean in their direction, researchers prefer to use the:
median
What’s a limitation of naturalistic observation?
** describes behaviours but cannot explain why they occur which is why we need experimentation
frequency distribution
a graphic representation showing the numner of times in which the measurement of a property takes on each of its possible values
sample
a partial collection of people or animals or things drawn from a population (n); representative of the population
population
a complete collection of people (N)
correlation
when variations in the values of one variable are synchronized with variations in the value of the other
what is direction (r) in a correlation?
the direction of the correlation = positive or negative;
- positive = “more is more”; ppl have a lot of one variable also have a lot of the other; r=1
ie: more health is associsated with more wealth - negative = “more is less”; r= -1
ie: more health is associated with less poverty
*the sign associated wit r
variability
the extent to which the measurements differ from each other (range, standard deviation, etc)
correlation coefficient (r) + scale
mathematical measure that shows both strength and direction of a correlation (how much of an impact the variables have on one another ie: 0.75 or -0.75 = strong correlation vs -.05 or -0.10)
scale:
weak = 0.10, medium = 0.30, strong > or = 0.50
what is strength (r) in a correlation?
the absolute value of r, thus tells you the amount of exceptions to the correlation/rule
r= 1
perfect positive correlation (positive direction)
every time the value of a variable increases (decreases) by a certain amount, the value of the second variable (decreases) increases by a certain amount.
r= - 1
perfect negative correlation
every time the value of a variable increases by a certain amount, the value of the second variable decreases by a certain amount.
r = 0
no correlation
every time the value of a variable increases by a certain amount, the value of the second variable NEITHER increases nor decreases by a certain amount.
imperfect positive correlation
most common in the real world.
there are always exceptions which means r will be somewhere between 0-1 and 0- -1
- few exceptions=r is closer to 1, more exceptions = r is closer to -1
imperfect positive correlation
most common in the real world.
there are always exceptions which means r will be somewhere between 0-1 and 0- -1
- few exceptions=r is closer to 1 (stronger correlation), more exceptions = r is closer to -1 (weaker correlation)
natural correlations
correlations we observe in the world around us
ie: height and weight, sleep and memory, etc
techniques used in experimentation?
manipulation: a technique for determining the causal power of a variable by actively changing its value
random assignment: avoids bias towards a specific type of person
steps of the experimentation technique
- MANIPULATE: manipulate the independant variable
- creates at least 2 conditions - MEASURE: measure the dependant variable (“depends “on the participants)
- COMPARE: compare the variable of the variable in one condition and the value of the variable in the other condition
- if values differ on avg, then the independent CAUSED changes to the DEPENDANT
self-selection
problem that occurs when anything about a participant determines whether they will be included in the experiment or control group;
a type of bias that can arise when study participants choose their own treatment conditions, rather than being randomly assigned
random assignment
a procedure that allows chance (ie coin flipping) to assign participants to the experimental/control group
**used in experimental/causation studies
can random assignment fail? is so, how?
yes.
assortment of participants into group appears biased/not 50/50 split of each “type” of participant in each condition.
- calculate the STATISTICAL SIGNIFICANCE of the data. if p>5%, the results are NOT statistically significant and random assignment failed, p<5%, then results are statistically significant
statistical significane
calculates how likely is it that the sample’s result came from a population in which there is no relationship (if the null hypothesis were true); determined by calculating the probabilities that random assignment has failed, there inferential statistics
- assess the probability that the effect size calculated from one just one sample came from a larger population with certain characteristics
- statistically significant results: p<0.05
- ** a result can be statistically significant but not necessarily important
internal validity
an attribute of an experiment that allows it to establish causal relationships; everything INSIDE the experiment is working exactly as it should
- how well it can rule out alternative explanations
- *conclusions drawn are restricted to the
external validity
when the variables in an experiment have been defined in a normal, typical, or realistic way (variables are representative of the real world); how well the results can generalize to a population of interest.
