SFL 290 MIDTERM 2 VERIFIED STUDY GUIDE
Reliability is... - Answers - CONSISTENCY
- you do the exact same thing every time
ex: height, weight, getting all of the darts in the same spot
Types of Reliability - Answers - 1) Test-retest
2) Parallel (Alternate)-forms
3)Representative
4)Inter-rater
5) Internal (Inter-item) Consistency
Test-restest Reliability - Answers - When the measure is given on 2 separate occasions,
is there consistency between scores?Repeat the experiment and get same results.
- questionnaires, observations
- same set of people both times
ex: IQ test, first midterm get a 45/50, should get the same score or around it the second
time that I take it
ex: depression test
ex: running a mile - getting same time
statistical test: correlation coefficient
Parallel (Alternate)-forms Reliability - Answers - Is there consistency between forms of
test?
ex: 5 different forms of ACT (if a student took any test they would get around the same
score), spencer writes up a new midterm and half receive one and other half receives
other
statistical test: correlation coefficient
Representative Reliability - Answers - Is the measure consistent across subpopulations
or groups of people?
- applies across age groups and cultural contexts
there is NO statistical test
Inter-rater Reliability - Answers - Is there consistency from rater to rater?
- observation research
- BYU preschool with Adam, measuring anger, then compare & get same results- TA's
grade evenly and fairly (consistency from rater to rater)
, statistical test: % agreement or Cohen's Kappa (2 people)
Internal (Inter-item) Consistency - Answers - Is there consistency between individual
items & the total score?
- written questionnaires
- are the questions related?
ex: thoughts of divorce are high
ex: risky sexual behavior and multiple sexual behavior are related
ex: good at subtraction, should be good at all subtraction questions
ex: answering you're happy, your life satisfaction should be high
statistical test: Cronbachs Alpha
How to obtain reliable measures - Answers - - only ask about things respondents know
- test-retest method
- split-half method
- use established measures
- train your interviewers
How can we increases reliability? - Answers - - increase # of items or observations
- eliminate items that are unclear
- maintain consistent scoring procedures
- standardize test conditions and instructions
- minimize the effects of external events
Validity is... - Answers - ACCURACY
- are you measuring what you WANT to?
- ex: getting the bullseye on the dark board
Types of Validity - Answers - 1) Face Validity
2) Content Validity
3) Criterion Validity (Current and Predictive)
4) Construct Validity (Discriminative and Convergent)
Face Validity - Answers - Do the items on the scale measure what they appear to
measure?
ex: depression scale, is it measuring sadness, lack of motivation, depression as a
whole?
ex: does 2 + 2 actually measure how good I am at addition?
Content Validity - Answers - Is the full content of the measure covered?
are you covering every possible measurement?
ex: mood ring is bad example, can't tell your mood
ex: are you male, female, or other?
Reliability is... - Answers - CONSISTENCY
- you do the exact same thing every time
ex: height, weight, getting all of the darts in the same spot
Types of Reliability - Answers - 1) Test-retest
2) Parallel (Alternate)-forms
3)Representative
4)Inter-rater
5) Internal (Inter-item) Consistency
Test-restest Reliability - Answers - When the measure is given on 2 separate occasions,
is there consistency between scores?Repeat the experiment and get same results.
- questionnaires, observations
- same set of people both times
ex: IQ test, first midterm get a 45/50, should get the same score or around it the second
time that I take it
ex: depression test
ex: running a mile - getting same time
statistical test: correlation coefficient
Parallel (Alternate)-forms Reliability - Answers - Is there consistency between forms of
test?
ex: 5 different forms of ACT (if a student took any test they would get around the same
score), spencer writes up a new midterm and half receive one and other half receives
other
statistical test: correlation coefficient
Representative Reliability - Answers - Is the measure consistent across subpopulations
or groups of people?
- applies across age groups and cultural contexts
there is NO statistical test
Inter-rater Reliability - Answers - Is there consistency from rater to rater?
- observation research
- BYU preschool with Adam, measuring anger, then compare & get same results- TA's
grade evenly and fairly (consistency from rater to rater)
, statistical test: % agreement or Cohen's Kappa (2 people)
Internal (Inter-item) Consistency - Answers - Is there consistency between individual
items & the total score?
- written questionnaires
- are the questions related?
ex: thoughts of divorce are high
ex: risky sexual behavior and multiple sexual behavior are related
ex: good at subtraction, should be good at all subtraction questions
ex: answering you're happy, your life satisfaction should be high
statistical test: Cronbachs Alpha
How to obtain reliable measures - Answers - - only ask about things respondents know
- test-retest method
- split-half method
- use established measures
- train your interviewers
How can we increases reliability? - Answers - - increase # of items or observations
- eliminate items that are unclear
- maintain consistent scoring procedures
- standardize test conditions and instructions
- minimize the effects of external events
Validity is... - Answers - ACCURACY
- are you measuring what you WANT to?
- ex: getting the bullseye on the dark board
Types of Validity - Answers - 1) Face Validity
2) Content Validity
3) Criterion Validity (Current and Predictive)
4) Construct Validity (Discriminative and Convergent)
Face Validity - Answers - Do the items on the scale measure what they appear to
measure?
ex: depression scale, is it measuring sadness, lack of motivation, depression as a
whole?
ex: does 2 + 2 actually measure how good I am at addition?
Content Validity - Answers - Is the full content of the measure covered?
are you covering every possible measurement?
ex: mood ring is bad example, can't tell your mood
ex: are you male, female, or other?
