psyc 217 midterm 1 material

5 ways of knowing (without research)

1. intuition

2. authority

3. folk wisdom/ common sense

4. logic

5. individual experience

scientific scepticism vs organised scepticism

scientific: recognise that our own ideas may be wrong and question them

organised: must be critical of hypothesis even if it supports your ideas

epiricism

gain knowledge based on structured, systematic observations of the world

intuition 2 advantages and 2 disadvantages

advatage: quick and accesible + easy to generate research question

disadvantage: subject to personal judgement and hence prejudice+ possibility of illusory correlation

folk wisdom 2 advantages and 1 disadvantage

Advantages: generate research ideas + appealing to what one expects everyone else to know

disadvantage: confirmation bias

authority 2 advantages and 2 disadvatages

advantages: experts can be authories + minimise burden of acquiring our own knowledge

disadvantages: not all authorities are experts on all topics (blind trust) + authorities may have other interests which skew their judgement

logic 2 advantages and 2 disadvantages

advantages: consistent reasoning and descisions + easy to analyse and crtisue (premise + claims)

disadvantage: requires correct premise/ info + may have nothing to do with the real world

individual experience 3 advantages and 3 disadvantages

advantages: very memorable and others will listen to it; captures a real-world experience + generate research questions

disadvantages: not representative of all experiences + hard to evaluate predictive power for future/ diff situations + does not explain causality

5 goals of psychological science

1. Describe behaviour (trends)

2. predict behaviour 

3. determine causes of behaviour

4. Understand/ explain behaviour

5. Apply knowledge to solve problems

scientific method definition, 2 advatages and 1 disadvantage

definition: data driven way of knowing

advantages: common language + set of tools to guide and encourage critical thinking

disadvatges: not everything can be measured (indiv feelings + spirituality)

4 ways to design tractable, scientific research questions

1. question common assumptions

2. observe the world around us

3. solve practical problems

4. test theories

4 norms of scientific inquiry

1. universality: objective evaluation using acceptable methods (eg. studies yielding diff conclusions)

2. communality: methods and results shared openly (for replication and meta analysis)

3. disinteedness: should be motivated by honest and careful quest for truth

4. organised scepticism: research should be critiqued even it supports your views

basic vs applied research

basic: attempts to answer fundamental questions on human behaviour (correlational mainly)

applied: address practical problem

Theory and what makes a good theory (3 points)

overarching framework that organises and explains phenomena and data and generates hypotheses that test boundaries of the theory

good theory: parsimonious, supported by data, falsifiable (capaciity to be wrong)

when should we choose a less parsimonious theory

when the less parsimonious theory explains the observation substantially better (eg. warm mother in early life vs just warm mother)

how does theories help to generate new knowledge ( 2 ponits)

it points us in the direction to discover novel aspects of behaviour:

• can show weakness or strengthen previous theories which we could the build on and amend

• generate new hypothesis

hypothesis

tentative statement about a phenomenon that may or may not be true

prediction

specific statement regarding the expectaed outcome of a study

deduction vs induction

deduction: theory —> hypothesis —> prediction

induction: prediction —> hypothesis —> theory

lit review vs metaanalysis

lit review: summarise research

metaanalysis: statistical technique to reanalyse data based a a lot of past researcg

operational definition

observable, measurable indicator of each avriable for the purpose of a particular stuy

nominal scale

scale that does not rely on numerical or quantitative indicators; for qualitative analysis

ordinal scale

relies on numerative ranking but the distance between each rank is not constant

no meaningful zero (starts from 1)

interval scale

relies on numerical rankings and the distance between each rank is the same

no meaningful zero

ratio scale

relies on numerical ranking and the distance between each rank is the same

has a meaningful zero (zero means absence) and therefore ratio can be obtainsed

non-experimental vs experimental design

non-experimental: no variables are manipulated, just measure (eg correlational) and statistically determine the relatinship between the two variables (coeffecient r)

