Chapter 3: Research Design

research design

collecting + analysing data

criteria to evaluate research

Research method

technique of collecting data

executed by research design

Variable

attribute where cases vary

Quality criteria - quantitive research

• reliability

• replication

• validity

  • measurement v

  • internal v

  • external v

  • ecological v

  • inferencial v

Reliability

same results of study if repeated
are measurements consistent?

replication

need for specific recording

is the procedure explained clearly enough?

validity

integrity of conclusion generated

how accurate are the findings?

measurement validity

if measure reflects the concept

Are you measuring what you think you are measuring?

internal validity

causality

How unambiguous is A the cause of B?

external validity

generalised

representative sample

ecological validity

applicable to everyday natural settings

are findings meaningful to everyday lives?

internal validity

warranty that the results are not bc of outside factors or methodology

parallels in quan research

internal - credibility

external - transferability

reliability - dependability

objectivity - confirmability

Experimental design: features

• trust worthy - high internal validity

• manipulate independent variable

• participants likely to be allocated into 1 or more exp groups

• how far differences between the groups are responsible for variations in the level of the dependent variable

threats to internal validity in experimental design

• history

• testing

• instrumentation

• morality

• maturation

• selection

• directional ambiguity

Experimental design - internal validity

able to control for possibility of dif explanation of findings

= high

face validity

assessment of how well an indicator reflects the content of the concept in question

Experimental research - threats to external validity

• interaction of selection + treatmenr

• interaction of setting + treatment

• interaction of history + treatment

• effects of pre-testing

• reactivity

laboratory exp

• high inf over arrangment = high internal v

• low eco validity

quazi exp

lack of random assignment

cross sectional design - features

sample of cases and at a single point in time

collects data (simultaneously) connecting with +2 variables - detects pattern of association

non-manipulable variables as reason to use cross sectional

socio - many unmanipulable (ethically)

relationship > causality

Longitudinal design

participating more than once

high cost + time = rare

strong casual inference

Panel study

type of longitudinal

often randomly selected + national

• people, households, organizations, schools

Cohort study

type of longitudinal

entire cohort / random sample

Case study design

detailed + singe case

• person fam org

often both quan + qual

types of case study design

critical

extreme / unique

representative / typical

revelatory - when observer has opportunity to obs a phenomenon previously inaccessible

longitudinal

Comparative design - features

2 contrasting cases + similar method

quan / qual

issues: high cost, comparability, sample, translations