RM1 W2: Measurements

Step 3:

Operationalise your variable (determine how you will define and measure your variable)

Constructs

Constructed variables that cannot be seen but can manifest through observable behaviour eg. love, happiness, intelligence

Operationalisation

A procedure for indirectly measuring a construct through the indication of its existence by other phenomena

i.e. converting an abstract variable (construct) into a concrete entity (operational definition) that can be observed and studied

Eg. Intelligence = construct, IQ test = operational definition

How do we evaluate the quality of our measurement procedure?

Using validity and reliability 

Validity

The degree to which the measurement process measures the variable that it claims to measure

6 types of validities: (FCCCCD)

1. Face validity

2. Content validity

3. Criterion validity (concurrent & predictive)

4. Construct validity

5. Convergent validity

6. Divergent validity 

Face validity 

Whether it LOOKS/SEEMS like it measures the variable it claims to measure 

When is it useful?

When you are trying to convince others to use the measure (people are more accepting of a measure when it has high FV)

Limitations

1. Subjective judgment, least useful type of validity

2. Could be wrong (eg. AI face recognition seems to analyse facial features deeply, but is a bad indicator of emotions)

Content validity

Whether it covers the entire scope/domain of a construct

Criterion validity

Whether the construct can be accurately manifested into a behaviour

2 types of criterion validity:

1. Concurrent validity

2. Predictive validity

Concurrent validity

Whether a measure can predict current behaviour

Predictive validity

Whether a measure can predict future behaviour

Construct validity

Whether a measure testing for a particular construct correlates with another measure testing for the same construct, and not correlate with a measure testing for a different construct

When is it useful?

When a construct has no well-established criterion

2 sub-categories of construct validities:

1. Convergent validity

2. Divergent validity

Convergent validity

Whether results on similar tests are correlated; have positive correlation

When is it useful?

Used to seek confirmation that a test if really measuring what it claims to measure

Divergent validity

Whether results on a test of an unrelated construct have no correlation with your test; weak correlation

4 Scales of measurements: (NOIR)

3 Modalities/Ways of measurements: (SPB)

1. Self-report measures

2. Physiological measures

3. Behavioural measures

Self-report measures

(+)

• easy, direct and convenient

(-):

• easy for participants to distort

Physiological measures

(+)

• very objective

• use of equipment provides accurate, reliable, and well-defined measurements that are not dependent on subjective interpretations by researchers of participants

(-):

• the presence of monitoring devices can create an unnatural situation for participants

• tend to be low in criterion validity (no evidence it can predict current/future behaviour)

• and low in construct validity (a physiological response like heart rate doesn't solely represent one psychological construct, can be influenced by many other factors)

Behavioural measures

(+)

• provides researchers with vast options in selecting the behaviour that best defines and measures the constructs  

(-):

• behaviours can be influenced by temporal situations —> best to measure a cluster of related behaviours rather than rely on a single one