chapter 2 Flashcards
limits of intuition and experience
The way to compensate for our potentially biased intuitions and narrow range of experiences is to systematically collect and average evidence across many ppl
This is what it means to study psych as sceince
We should draw conclusions usuing scinetfiic methods rather than relying on intuition or experince
Scientific method
Scientists are empircists meaning they base beliefs on systematic objective observations of the wolrd
Instead of using experince or intuition, psycholgists test theri ideas by observing ppl directly
Theory data cycle
Scientifc method follows process of theory data cycle
Theory data cycle involves developing a theory abt what ppl do and collecting data that are compared with the theory - the data either cofnrim or disoncrim the theory
Theory
Set of propositions abt what ppl do and why
To test a theory researchers design a study with a specific hypothesis
Hypothesis
A prediction about what will happen based on the theoiry
Data
oberservations from a study usually in numerical form, collected from ppl at a certain time or in certain situations
Replication
Even when a single stidy supports their theory, researchers strive for more evidence because an imprtant feature of the scientifc method is replication
Replication means that the stidy has been conducted more than once on a new sample of partipcants and found the same basic results
Journals
In final step of sicnetific investigation, rsearchers make reuslts public
They share their scientfic research in specialized publications known as joruansl
Peer reviews ensure that teh research published in horunals is innovative, important and well conducted
Variable
To test a theory, researchers must choose and define the variables taht interest them
Variable is something of interest thatc an vary form person to person or situation to sitation
Variable has at least two leevls or values
Measured and manipulated variables
VAriables can be measured or manipulated
Measured variables are used in every study - they are observed and recorded
Manipulated variable is one whose levels the researcher control by assigning diff partipcants to diff levels of that variable - usually only used in expereimts
Some variables can be measured or maipulated and some can only be measured (like gender)
Operational definitions
To test a theory using sci method, research have to convert concpet of variable to qunatifiable and measurable variable - they must operationlize variable but creating operational definitions that specifiy the exact process for dterming the levels or values of each variable
Usually just turning a variable into a number so researchers can statistcally analzye data and evaluate strength of evidence for hypothesis
Self report
A way of operationlizing variable
Ppl asjed to describe temselves in survey or interview
Usually asks many similar questions
Direct observation
anotehr form of operationlzing
Researchers will specify what behaviours they are looking for
Tech can be used to observe behaviour like social media or GPS
tech can operationlize variables taht are not easily observable by researchers or particpants themselves
Types of research emthods to test theroies
Descriptive, correlational and experimental
Descriptive research tells us what do ppl do
Correlational tells us what kinds of ppl do this or whats associated with what
Experimental tells us why do ppl do this or what causes these behavours
Descriptive research
Researched focus on one measured variable at a time with the goal of describng what is typical
Descriptive research asks what do ppl do on average
Self report
Descriptive research is based on self report which usually takes teh form of a survey
Survey research provides concise summaries of a lot of ppl
Surveys allow reseacrhers to use a small group of ppl to draw conclsuions abt the larger group of ppl they came from
- The small group of ppl is sample and they particpate in reseach
- They are selecred from a larger group known as the population of interest which are the larger set of indivduals teh researcher is trying to understand or describe
Random sampling
For a sample to generalize to a population, ppl of the population of interest must have equal chance of sleection which is done by random sampling
ex. like calling random digits
Truly random sampling techniques require special planning to avoid bias
Naturalistic observation
Other descriptive method rely more on observational data where psycholgists measure vairables of interest by observing and recoridng ppls behaviour or performance
In naturalistic observatuion, psychologists observe the ebhaviour of animals or ppl in their normal, everyday wolrds and envirmnets
The goal of naturalistic research is to observe iwthout interfeing with the uaual behaviour of those being observed - so scientists will maje tehmselves unobtrusive, bledning in so the ppl theyre observing dont notice their presence
Observational research
Another method for descriptive reseacrh
Psycholgists measure their variable of interest by observing and recording what ppl are doing
Can use tech for this
Researchers also use large data sets called big data to conduct observational research
Case study
another type of method for descriptive research
These are rare conditions that are studied
An indepth exmintion of one persons experience, abilities and behaviours
These stuides can lead to theortical insights that inspire other types of research
Correlational research
Researchers measure two or more variables in order to understand the relationship between them
Can tell us how two or more variables relate to one another
Scatterplot
Data from correlational studies can be presented in a scatterplot
Each dot reps a participant, the x axis is one variable and y axis is the other
When researchers conduct correltaional research the association they observe betwene two variables can be pos, neg or zero
Correlation and causation
Even though a correltaion that is pos or neg can help us predict one vairbale from another, a correltaion, even a strong one, doesnt allow us to say that one variable causes the other
To be convinced that one variable causes the other we have to satisfy 3 conditions:
- The two variables must be correlated
- One variable has to come before the other
- There must be no reasonable alternative explanation for the pattern
Correlational studies can satisfy the first but uually not the other two
Third variable problem
Whenevr a correlation observed between two variables is actually explained by the infleunce of some third variable
one correlational study cannot rule oyt all pissble third variable probelms
Correlational studies can never support causal claums becaus ethey can never rule out all possible third variables
Experimental research
conducted in a way that can support causal claims