type I error
error that occurs when researchers conclude that there is a causal relationship between two variables when in fact there is not (a false positive conclusion)
type II error
error that occurs when researchers conclude that there is not a causal relationship between variables when in fact there is (a false negative conclusion)
cross-sectional study
a study in which people of different ages are compared with one another at the same time
- get a snapshot of people at the same time
- more feasible alternative to longitudinal study
longitudinal study
research in which the same people are restudied and retested over a long period of time
- useful for developmental psychology (how behaviours, personality, etc change over time)
disadvantages to longitudinal research
takes alot of money and time
difficult to retain research participants (some move away, pass away, etc)
disadvantages of the cross-sectional studies
cohort effect: the way ppl are affected by their coming from a particular time in history (50year old in 1950s vs 2010)
central tendency
the typical, or representative score in a distribution, often referred to as the “average”
inferential statistics
use sample results to infer what is true about the broader population; #s that are calculated from a distribution of scores to provide evidence supporting or opposing a hypothesis
(indicates the level of confidence in results)
experimental group
participants in an experiment who receive the treatment level of the independent variable
control group
participants in an experiment who do not receive the treatment
null hypothesis significance testing (NHST)
an approach to evaluating research results that compared the observed outcome to what would be expected if the null hypothesis is true
= p>0.05 or p<0.05
when the sample’s result (or one more extreme) would happen less than 5 percent of the time if the null hypothesis is true = reject null and say the research is statistically significant
null hypothesis
no difference between the variables in a population
ie, no difference between the experimental group and the control group
effect size (“d”)
the magnitude of the relationship between manipulated or measured variables.
- the absolute value of the size of r captures the effect size = r of 0.30 = medium effect size, r of 0.80 = strong effect size.
a calculated number that indicates the size of a difference between two values; not affected by the sample size
“d” is used to represent the statistical calculation of the effect size
confidence interval
a range of scores calculated such that there is a specific probability (usually 0.95) that the value of interest (such as the estimated mean of the population) actually falls within that range
- if the confidence intervals don’t overlap = indicates that differences between the two groups are reliably different
p-level
probability of finding a difference that is equal to or greater than what was actually measured, assuming the null hypothesis is true
the probability/odds that the results would have occurred if the random assignment had failed
reliability
ability to detect the absence of differences or changes in the magnitude of a property
- findings can be consistently repeated
3 criteria for causation (making casual claims)
- must show correlation between the variables
- establish the temporal precedence of the causal variable (one variable must come before the other)
- rule out alternative explanations (third-variable)
temporal precedence: establishing that the cause (i.e., independent variable) occurs before the effect (i.e., outcome)
convenience vs random sampling
Random:
ie. random sampling because it used a random-digit dialer to select who was going to be in the study sample.
Convenience:
- there was no manipulated variable in the study, so there was no random assignment.
- a convenient group/choice given the study’s circumstances ( ie. psyc class in uni vs random ample of another population)
- whoever is there for observers to observe at that time
note
Each method has its own strengths and weaknesses, so each one is better for answering some questions than others.
experimental research
- testing whether one variable causes another; establishes causal relationship between variables
- independent (manipulated); dependant (control group)
- ONE OF THE VARIABLES MUST BE MANIPULATED
- uses random assignment
- determines causation
descriptive research
about a single variable and not about a relationship
- Most descriptive studies try to provide concise summaries of one variable at a time.
- often based on self reports
- usually expressed as percentages or frequencies
what happens when you add a data point lower than the mean
, the mean decreases but the standard deviation increases.
what happens when you add a data point higher than the mean
both the mean and the standard deviation increase
how are mean and standard deviation similar?
both examples of descriptive statistics; Both means and standard deviations summarize participants’ different responses.
how are the r and d statistic similar?
Both r (strength) and d (effect size) indicate the strength of the relationship between two variables.
TRUE OR FALSE: a low p-value means the hypothesis is true
FALSE
it just means the data are unlikely to occur under the null hypothesis
- significant result
TRUE OR FALSE: a high p-value means that the hypothesis is false/should be abandoned
FALSE
it just means that the data are likely to occur under the null hypothesis
- non-significant result
meta-analysis
- provides a quantitative estimate of the effect size in the previous body of research
- demonstrate research replication
Saying that a finding is “statistically significant” means that it
is unlikely to have happened by chance if the null hypothesis were true
TRUE or FALSE: every study must have a strong external validity, or the results will not be considered reliable
Many studies sacrifice external validity for internal validity; studies that make this sacrifice are sure not to make extensive claims about their generalizability.
TRUE or FALSE: every study has to reflect the global population
Not every study has to reflect the global population; rather, the participants used in a study should be parallel to the specific types of people that the study aims to analyze.
idea
unorganized principles or thoughts about social behaviour
what’s a variable?
event, situation, behaviour, or characteristics that take on more than one value
conceptual variable definition:
concept or meaning of a variable (what is it)?
ie:
age is conceptual, operationalized to be days since birth (for example)
exercise (conceptual), operationalized (ie: miles run per month)
predictions
- more specific than the hypothesis
- – >uses operational definitions of variable
placebo
A placebo is a sham substance or treatment which is designed to have no therapeutic value.