1A - elements of researching behaviour

hypothesis

a tentative statement or prediction that proposes an explanation for some psychological phenomenon

a hypothesis must be both….

testable and falsifiable

alternative hypothesis (H1)

predicts an effect

(a manipulated variable is expected to cause some change in a measured variable)

null hypothesis (H0)

predicts no effect

(manipulated variable causes no change in the measured variable - ie, any change is due to chance)

*NOT the opposite of the alternative hypothesis

one-tailed hypothesis

predicts effect in one direction (one of either positive or negative)

two-tailed hypothesis

predicts an effect in either direction (can cause both positive and negative)

significance testing

one of H1 or H0 is accepted and the other rejected, depending on statistical analysis results that determine if results are statistically significant or occurred by chance

hypothesis testing

H1 and H0 are compared using statistical error rates and collected data, supporting one hypothesis over the other

reification

defining an abstract concept as a real/concrete variable. *CANNOT DO

quantitative variables

a type of variable that is a numerically measurable quantity

categorical variables

a type of variable where qualities can be used to assign categories to individuals/groups

operationalised variables

variables that have been expressed specifically in terms of how they will be measured

variables that are categorical / not already expressed in quantitative terms can still be operationalised by…..

quantifying them on a scale

independent variable

the factor or variable being manipulated by the researcher

aka the CAUSE

dependent variable

the factor or variable being measured by the researcher

aka the EFFECT

the two key components of a psychological theory

1. must describe a behaviour

2. makes a prediction about future behaviours

TEACUP acronym for evaluating theories

Testable - has to also be falsifiable so that it can be tested

Empirical support - supporting evidence that is reliable and can be replicated

Application - can be applied to various different situations or improves a specific behaviour

Clearly defined variables - can be reliably measured

Unbiased - not androcentric (gender bias) or ethnocentric (culture bias)

Predicts behaviour - has a high predictive validity

two types of sampling methods

probability and non-probability

types of probability sampling

simple random , stratified , systematic

types of non-probability sampling

convenience / opportunity , purposive , snowball

what does the incorrect sampling of participants affect?

the validity of a study

target population

the wider group of people to whom a study’s results can be generalised

typically include a set of traits/characteristics that are relevant to the research

sampling frame

the accessible portion of the target population that researchers can identify

sample

those from the population who actually participate in a study

random selection

every member of a population has an equal chance of being chosen for participation

*only possible when a researcher has access to an ENTIRE population, which is why it is rare that psychological research draws a truly random sample

the only way to obtain a properly representative sample is through….

probability sampling methods using random selection

how to obtain a simple random sample

1. organise the sampling frame to identify every member of the accessible population

2. use a statistically random method to select the sample (eg: random number generator)

what is the challenge for simple random sampling?

it is often not a viable technique for many psychological studies as EVERY member of a target population may not always be identifiable / accessible

how to obtain a sample using stratified sampling

1. identify strata (key subgroups within a target population)

2. a simple random sample is taken from each strata to make the overall sample

*effective for small populations with known characteristics

benefit of stratified sampling

allows for a more representative sample as all important subgroups are included

disadvantages of stratified random samples

prone to researcher bias due to how a researcher may define strata

if members of a population belong to 2+ strata, stratified sampling becomes ineffective

what is a proportionate sample drawn using stratified sampling?

when the sample group contains a number of participants from each subgroup that is relative to their fraction of the overall population

what is a disproportionate sample drawn using stratified sampling?

when certain strata are weighed more heavily than their fraction of the population, resulting in overrepresentation of some subgroups

how to obtain a systematic random sample

1. randomly number members in the population

2. determine sample size

3. calculate k using the following formula

   k = N / n , where N = population size / n = sample size / k = interval value

4. draw a random number between zero and k

5. from the drawn number, every kth participant is chosen for the sample

disadvantages of systematic sampling

every kth member may share similar characteristics, so the sample may not be representative

the technique is based on the assumption that population sources are randomly organised to begin with

can data collected using non probability sampling methods be analysed using inferential statistics?

no !!!

convenience / opportunity sampling

sampling frames that are readily available , nearby , and willing to participate

**sample is selected by researchers

disadvantage of convenience sampling

technique is biased, so samples drawn are not statistically generalisable to the target population