experimental: independent variables are manipulated and dependent variables are measured (can be used to determine causal relationship)

siuational variable vs response variable vs particpant variable

situational: the conditions external to the participants that are varied

response variable: response/ behaviour by the participant

participant variable: characteristics that individuals bring with them to the study

correlation matrix

table which shows correlation between >2 variables

2 limitations of correlational design

1. relying on R misses potential non-linear relationships

2. cannot determine causality

internal validity

ability to infer that one variable causes changes in another variable

2 tools to increase internal validity

1. experimental control

2. random assignment

3 criteria for claiming causality

1. covariation of cause and effect

2. temporal precedence

3. ruling out alternative explanations

covariation of cause and effect

the change in one variable is accompanied by change in the other variable

3 ways to operationalise dependent variable

1. self report

2. behavioural

3. physiological

third variable

a separate variable that impacts both variables of interest and the impact it has on X and Y is not measured in the study

for correlational/ non-experimental studies

confound + its effects on study

nuisance variable that covaries with the variable of interest and is often impossible to separate and it is not measured in the study

makes operational definition less valid and could explain the resulsy

mediator variable

a process that explains the relationship between the two variables and is measured during the research; the fact that this variable drives the relationship should be testable

random/ error varibility

phenomenon that we can’t predict or explain and results from variables that are not of interest

quasi experimental design

used to compare groups when true experiment is impossible (unethical to assign people into groups) but cannot support causal inferences

random assignment effects

makes groups equivalent and helps rule out alternative explanations that relate to particpant characteristics

works better for bigger groups

2 criterias for designing an appropriate comparison

1. needs to be as similar as possible to experimental conditon except for IV

2. small things like time of day, wording and staffing matters

placebo control

expectations are more equal across group hence any difference could be attributable to IV (which is often a drug)

demand characteristics vs experimenter expectancy

demand characteristics: features of the study that might inform particpants of the true purpose in a way that changes their performance

experimenter expectancy: experimenters knows the condition that participant is in and may unintentionally act to create the expected effect

hawthorne effect vs evaluation apprehension effect

hawthorne effect: people’s behaviour changes when they become aware that they are being watched

evaluation apprehension effect: unconsciously perform worse due to the feeling of uneasiness

5 ways to manage demand characteristics

1. use covert measures

2. use deception (must debrief at the end)

3. use distractor items to obscure inerests

4. order of events

5. single blind study (keep participants blind to hypothesis) —> via placebo maybe

how to manage experimenter expectancy (2 ways)

1. double blind (researchers and participants blind to the hypothesis)

2. computerised delivery to minimise contact between researcher and participant

interaction effect and its 2 subgroups

interaction effect: interactions between researcher and participants may affect the participant’s response

biosocial: characteristics of researchers (age/ race) can affect participants

psychosocial: attitude of the researcher can affect the participant

4 ways to operationalise DV well

1. avoiding ceiling and floor effects

2. attention and manipulation check

3. consider IV strength

4. consider DV sensitivity

ceiling effect vs floor effect

ceiling effect: everyone do too well, too little variation unable to see effect of IV

floor effect: everyone dont do well, too little variation unable to see effect of IV

strong vs weak IV

strong: likely to generate a large change in DV

weak: unlikely to generate large changes in DV

2 problems related to strong IV

1. Ethical

2. May not be relevant to the real world

high vs low sensitivoty DV

High sensitivity: DV that changes easily (mood/ attention span etc)

