Samenvatting testgebruik
Contents week 1
Chapter 1: Psychometrics and the importance of psychological
measurement
Psychological measurement
Examples:
- Clinical context: depression test has an effect on the treatment plan
- Educational context: cognitive ability test determines if the student gets an internship
- Political context: result of mathematics test determines if more money will be spent on
mathematics education
- Scientific context: direct observation tests have an effect on the outcomes of the study
Important to get familiar with the principles of psychological measurement:
- To evaluate the quality of psychological tests
- To draw accurate conclusions based on psychological test results
Unobservable psychological attributes
Psychological attributes
- We cannot directly observe psychological attributes like anxiety, depression and
mathematical ability (latent variables), so they are unobservable
- We measure them by measuring some type of observable behavior that we think represents
that attribute
Three important assumptions:
1) The observable behavior reflects the unobservable psychological attribute
2) The task is theoretically linked to the unobservable psychological attribute
3) The unobservable psychological attribute exists
Psychological tests
Psychological test: a systematic procedure for comparing behavior between or within persons
- 3 important elements: behavior, systematic procedure, comparing
Can take many different forms:
- Paper-and-pencil test
- Laboratory setting
- Can differ in:
- Content, type of response, method of administration, purpose
- Criterion referenced tests: if someone meets a prespecified criterion
- Example: exam at university
- Norm referenced tests: how someone compares to others using a reference sample
- Example: WISC
- Speeded tests: answer as many questions correctly as possible in a fixed time
- Power tests: allowed to use all the time that is required to finish the test
Psychometrics
Psychometrics: the science concerned with evaluating the attributes of psychological tests
- The type, the reliability and the validity of the scores
Chapter 2: Scaling
,Scaling
Scaling: the way in which numerical values are assigned to psychological attributes (such as
numerical score on an IQ test)
- Can heavily influence the way in which test scores should be interpreted
Three properties:
1) The property of identity: the numbers provide information about whether or not a feature of
an attribute is present (yes or no)
2) The property of order: the numbers provide information about the relative amount of the
attribute (who has more)
3) The property of quantity: the numbers provide information about the amount of differences
between people with respect to the attribute (how much more)
The number 0 can have different meanings:
- Absolute zero: 0 reflects the non-existence of an attribute (such as 0 grams)
- Arbitrary zero: 0 reflects an arbitrary quantity (0 degrees)
Units of measurement
Units of measurement: definite amounts that are used as a standard for measurement
- Example: centimeters, meters, kilometers; IQ points; reaction time
Can be arbitrary in 2 ways:
1) The size of the unit: a gram is arbitrary in size
2) The link to the type of object: a meter is not tied to one single object
- Psychological units are usually only arbitrary in the first way
Scales of measurement
Nominal scale of measurement:
- Numbers provide information about whether or not a feature of an attribute is present or not
- Numbers solely reflect a specific category and do not have any mathematical value
Ordinal scale of measurement:
- Numbers provide information about the relative amount of the attribute
- Numbers can be used to rank people according to the amount of the attribute
Interval scale of measurement:
- Numbers provide information about the precise amount of differences between people
- Does not have an absolute zero
- Numbers are additive but not multiplicative
- Example: temperature
Ratio scale of measurement:
- Numbers provide information about the precise amount of differences between people
- Does have an absolute zero
- Numbers are additive and multiplicative
- Example: weight
- Most scores in the social sciences are treated as interval level → there is no absolute zero
because a score of zero does not represent the absence of an attribute
, - Example: every person has some level of social skills, a score of 0 should not indicate
the absence of social skills
Chapter 3: Individual differences and correlations
Variability
Variability: the degree to which scores on a psychological test differ from person to person, or from
time to time
- Variable: an attribute, characteristic or quantity of which the value can vary from person to
person (example: age, test scores)
Central tendency: the most representative value of a set of scores (most common: the mean)
Variability: provides information about the size of differences among a set of test scores, most
common:
- Variance (the average squared deviation from the mean) and standard deviation (the average
deviation from the mean)
Covariability
Covariability: the extent to which the scores on two variables vary in a corresponding way
Two types of information:
- Direction of association: positive or negative
- Magnitude of association: strong (consistent) or weak (inconsistent)
Statistics:
- Covariance: provides clear information about the direction of the association
- Does not provide clear information about the magnitude of the association
- Correlation: provides clear information about the direction of the association, does also
provide clear information about the magnitude of the association (does not depend on metric)
- -1 → 1
The variance and covariance of composite scores
Composite scores: scores computed by summing or averaging a set of responses made by an
individual
- Variance of a composite score depends on:
- Variance in the item scores
- Correlation between the item scores
-
- To maximize composite score variance:
- Add more items
- Use highly-correlated items
The mean and variance of binary items
Binary or dichotomous items: items for which only two outcomes are possibles
- Coding:
- 0: negatively valenced answer
- 1: positively valenced answer
- Mean of binary items: p
- Variance of binary items: pq (positive x negative answers)
- Largest possible variance: p= 0.5, q= 0.5, variance = 0.25
