Research Methods

Empirical Evidence

Empirical Evidence

Data collected directly from observation or experimentation

Experiment

Controlled research study that tests if one variable influences or causes a change in another variable.

Quantitative

Numerical Data

Independent Variable

What is manipulated to assess effect of behaviour

Dependent Variable

What is measured to assess effects of IV

Control Variable

Kept constant

Lab Experiment

Done in highly controlled conditions, but does not have to be conducted in a lab

True Experiment

IV manipulated and DV measure under controlled conditions, randomly allocated.

Field Experiment

In a natural setting where there is less control

Quasi-experiment

IV is not manipulated and not randomly allocated. Participants are set apart by traits (age, gender etc)

Natural Experiment

Result of ‘naturally occurring event’ E.g. vaping behaviour before and after law changes

Operational Variable

How it is implemented, observed or measured.

Scientific Method

• Research Question

• Aim - limited to certain population

Hypothesis

Prediction how IV affects the DV (cause-and-effect)

Research Hypothesis

Relationship of IV and DV. A statement of the difference between 2 conditions “this will cause that to happen”

Null Hypothesis

Assumes there is no significant difference for a target population under 2 different conditions

Extraneous Variable

Has the potential to cause unwanted effect to DV (other than IV). Can affect results in an unwanted way. All variables must be controlled to manipulate the IV that causes changed to DV (NOT POSSIBLE TO CONTROL EVERYTHING)

Confounding Variable

Having a direct and systematic effect on DV. Occurring throughout the process and related to design of experiment. If CV exists, research becomes invalid and cannot have an conclusion.

EV and CV

All CV are EV but not all EV are CV. An EV only becomes a CB if it is not been controlled.

CV = EV

EV = CV

Implications

It is important to anticipate, minimize and/or control EV and CF during an experiment so that the researcher can be confident when observing the DV. The results will be more likely to be valid and reliable too.

Repeated Measures (minimizing EV and CV)

Each participant is in control and experimental group (repeating the study 2 times).

Advantages:

• Participant difference eliminated as it is the same participants

• Less participants needed

Disadvantages:

• Participants may do better or worse in second condition ‘order effect’

• Demand characteristics (work out purpose of the study)

• May drop out

• May need to spread out two conditions (time consuming)

Counterbalancing

• Manipulating the order participants are exposed to experimental condition

• Each group receives different treatments in different orders

• Results are combined so order effect is controlled

Matched-pair designs

Pairs that are similar in key characteristics that can influence IV, random, one to each condition. E.g. twins.

Advantages:

• Very similar key characteristics

• Twins

• Attrition less common

• Don’t need to spread out time between two conditions

• No order effects

Disadvantages:

• Characteristics don’t reflect all human characteristics as groups will differ

• Pre-testing is time consuming

• Attrition, loss of two sets of data

Independent Sample Design

Participants are randomly allocated to control or experimental groups that have an equal chance of being assigned. Completing only one condition.

Advantages:

• No order effects

• Attrition less common (commitment not as significant as repeated designs)

• Don’t need to spread out

Disadvantages:

• Less control over characteristics, participant difference mat affect results

Ethics

Standards that act as a guide to identify good, desirable and acceptable conduct. Which actions are appropriate and which are not.

DUDCAR

D - deception

U - Undue stress

D - debriefing

C - consent

A - Anonymity

R - Right to withdraw

Deception

Providing information may influence how participants think, feel and behave. Deception may be employed if it is justified, no harm and stress done and participants are debriefed after.

Undue Stress

Participants must be protected from all forms of mental and physical harm and must not suffer negative consequences. Must be extra careful if there are vulnerable groups

Debriefing

Participants are informed of true aim after experiment. If deception is used, that must be explained. Must provided opportunities for participant to access information about the study and refer them to relevant assistance if needed.

Consent

Participants must be voluntary and fully informed. Prior to study, they must be made aware of nature and purpose, method, withdrawal and confidentiality, any risks/discomforts they may encounter. Legal guardians have to give consent for children/people unable and no participants can be psychologically or physically vulnerable.

Anonymity

Participants are not identified or identifiable without consent. It must be explained prior to study that details of involvement cannot be revealed or connected to them. Confidentiality refers to data and results being stored and disposed of appropriately.

Right to withdrawal

They have the right to leave at anytime with no reason and no negative consequences. Right to remove results at any time.

Sources of Error

Errors that can make a research invalid or unreliable

Individual Participant Variables

• Age

• Mood

• IQ

• Culture

• Education

Differing backgrounds, characteristics and abilities that may impact their response in an experiment

Minimizing:

• Appropriate experimental design

• Random/stratified sampling

• Random allocation of participants to control or experimental

Order Effects

Performance measured by DV and influenced by specific order of tasks. Treatments or conditions are presented rather than IV.

