WEEK 1 - STATISTICAL ANALYSIS & DESIGN

- differentiate between conceptual and operational constructs - apply operationalism to concrete examples - compare and contrast experimental, quasi-experimental and non-experimental research designs

the research process (6)

1. finding a research topic and buillding rationale

2. generating research question

3. develop hypothesis

4. design research study

5. analyse the data

6. drawing conclusions and reporting results

what do we need to do when designing a research study (3)

1. identiify and define variables

2. research design (experimental vs quasi-experimental vs non-experimenta;l)

3. identify the population (sampling)

what is a conceptualising construct

something hypothetical / abstract concept e.g. motivation, aggressiveness, depression etc

what is operationalisation (3)

1. turning a construct into an operational variable

2. something we can objectively measure

3. helps others to replicate our study / findings

visualisation of construct to measure

construct —-operationalisation—→ measure

when we operationalise a construct, we are creating measured variables

an example of operationalisation

happiness ——operationalisation—→ physiological (hormones, neurotransmitters), behavioural, other-report (family/friends), self-report

when do we do operationalisation

beginning of each project

what does operationalisation require

a decision on precise definitions to be able to quantify - subjective 

what can different operationalisations lead to

different conclusions e.g. if RQ whther background music affects task performance

example of how operationalisation can be subjective (3)

• operationalise task performance as accuracy - positive

• operationalise task performance as speed - negative

• operationalise bg music - type/volume

what are quasi-experiments

provide some evidence for cause-and-effect

what are conditions of quasi-experiments (3)

1. manip of independant variable

2. no comparison/control group or not random assignment to conditions/groups

3. little to no control over extraneous variables

what are experiments

explain cause-and-effect relationships

conditions of experiments (3)

1. manip of independant variable

2. random assignment to conditions/groups

3. control over extraneous variables

what do true experiements include

control groups

what are non-experiments

establish association between variables

conditions of non-experiments (3)

1. no manip of independant variable

2. no random assignment to conditions/groups

3. little to no control to over extraneous variables

what are within-subkects design

re[peated measures design

• all ppts are exposed to all levels of independant variable

whatis between-subject design

independant designs

• different ppts are exposed to different levels of independant variable

positives of within-subject (3)

• max control of extraneous ppt variables

• more powerful as noise in data is reduced

• smaller N

negatives of within-subject (2)

• not practical with all designs

• carryover effects (background factors, repeating the same thing so they know what to expect) → solution is counterbalancing

positive of between-subject (4)

• no carryover effects

• independant scores

• less time per ppt

• practical with designs where it is not possible for ppts to be in more than one condition

negative of between-subject

needs larger N

what is complete counterbalancing

equal number of ppts complete each possible order of conditions e.g. 3 note-taking conditions (laptops, tablet, pen and paper)

what is the problem with counterbalancing

might not be possible with large number of conditions (view table)

what is the solution to counterbalancing

latin square

what is the latin square

each condition appears exactly once in each position

what is the goal of latin squares

ensure that each condition has an equal chance of being experienced first, second, last or by different ppts, this is efficient

what are correlational designs for non-experimental designs

associations/ relationships between variables

correlational design =/ statistical correlation

can include continuous and/or categorical variables

how can we study change over time (3)

1. cross-sectional

2. longitudinal

3. cross-sequential

what is cross-sectional

data collected at one point in time that compares two or more pre-existing groups of people

what is longitudinal

one group of ppl are followed oevr time as they age

what is cross-sequential

combo of longitudinal and cross-sectional elements with different cohorts observed across time

positive of cross-sectional (2)

1. efficient for large samples

2. time and vost effective

negative of cross-sectional (3)

1. causality

2. vulnerable to cohort effects

3. misses dynamics over time

what does cross-sequential monitor

individuals of different ages across a short period of time

3 types of quasi-experimental research

1. one group post-test-only design

2. one group pretest-posttest design

3. non-equivalent pretest-posttest design

why is one-group post-test-only design the weakest form of quasi-experimental design (2)

• no baseline

• no comparison group

positive of one-group posttest-only design (2)

• quick to implement

• good as pilot study/initial exploration

what is one-group pretest-posttest design

dependent variable is measured before and after treatment

negatievs of one-group pretest-postest design (3)

• practice effects

• potential history effects

• regression to the mean

positives of one-group pretest- post-test design (2)

• quick to implement

• each [erspn os theit own baseline

what is non-equivalent pretest-posttest design

a quasi-experimental design that compares two existing, intact groups that are not formed through random assignment

positives of non-equivalent pretest-posttest design (5)

1. more info than one-group designs

2. pretest allows baseline comparison

3. practical and feasible

4. ethically easier

5. some causal evidence

negatives of non-equivalent pretest-posttest design (3)

1. no random assignment possibke

2. baseline differences

3. regression to the mean