Contents week 1
Chapter 1: Psychometrics and the importance of psychological
measurement
Psychological measurement
Examples:
- Clinical context: depression test has an effect on the treatment plan
- Educational context: cognitive ability test determines if the student gets an internship
- Political context: result of mathematics test determines if more money will be spent on
mathematics education
- Scientific context: direct observation tests have an effect on the outcomes of the study
Important to get familiar with the principles of psychological measurement:
- To evaluate the quality of psychological tests
- To draw accurate conclusions based on psychological test results
Unobservable psychological attributes
Psychological attributes
- We cannot directly observe psychological attributes like anxiety, depression and
mathematical ability (latent variables), so they are unobservable
- We measure them by measuring some type of observable behavior that we think represents
that attribute
Three important assumptions:
1) The observable behavior reflects the unobservable psychological attribute
2) The task is theoretically linked to the unobservable psychological attribute
3) The unobservable psychological attribute exists
Psychological tests
Psychological test: a systematic procedure for comparing behavior between or within persons
- 3 important elements: behavior, systematic procedure, comparing
Can take many different forms:
- Paper-and-pencil test
- Laboratory setting
- Can differ in:
- Content, type of response, method of administration, purpose
- Criterion referenced tests: if someone meets a prespecified criterion
- Example: exam at university
- Norm referenced tests: how someone compares to others using a reference sample
- Example: WISC
- Speeded tests: answer as many questions correctly as possible in a fixed time
- Power tests: allowed to use all the time that is required to finish the test
Psychometrics
Psychometrics: the science concerned with evaluating the attributes of psychological tests
- The type, the reliability and the validity of the scores
Chapter 2: Scaling
,Scaling
Scaling: the way in which numerical values are assigned to psychological attributes (such as
numerical score on an IQ test)
- Can heavily influence the way in which test scores should be interpreted
Three properties:
1) The property of identity: the numbers provide information about whether or not a feature of
an attribute is present (yes or no)
2) The property of order: the numbers provide information about the relative amount of the
attribute (who has more)
3) The property of quantity: the numbers provide information about the amount of differences
between people with respect to the attribute (how much more)
The number 0 can have different meanings:
- Absolute zero: 0 reflects the non-existence of an attribute (such as 0 grams)
- Arbitrary zero: 0 reflects an arbitrary quantity (0 degrees)
Units of measurement
Units of measurement: definite amounts that are used as a standard for measurement
- Example: centimeters, meters, kilometers; IQ points; reaction time
Can be arbitrary in 2 ways:
1) The size of the unit: a gram is arbitrary in size
2) The link to the type of object: a meter is not tied to one single object
- Psychological units are usually only arbitrary in the first way
Scales of measurement
Nominal scale of measurement:
- Numbers provide information about whether or not a feature of an attribute is present or not
- Numbers solely reflect a specific category and do not have any mathematical value
Ordinal scale of measurement:
- Numbers provide information about the relative amount of the attribute
- Numbers can be used to rank people according to the amount of the attribute
Interval scale of measurement:
- Numbers provide information about the precise amount of differences between people
- Does not have an absolute zero
- Numbers are additive but not multiplicative
- Example: temperature
Ratio scale of measurement:
- Numbers provide information about the precise amount of differences between people
- Does have an absolute zero
- Numbers are additive and multiplicative
- Example: weight
- Most scores in the social sciences are treated as interval level → there is no absolute zero
because a score of zero does not represent the absence of an attribute
, - Example: every person has some level of social skills, a score of 0 should not indicate
the absence of social skills
Chapter 3: Individual differences and correlations
Variability
Variability: the degree to which scores on a psychological test differ from person to person, or from
time to time
- Variable: an attribute, characteristic or quantity of which the value can vary from person to
person (example: age, test scores)
Central tendency: the most representative value of a set of scores (most common: the mean)
Variability: provides information about the size of differences among a set of test scores, most
common:
- Variance (the average squared deviation from the mean) and standard deviation (the average
deviation from the mean)
Covariability
Covariability: the extent to which the scores on two variables vary in a corresponding way
Two types of information:
- Direction of association: positive or negative
- Magnitude of association: strong (consistent) or weak (inconsistent)
Statistics:
- Covariance: provides clear information about the direction of the association
- Does not provide clear information about the magnitude of the association
- Correlation: provides clear information about the direction of the association, does also
provide clear information about the magnitude of the association (does not depend on metric)
- -1 → 1
The variance and covariance of composite scores
Composite scores: scores computed by summing or averaging a set of responses made by an
individual
- Variance of a composite score depends on:
- Variance in the item scores
- Correlation between the item scores
-
- To maximize composite score variance:
- Add more items
- Use highly-correlated items
The mean and variance of binary items
Binary or dichotomous items: items for which only two outcomes are possibles
- Coding:
- 0: negatively valenced answer
- 1: positively valenced answer
- Mean of binary items: p
- Variance of binary items: pq (positive x negative answers)
- Largest possible variance: p= 0.5, q= 0.5, variance = 0.25
