Chapter 2 Methods of Psychology Flashcards
Bias
&
Random variation
👽BIAS
(= nonrandom effect caused by factor/s extraneous to research hypothesis)
- > researchers might think that hypothesis is supported but
- > factor/s irrelevant to hypothesis (BIAS) cause observed results
SERIOUS PROBLEM
-> statistical techniques cannot identify/correct it
-> NOT correctable by averaging
⚡-> results are ✔️NOT statistically significant
👽Random variation / error
-> average can correct it
⚡-> the higher variability of data the less likely results are to be statistically significant
But they ✅😊ARE statistically significant (oppsed to when results are biased)
- > RANDOMLY ASSIGN PEOPLE
- differences only due to error/random variation
Biased sample
Biased sample NOT representative of larger population
-> can’t draw general conclusions from the sample for the population
PROBLEM
-human subjects that are easily available to be studied (like psychology students) may not be representative of population
Reliability
of measurement procedure
If we do the same study with the same measurement procedure again, how likely are we to get the same results?
How reliable is our measurement (procedure)?
🔥⚡LOW RELIABILITY =
SOURCE OF ERROR / RANDOM VARIABILITY
F. Ex. Measurement procedure: psychological test
- > greatly affected by mood of subjects - > observed results subject to random variation (because of random variation in mood)
Interobserver / interrater reliability
Is the same behaviour seen by one observer also seen by other?
Need to carefully define which behaviour we wanna observe
Operational definition
Specifying exactly what observable behaviour (which we measure) should look like
(-> so we can observe it and know if we’re observing the behaviour we wanna observe)
F. Ex. Operational definition of aggression
- Child hits others
- questionnaire to measure aggression
Validity
Are we measuring what we want to measure?
-> is our measurement (procedure) valid?
(-> then it has “face validity”
F. Ex. Test that assesses degree of shyness (measurement procedure) has face validity for measure of personality)
⚡🔥LACK OF VALIDITY = SOURCE OF BIAS
F. Ex. Biased group: people who are motivated to better their depression are all in one group
-> now we’re not measuring what we want to
- > we’re not measuring if people with depression respond well to psychotherapy in general
- > we’re measuring if people who are motivated to better their depression respond well to psychotherapy
- > we’re not testing if Hypothesis is true
Measurement procedure
Can be RELIABLE (same results when study reproduced)
But NOT VALID (measuring sth else than we want to)
Within-Subject Experiment
Each subject each tested in each condition of independent variable
(subjects repeatedly tested)
OR
just 1 subject tested under varying conditions of independent variable
F. Ex. Clever Hans
Between-Groups Experiment
Different groups tested under varying conditions of independent variable /
Manipulations of independent variable applied to different groups of subjects
observer-expectancy effects
=> Biases
Observer has certain wishes/expectations that affect how they behave and what they observe
F. Ex. Clever Hans
Researcher wants/expects horse to respond in particular way and unintentionally communicates this expectation & influences subjects behaviour
blind observer
How does blind observer prevent 2 biases of observer-expectancy effect?
observer blind (uninformed) about aspects of study that could lead him/her to form biasing expectations
F. Ex.
Observer doesn’t know which group gets treatment
-> Doesn’t have expectations regarding the behaviour of the groups
1.
-> no confirmation bias (f. Ex. expects them to smile more, interprets more facial expressions as smiling)
2.
-> doesn’t influence their behaviour through behaving differently with each group
Subject-expectancy effect
=bias
Expectation of subject leads to effect and not treatment itself
F. Ex. Believe that psychotherapy treatment will work -> improve because of that (placebo)
Placebo effect = Subject-expectancy effect
Hawthorne effect
Workers got more productive because they believed they were receiving special treatment (& bc they knew they were being watched)
Double-blind experiment
Subjects and observers blind to / uninformed about treatment
To prevent BIASES
F. Ex.
Some subjects receive drug, some placebo (inactive substance that looks like drug)
- double-blind experiment - NO BIASES
- all subjects take sth - all subjects have PLACEBO EFFECT (belief that it will work causes it to work) /Hawthorne effect
=> any observed difference between subjects who did and didn’t get the drug due to drug’s chemical qualities
Theory
idea / conceptual model / explanation
explains existing observations
makes predictions about new observations (hypothesis)
Observation -> Theory -> Hypothesis
Hypothesis
Prediction about new observations made from a theory
Then: testable prediction
F. Ex.
Theory: Horses have humanlike intelligence
Hypothesis: Hans can give correct answers to verbally stated problems/questions
Variable
Anything that can change / assume different values
(Anything that’s observed in a study)
F. Ex. Temperature, amount of noise, score on test, eye colour
Experiment
Procedure in which researcher systematically
manipulates/varies on or more independent variables
to see the changes in the dependent variable
while keeping all other variables constant
=> change in dependent v. CAUSED BY change in independent v.