*results may still be transferable (?)

volunteer sampling

sample obtained/chosen by people who choose to participate in the study themselves (typically through advertisements)

volunteeers accepted until researcher has adequate participants

disadvantage of volunteer sampling

biased technique, not representative of target population

purposive sampling

participants that are chosen for a sample according to predetermined criteria that is relevant to the research

*sample sizes are determined by theoretical saturation ; the point at which additional / new data will not provide additional insight into the research question

disadvantage of purposive sampling

tends to overrepresent sub groups in the population

when is purposive sampling most effective?

when data review and analysis are completed as data is being collected

snowball sampling

where participants already involved in a study are asked to recruit / recommend others who meet the selection criteria

*type of purposive sampling

**typically used when target population is rare / hard to find / hidden

disadvantages of snowball sampling

anonymity of participants is often lost

unlikely to yield a representative sample

the two types of groups in a research design

1. experimental group (one or more - group subject to manipulation of the IV)

2. control group (typically exposed to a placebo instead of the IV to gather baseline data)

independent measures (between-subject design)

two separate groups of participants , where each participant is tested on only one condition of the experiment

demand characteristics

participants of a study figuring out the aim/hypothesis, and changing their behaviour to deliberately please/confound it

participant variables

differing personal characteristics of individuals across different groups that may affect the data

*a type of EV

strengths of independent measures

• prevents order effects

• prevents demand characteristics

• participants are more likely to complete the study ; less risk of losing data

• more time efficient ; experimental and control groups can be researched simultaneously

weaknesses of independent measures

• affected by participant variables

• greater number of participants required

random assignment conditions

1. each participant has an equal chance of being assigned to each experimental condition

2. each participant’s assignment to either condition is independent of other participants

*mitigates researcher bias affecting allocation

repeated measures (within subject design)

each participant is exposed to all experimental conditions ; each participant yields at least two pieces of data

order effects

the effect of participation in one condition of the experiment then effects participant behaviour in a later condition

*when not controlled, these become CVs

types of order effects

1. practise / learning — participants perform better in later conditions bc they have learned the task in earlier conditions

2. fatigue / boredom — participants perform worse in later conditions bc they are tired/bored from the earlier ones

3. context — testing in earlier conditions may influence how participants interpret tasks in later conditions

*when not controlled, these become CVs

counterbalancing

when a sample is split in half, with one half completing conditions in the order A-B and the other half in the order BA

a more complex version is having all participants complete the conditions in the order A-B-B-A, and taking the mean score across both trials

*counterbalancing mitigates order effects’ impact on results by applying them to both sets of data

ways to mitigate order effects

• counter balancing

• including a time lag between conditions (so that order effects can wear off)

• randomising stimulus material

• randomising the order in which conditions are presented to each participant

strengths of repeated measures

• requires fewer participants

• reduces impact of participant variables as they are the same for all groups

weaknesses of repeated measures

• subject to the impact of order effects

• subject to the impact of demand characteristics

• requires more time from each participant

matched pair design

participants sampled are similar in a range of specific characteristics relevant to the study, and each partner is randomly assigned to the control/experimental group

strengths of matched pair designs

• controls impact of participant variables

• avoids order effects

weaknesses of matched pair designs

• highly time consuming

• only identical twins allow for exactly matched pairs

• high risk of lost data ; if one participant drops out, the partner’s data also becomes unusable

standardisation

the elimination or control over an variable or factor other than the IV that may affect the results of a study

elements of a standardised procedure

1. standardised environment (location)

2. standardised instructions

3. standardised materials and tests (researchers also trained in administration and scoring)

4. standardised procedure (researcher-participant interaction)

single-blind control

when the participant is unaware of whether they are in the control or experimental group

*reduces influence of participant expectancies

double-blind control

when the participant’s allocation to the experimental/control group is hidden from both researchers AND the participant

*reduces influence of participant expectancies

ethical considerations in psychological research

• informed consent

• protection of participants

• right to withdrawal

• confidentiality and anonymity

• deception

• debriefing

ethical considerations in reporting results

• confidentiality and anonymity

• stigmatisation

• scholarly integrity