When conducting an experiment, researchers manipulate a variable that is hypothesized to be causal - researchers then assess the effect of that manipulation on one or more measured variables
Independent variable
In an experiment, the manipulated variable or the hypotheiszed cause
Dependent variable
the measured variable or the hypothesized effect
Random assignment
A procedure used in experimental research in which a random method is used to decide which participants will be in which group
It allows reseacrhers to assume that at the beggining of the study the people in each group are on average similar
Experimental and control group
The group where the change is applied is experimental group
The control group is the condition in which it is absent
The control and experimental groups should have the same experinece in the study besides the one vairable being manipulated
Double blind procedure
Both the particapnt and researcher who has contact with them are unware of the condition assignment
Having partcapnts blind to condition ensures they act spontaneosuly not in a bias way
This precaution ensures that reseacrhers interacting with particants do not subtly infleuce them or have bias in percpetions
Random sampling vs Random assignment
Random sampling also known as random selection is a method for selecting particapnts who will be in a study - if it is used, every person in the population of interest has an equal chance of being selected - when a stdy uses random smapling the results can be generalized to the population of interest - it can be used in any study - survey, obervational, correltaion or expierment
Random assignment is used in experiemnts - method used to assign participants to diff levels of the indepent variable - it helps rule out alternative explanations for any differene between experimental groups
Comparing diff research methods
Each method has both strengths and limitations
Descriptive emthod provide info on what ppl typically do but not abt who does the vehaviour in what circusmtances - descriptive studies can assess many variables but researcher needs correlational methods to assess relationships between variables
Correlational studies dont allow for causal claims but can tell how variables are related to other things
Experimentas have value of isolating causal effects but reseacrehs are limited to manipulating one or two things
Reseacrhers might start with descriptive study to find out what is typical, then move onto correlational methods to establish relationship between variables and then use experiments to estabslih causation
Validity
The appropriateness or accruacy of a claim or conclusion
Ask:
- How well did researchers operationlize variable
- Are the ppl they studied reps of the population of interest
-Can we rule out most plasuible alt explanations
Construct validity
How well the variables were operationalized - how well do they capture the variables of interest
How were the variables operationlized
Are the questioanrres, behaviours or data good
Threat to consyruct validity can come form measures that are unreliable - reliability is the degree to which a measure yields consistent results each time it is administered
A form of relaibility is the consitency between two observers
How well was the independt variable manipulated
External validity
When a sample in a tstudy can generalize to the populatuon of interest, the stdy is said to have extrnal validity
Ask who did they study, what types of ppl did they include and how did they select them
Random sampling can help create good external validity
External valdity aslo concerns whether results form one pop of interest can generlize to another
Threats to external validty are often tolderated in research as long as researchers are careful to not claim their findings genrlaize more broadly than the sample allows
Internal validity
can we rule out alterntve explanations
When the study is experiment then its easy to rule out third variables
Ask whether tehre could be other explanations for the relationshio (thrid variable prob)
Studies with confounds have poor internal validity because those confounds become alt explanations
Confounds
When the experimental groups accidnetally differ on more than just the independent variable
5Q for helping determine valid claims in media
- What am i being asked to believe
- What kind of evidence is there?
- How strong is the result?
- Are they making a causal claim?
- Has the study been replicated?
Descriptive statistics
These summarize partcipants differing resposnes in terms of what was most typical and how much ppls responses varied from average
Frequency distribution
Bar graph in which the possible scores on a variable are listed on x axis from lowest to highest and the total number of ppl who got each is plotted on y axis
Central tendency
The centre of batch scores
- mean or average of scores
- median: line up from lowest to highest and find middle
- mode: most common score
Variability
The extent to which the scores in a batch differ from one another is variablity
Standard deviation - calculates how much a bacth fo scores varies around its mean - finding distance between each indivdual score and then the mean and then computing the average of these distances
Effect size
descirbes magntidue of relationship bewteen manipulated or measured variables
correlation coeficant r = -1 to 1
As standard deviations of two groups get larger, the socres overalp more meaing there are more exception to the overall difference and tehre is smaller effect szie
d is standard deviation - 0.20 is small and 0.80 is large
Inferntial statistics
use sample results to infer what is true abt broader population
Statistcal signifance testing - when researchers test for statiscal sig they apply rules of logic and proability to estimate wheehr the result obatined came from particular pop
HARKing
hypothesizng after the rrsults are knwon - can mislead readers abt stength of evidendce
pHacking
Researchers cana anlyze a studys reulst in a wide vareity of ways they might remove scores, compused several score diff way or find p value of 0.5 (probability being null hypothesis)
Why is a study not replicable
false positive
small samples
HArking
pHacking
Underrporting nonsignifant effects
Intrsutcinal review board
panel of researchers, techers and other iztnes who deteermne whtehr a study upholds communities ethical standards
Based on
- Autonomy: ppl must give informed consent for partycipating
- beneficence: proposed researcg is evaluated on its rsiska dn beenifts to partcapnts and researchs outcome poetanl benefit to society
- Justice:research shouldnt be condcted disporpatantly on one segment of opulation
Three Rs for working with aniamls
- replacment: researchers should find altenrtaives to using animals inr eserch when possible
- refinment: researchee should modify expeirmenal procdueres an other aspects of animal care to minimize or elimnate animal distress
- reduction: adopt expeirmental designs that requrie fewest aniaml subjects