Correlational study
Researcher doesn’t vary/manipulate any variable
Observes two or more already existing dependent variables to find relationships between them
=> when we identify relationships we can make predictions about one variable based on knowledge of another
(f.Ex. high previous test scores predict high test scores in future tests)
=> CANNOT say whether change in one v. is CAUSE for change in another
Descriptive study
Describes behaviour subject/s without assessing relationships between them
(assess=ermessen, feststellen)
descriptive studies can or can’t make use of numbers
F. Ex. No numbers
Observing courtship behaviour of ducks to describe sequence of movements involved
(courtship b. = Umwerben, Balzverhalten)
Laboratory study
➕Advantages?
➖Disadvantages?
Subjects brought to designated area
Researcher has control over subject’s experiences
-> EXPERIMENTS most often conducted in laboratory studies
➕ Greates control over variables
➖ Beaviour could be unnatural (due to unnatural env. and knowledge of being observed)
Field study
➖Disadvantages?
➕Advantages?
Subjects’ natural environments
Researcher has no control over their experiences
-> CORRELATIONAL & DESCRIPTIVE studies most often conducted in field studies
➖ Less control over variables
➕ Natural behaviour more likely
(BUT observer can’t always be completely unobtrusive f. Ex. observing children in classroom -> they will notice)
Self report methods
➕Advantages?
➖Disadvantages?
Rate/describe own behavior/mental state
- questionnaires
- interviews
➕ Information that can’t be obtained from observing behavior
➖ Validity of data limited
Subjects don’t report truthfully
-wanna look good in front of researcher
-biased
f. Ex. Availibility heuristic
(people put more weight on info. That comes to mind easily)
F. Ex. I behaved very openly and communicative at last party - report that I am open in general but I am not
Introspection
➖Disadvantages?
➕Advantages?
one SELF-REPORT-METHOD
personal observation of one`s thoughts, feelings, perceptions
➖ not directly observable, can`t be confirmed -> can never be sure if info. is correct
➕ modern methods -> measuring neural activity
- > neural activity correlates with introspections reported
- > provides more objective “observable behavior”
Observational methods
Researchers observe & record behavior
(rather than relying on subject`s self-report)
- tests
- naturalistic observations
Tests
➕Advantages?
➖Disadvantages?
researcher presents problems / tasks / situations
to which subject responds
F.Ex. subject wins money and can decide how much to donate -> test generosity
➕ convenient, easily scored
➖ artificial,
relevance to everyday behavior is not always clear
F. Ex. What is relationship between score on IQ-Test and ability to solve real-life problems?
Naturalistic observations
➕Advantages?
➖Disadvantages?
=> Field studies, Descriptive studies
=> Observing behavior in natural env. without interfering
F.Ex. Watching people pass by charity booth to see if/how much they donate
➕ learn firsthand about subject`s natural behavior
➖ limited practicality
- great amount of time
- difficulty of not interfering
- difficulty of coding results so they`re usable for statistical analysis
Hawthorne effect
changes in subjects behavior as a result of knowing they are being watched & BELIEVING THAT THEY
RE RECEIVING SPECIAL TREATMENT
=> SUBJECT-EXPECTANCY EFFECT
- > diff. techniques to improve worker`s performance
- > improved bc they were being watched + belief that they were receiving special treatment NOT due to the diff. techniques (f.Ex. diff. lightning, schedules)
descriptive statistics
summarize sets of data
inferential statistics
mathematical methods
that help researchers determine how confident they can be in drawing general conclusions (inferences) from specific sets of data
-> determine how likely the results observed are due to chance
mean
arithmetic average
sum of scores divided by number of scores
median
50th percentile
center score in a set of scores that have been rank-ordered
if it`s an even number of scores you take the 2 scenter scores, add them and divide the sum by 2
f.ex. 1, 2, 4, 5, 7, 8, 9, 9
Median: (5+7) : 2 = 6
Variability
measures of variability?
degree to which scores differ from one another and from the mean
common measures of variabilty:
variance s^2 & standard deviation s
standard deviation
when is standard deviation s the greatest?
typical distance of an observation from the mean
the farther most individual scores are from the mean, the greater is the standard deviation
F. Ex.
Set A 7, 7, 8, 11, 12, 12, 13 Mean = 7 s = 2.39
Set B 2, 4, 8, 9, 14, 16, 17 Mean = 7 s = 5.42 !!!
correlation coefficient
in correlational studies we observe a correlation between 2 or more variables (X and Y)
if both/all are measured numerically, we can calculate the correlation coefficient r
ranges from -1 to +1
positive correlation: X increases while Y increases
negative correlation: X increases while Y decreases
SRONG CORRELATION: r close to -1 / +1
-> can predict value of one variable by knowing other
F. Ex. high previous test scores predict high scores in future tests
MODERATE CORRELATION: r between 0 and -1 / +1
f.ex. r = 0.51)
WEAK / NO CORRELATION: r close to 0 / 0
=> variables statistically unrelated
-> can`t predict value of one variable by knowing other
F. Ex. Students hight not correlated to their grades
statistical significance
how small/big is likelihood of observed results being due to chance
f.Ex. BY CHANCE a lot of people that are motivated to better their depression in one group -> corellation of treatment & degree of depression due to CHANCE
statistically significant: p < 0.05
-> bserved effect/ data is right to 95% (only 5% due to chance)