Chapter 4: Defining And Measuring Variables Flashcards
Variable
Any factor that has 2 or more values
Constant
Something that is non-changing
Qualitative Variables:
properties that differ in type (Sex, religions, eye colour)
can still be statistically analyzed measuring number of instances in each category
Quantitative Variables:
properties that differ in amount (height, weight, distance)
Discrete Variables
between 2 adjacent values no intermediate values are possible
Continuous Variables
Intermediate values are possible between any 2 adjacent scale values
CONVERTING continuous variables into discrete variables in order to measure findings is very common
(example: intensity of depression, measured on a 5 point scale)
Independent Variable
- presumed cause in a cause effect relation
- manipulate or systematically vary in order to see effect on a behaviour or outcome
Dependent Variable
presumed effect in a cause effect relation
-behaviour or outcome that researcher measures to determine whether independent variable has produced an effect
Situational Variable
A characteristic that differs across environment or stimuli
Subject Variable
A personal characteristic that differs across individuals task
Can independent variables be used in descriptive research?
Yes, they are not manipulated, but the variable can be discussed and conceptualized, an example of this would be measuring through interviews how religious a subject is
Hypothetical Constructs and example!
underlying characteristics or processes that are directly NOT directly observed but instead are inferred from measurable behaviour or outcomes
Example: can’t observe hunger, but can infer it
-must be translated into something measurable by operalizatinoalism
Manifest Variables
are variables easy to observe like physical characteristics
Operationalization
translate a construct into something that can be measured
need to properly conceptualize and communicate how variable will be manipulated or measured
Mediator Variables (why)
Provides a causal link between dependent and independent variable and explains relationship
Moderator Variables (when, and whom)
A factor that alters strength or direction of the relationship between the independent and dependent variable
An Independent Variable may have different effect on dependent variable depending on age, gender, self esteem
Conceptual definitions:
Planning errors, execution errors: defining key concepts
Measurement
Systematically assigning values(#’s, labels) to represent attributes of organism, objects or event.s
Scales of Measurement 4 MEASUREMENTS
-Impacts how data can be represented, analyses and interpretted
Nominal
Ordinal
Interval
Ratio -most precise
Nominal Scale
- Meant for organizing data
- scale values represent qualitative differences
- Group variables into categories
- Differ in type NOT degree
- numbers can be assigned to represent category. but not numerical
Ordinal Scale
- different scale values represent relative difference in the amount of some attribute
- rank order
- can be represented by numbers or category
- tells what level an item belongs to, but not difference between rank orders
- doesn’t say how distant an A is from a B just provides info of greater, worse, more than, less than
Interval Scale
When equal difference between values on the scale reflect equal differences in the attributes being measured
- numbers reflect the actual amounts
- 0 point is arbitrary
- Difference between 4 and 8 degrees same as 10 and 14
- often record psychological measures on this scale, but sometimes if can’t draw equal intervals it is an ordinal scale
Ratio scale
when equal values on the scale reflect equal difference in the amount of attribute being measured
- like Interval scales, but has a true 0 point
- Provides most info about an attribute being assessed
- A score of 0 means absence of attribute
- time length, annual income is measurable
Accuracy of Measurement
degree to which the measure yields results that agree with a known standard.
If measurement tool to measure weight isn’t collobarted properly then weight won’t be accurate
Measurement error
Things we don’t want to measure but do, due to imperfections of our measurement tool
True score
actual degree of characteristic of individual being measured
Systematic Error
A consistent degree of error that is inaccurate with each measurement
Reliability
Consistency of measurement tool
Random measurement error
random fluctuations during the measurement and cause obtained scores to deviate from try score
Test-Retest Reliability
administering the same measure to the same participants on 2 or more occasions under equivalent test conditions
Alternative forms
A method where 2 versions of the same test are administered to the same participants and different times, to avoid bias effect to reassure reliability of results
Split-Half Reliability
what is the average of all split half combinations
Items that compose a test that are divided into 2 subsets; and correlation between 2 subsets is determined
-assesses internal consistency
-split-half combinations average: Cronbachs Alpha
Interobserver reliability
degree to which independent observers show agreement in their observations
Validity of measure
Alignment between construct and measurement tool we employed to gain insight into the construct
Face Validity:
The degree to which items talked about in the measure seem reasonable to participants, and makes sense based on studies purpose
Content Validity:
degree to which the items on a measure adequately cover all domain of interest (range or set of items)
Criterion Validity:
and what are the 2 different types
- Addresses relation between scores on a measure and an outcome
- Predictive validity: scores on a job test predict increase job performance in the future
- Concurrent Validity: Give GRE test to current college students to see if current results are a factor of success
Construct validity:
Measure that truly assesses the construct that it is claimed to assess
Convergent Validity:
scores on a measure should highly correlate with scores on a measure of the same construct
-ex/ scores on new depression scale should correlate high with scores on previously established depression scale
Discriminant Validity:
Scores on a measure should not correlate strongly with scores on a measure of other constructs