Two types:

1. Practice Effect - prior experience and repeating the task = better or worse/bored

2. Carry-over effects - influence in task performance by e.g. drinking

Minimizing:

• Counterbalancing - systematically changing order of conditions so each conditions occurs equally as oftion in each position. I.e. half participants do experiment and half do reverse

Demand Characteristics

Participants interpretation of aim or purpose causes them to change their behaviour consciously of unconsciously to fit interpretation.

Minimizing:

• Single-blind procedure - participants do not know which group they are in

Experimenter Effect

1. Experimenter expectancy - provides cues or hints about responses participants should make

2. Experimenter bias - unintentional bias in collection and treatment of data

Minimizing:

• Double-blind procedure - the experimenter and participants don’t know which group they are/observing

Non-standardized instructions and procedures

Are not uniform, or the same for all participants.

Standardised:

• Selection go participants

• Instructions for participants in different groups

• Interactions with participants

• Use materials or apparatus

Minimising:

• Use standardised - predetermined and identical. It is simple clear and contains no ambiguity or variations

Placebo Effect

An inactive substance or fake treatment which subs for the real thing/treatment. A change in behaviour or response of participants due to their expectations and response of participants that they are receiving some treatment.

Minimising:

• Single-blind

• double-blind

• Provide placebo to control groups as well as experimental group to ensure it impacts groups equally and cancels out

Population

“Specific group of people whom the researcher is interested in studying”

Sample

A subset of population who participate in the study

Sampling

The process of selecting participants for research

A good sample

• Representative sample = good sample

• Larger = better

• Representative of everyone in population (bias/error)

• Characteristics are taken into account

• Characteristics represented are in the same proportion of population

Random Sampling (Representative)

• Employs a systematic method of selecting

• Every member of population has an equal chance of being part

E.g.

• Lottery

• Table of random #: allocate member w/ a number

Advantages

• Results can be generalised

• If sufficient size, may be ‘rep’ of population

Limitations

• Large sample

• Almost impossible to be truly random (refuse to participate)

• Time consuming to ensure everyone has a equal chance of being selected

Stratified (rep)

1. dividing into subgroups based on shared characteristics

2. Randomly selected from each subgroup in same proportion that exist in population

Advantages

• Control rep of some key chara

• Useful of large isn’t possible

  • Clear chara need to be controlled

Limitations

• Time consuming (req pre-knowledge of participant chara)

Convenience

• readily available w/out attempt to make sample rep

• Based on accessibility

Advantages

• quick + easy

• Considered adequate when investigated mental processes in behaviour

  • assumed similar in ‘normal’ ppl

• Widely used, pilot to gain prelim indication of responses before

Limitation

• Biased Sample

  • time + location

• Data may be misleading and results cannot be generalised

• low external validity

Volunteer

• Recruiting people who ‘self-select’ themselves

• often done through ads + newspaper

Advantages

• quick and easy

• Participants motivated to take part

Limitations

• Sample not rep of population, limits ability to generalize

Snowball

• People asked to provide others

  • family + friends

• Used when it is difficult to access research participants

Allocation

• After sampling

• Systematic

• personal chara that affect results to be controlled

Experimental Group

• Expose to experimental condition, IV investigated

Control Group

• Exposed to control cond, IV is absent

Occur?

• Randomly

• Participants likely to be in one group or another

  • Every participant has an equal chance of being selected for any group

Random Allocation

Purpose

• equivalent groups before intro IV so effect can be estimated

• W/ a sufficiently large sample, reasonable to assume each group contains equal # participants w/ personal chara that affect result

Reliability

How consistent the results

E.g. blood alcohol levels or experiment done twice with similar results

Not expect to be identical (reliability)

• Psych experiments are complex

  • Diff chara

• Similar results over time

• If results can be replicated = high reliability

Internal Reliability

How consistent a method measures within itself

• Ruler w/variable cm, LACK internal validity

• IQ test composed only of easy or hard questions

External reliability

How consistent a method measures overtime when repeated

• Should give similar results when repeated on same ppl under same circumstances

  • Same method w/diff results = LACK external reliability

• Assess using test-retest measure to assess correlation in results to determine validity

Validity

Results accurately measure what they aim to measure

• IQ tests intelligence not memory or language

• Not all psych chara can be measured directly, other indirect measures are used to reference inferences about trait

Construct Validity

Does test/method support underlying theory? To what extent do the operationalization’s reflect the construct?

Internal Validity

Accurate methodology and procedures used? Is study testing hypothesis? Cause-and-effect relationship? Sample a rep of the population?

External Validity

Findings apply to other people, setting, situations and time periods? Most discusses types are:

• Ecological - environment in study affect results, less rep of behaviours of uncontrolled environment?

• Population - sample biased? How affect results?

Measurements can be reliable w/out being valid

• BUT measurements cannot be valid if they are not reliable