Low sensitivity: Resistant to change (core attitudes etc)

straightforward manipulation

manipulating independent variable using instructions/ oher stimulus materials in a simple way (no deception)

staged manipulation

operationalisation of independent variable that involve creating a complex situation which would elicit a state that would be elicited in a real situation

experimental realism vs mundane realism

experimental: extent to which experiences in the study are experienced by participants as impactful and engaging

mundane realism: extent to which experiences in a study resemble closely an experience in everyday life

attention checks (definition and function)

attention check: to check whether people understood the instructions or is answering the questions properly

ensure validity of the answers and ensure that participant are aware of the instructions (which may be manipulated)

manipulation check

confirm that manipulation has the intended effect on the participant

pilot studies and things to look out for

collect data from few participants

be sure that peaple in the pilot studies are not included in the actual study as they are not randomly assigned

researcher’s commitment

Researchers are to do what they promise when they collect data

open ended questions vs close ended question

open ended: flexbility in their answer, can come in multiple forms

close ended questions: a limited set of possible answers are provided (mcq, true/ false, rating scales)

open ended questions (2 pros 2 cons) + ecological validity

pros: freedom for participants to respond how they like + ecological validity (match between real world and what is measured)

cons: tough to quatify + coding responses is tough

close ended questions (3 pros 2 cons)

pros: easy to implement, restricts answers, easy to compare results between studies using the same methods

cons: restrict oarticipant responses (miss out info that might be relevant), designing questions tricky

considerations for rating scales

1. labelling alternative response

2. anchors

3. how many numbers (even numbers will skew vs odd numbers)

semantic differentiation scale

give two anchors on opposite ends and a series of dash lines for particpants to rate where they ar are

3 types of semantic differentiation scale

1. evaluation: how something is valued

2. ativity: active-passive, slow-fast, excitable-calm

3. potency: weak-strong, hard-soft

non-verbal scales

scales without words/ number —> for kids or those intellectually disabled

forced choice

give a fixed number of options and have to choose one even though it may not represent you

6 types of questions to avoid

1. double barrelled questions

2. loaded questions

3. negative wording

4. unnecessary complexity

5. yea-saying/ nay-saying

6. fence sitting

response bias

common pattern of inattentive responding (ie always say yes/no/ sitting in the middle

social desirability

responding in sociallu acceptable way rather than what they actually think

2 ways to deal with response bias

1. reverse coded questions

2. get rid of neutral alternatives (for fencing sitting)

why do question order matter

the response to the first question may impact response to subsequent question

how to ensure that question order does not affect reseults

1. insert filler questions

2. counterbalance (one group one order, other group do other order)

longitudinal/ panel study

administer questions to the dame people at 2 or more points in time

interview bias

intentional/ unintentional influence on respondent exerted by interviewer which might encourage certain responses consistent iwth interviewer’s expectation

focus group interviews

mainly qualitative and gather people with particulay knowledge

when do you measure DV for pretest-posttest design

before random assignment

4 pros of pretest and 1 con

1. allow researchers to see the changes in scores due to to the IV

2. show that groups are equivalent in the prettest hence more evidence to support the claim that the IV influences the dv

3. can help with matching design hence curb the problem posed by small sample size

4. can account for participant drop out —> selective attention

con: pretest may alert participants to the research hypothesis (may threaten internal validity)

matched pair design + pros

match participants on crucial characteristics and randomly assign one from each pair to control vs experimental group

can get rid of participant bias and porblems with small sample size

solomon 4 group design

when there is concern that prettest could create demand characteristics

pretest - experimental + control

no pretest - experimental + control

2 pros of within subject design

1. more statistically powerful as there is euivalent groups and better chanceat detecting small differences

2. fewer participants necesseary

repeated measure design vs concurrent method design

repeated: particpants exposed to all IV and DV recorded after exposre to each IV

concurrent: particpants exposed to all IV at the same time and comparative DV is measured at the end (behavioural/ attitudinal)

3 types of order effect

1. fatigued

2. practice

3. contrast: answer to first question affect the answer to subsequent questions

2 types of counterbalancing + definition of counterbalancing

counterbalancing: switching up the order of conditions across participants in a within- subjects design

complete: all possible orders

partial: latin square —> each condition appears directly after each other exactly once (for too many conditions)

3 restrictions of concurrent measure designs

1. restricts research question and DVs

2. potential for strong demand characteristics

3. hard to generalise