Unit 3: Ch. 11 Flashcards
measurement
assignment of numbers to represent an amount of an attribute
advantages:
- removes guesswork
- provides precise information
- less vague than words
what are the 4 levels (classes) of measurement?
*THIS WILL BE ON THE EXAM!
- nominal
- ordinal
- interval
- ratio
A variable’s level of measurement determines the mathematic operations that may be performed in a statistical analysis
The higher the level of measurement the more you can do with it
Nominal measurement
*THIS WILL BE ON THE EXAM!
classification into categories (ex: male/female); numbers can be assigned to nominal data but it’s just to tell the computer which is which
lowest on the measurement list
*a variable is nominal if the values could be interchanged (e.g. 1 = male, 2 = female OR 1 = female, 2 = male)
Ordinal measurement
*THIS WILL BE ON THE EXAM!
ranks objects based on their relative standing of the attribute (ex: grades [ABCDF]; hot, warm, cold; high, medium, low); numbers may be assigned but they’re just the name of the variable (not a meaningful quantity)
*a variable is usually ordinal if there is a quantitative ordering of values AND if there are a small number of values (e.g., excellent, good, fair, poor)
Interval measurement
*THIS WILL BE ON THE EXAM!
order variables on a scale that has equal distances between the points but there is no meaningful zero point
ex: temp in Fahrenheit - it has a 0, but it doesn’t mean that it’s freezing (that’s 32 degrees)
* a variable is usually considered interval if it is measured with a composite scale or test
Ratio measurement
*THIS WILL BE ON THE EXAM!
order distances into equal distances between score distances and there is a meaningful zero point
ex: Celsius - zero degrees is freezing
Best form of measurement (highest on the list)
*a variable is ratio level if it makes sense to say that one value is twice as much as another (e.g., 100mg is twice as much as 50mg)
Errors of measurement
obtained score = true score +/- error
Obtained score: an actual data value for a participant (e.g. anxiety scale score)
True score: the score that would be obtained with an infallible or perfect measure in a perfect world
Error: the error of measurement, caused by factors that distort measurement
-lots of things that may distort measurement
Want to minimize the error and maximize the true score (with any data)
what are the 4 factors that contribute to errors of measurement?
- situational containments
- transitory personal factors
- response set biases
- item sampling
situational containments
things outside of the person but within the person’s surroundings
-ex: noise during an exam
transitory personal factors
things like fatigue, anger
- usually personal factors that are temporary
- things inside the person that may change how the person responds
response set biases
measurement error resulting from the tendency of some individuals to respond in characteristic ways independent of the content
-answering in politically correct ways, etc.
item sampling
when the item itself doesn’t cover the domain of the contents
-not valid (not measuring what it says its measuring)
reliability
the consistency with which an instrument measures the target attribute
reliability assessments involve computing a reliability coefficient
- reliability coefficients ranges from .00-1.00 (perfectly reliable)
- coefficients below .70 are considered unsatisfactory (not reliable)
- -> things below .70 may not be published
- coefficients of .80 or higher are desirable
qualitative: reliability is located in methods section
quantitative: reliability coefficients are listed where each measurement is
what are 3 aspects of reliability that can be evaluated?
- stability
- internal consistency
- equivalence
stability
the extent of which scores are similar on 2 separate administrations
test-retest reliability: must complete same instruments on 2 separate occasions
creativity, traits (ex: how anxious you are)
internal consistency
aka homogeneity. Extent of which all items in an instrument are measuring the same attribute
-low coefficient in this means not all items are measuring the same thing
equivalence
degree of similarity between alternative forms of an instrument
-to determine equivalence, you administer both of those instruments and want there to be some similarity between results. Pick one you already know is reliable and then one you’re not sure about - if the new instrument is doing what it should do, you’ll get a high reliability coefficient
reliability principles
low reliability can undermine adequate testing of hypotheses
- if an instrument isn’t consistently measuring what it’s supposed to be measuring, you don’t know what results you got
- must have good reliability
reliability is lower in homogeneous than in heterogeneous samples
-different types of participants (heterogeneous) are answering the question in the same way means the instrument is really tapping into whatever you’re trying to measure
reliability is lower in shorter than in longer multi-item studies
- the longer the instrument the higher the reliability
- find the perfect length of instrument (longer instrument = more time consuming but more reliable)
validity
the degree to which an instrument measures what it is supposed to measure or purports to measure…accuracy.
4 aspects of validity:
- face validity
- content validity
- criterion-related validity
- construct validity
face validity
instrument looks as though it is measuring the appropriate construct
based on judgment, not necessarily based on objective criteria
content validity
tapping into the domain of interest
concerns the degree to which an instrument has an appropriate scale of items for the constructs being measured
can calculate a content validity index (CVI) that indicates the extent of expert agreement
-book suggests CVI level of .90 or higher as standard for establishing excellence in a scale’s content validity
researchers designing a new instrument should begin with a thorough conceptualization of the construct, so that the instrument can capture the full content domain.
**such a conceptualization usually comes from a thorough literature review, a concept analysis, or findings from a qualitative inquiry
criterion-related validity
degree to which the instrument is related to an external criterion
-researchers examine the relationship between scores on an instrument and an external criterion
calculated by analyzing relationship between scores on the instrument and some outside criterion
predictive validity - ex: ability of your high school GPA to predict how you’ll do in college
-taking something that happened in your past to predict how you’ll do in the future
concurrent validity: ex: your med surg scores predicting your mental health scores on exams
-2 things that are basically happening in the same space of time
construct validity
has to do with the constructs that make up the measure
-asking what the instrument really measures (is it really measuring fear, or is it measuring dread, apprehension?)
“what is this instrument really measuring? Does it validly measure the abstract concept of interest?”
construct validation is essentially a hypothesis-testing endeavor, typically linked to theoretical conceptualizations.
criterion-related validity: predictive validity
ex: ability of your high school GPA to predict how you’ll do in college
- taking something that happened in your past to predict how you’ll do in the future
criterion-related validity: concurrent validity
your med surg scores predicting your mental health scores on exams
-2 things that are basically happening in the same space of time
sensitivity
the instrument’s ability to correctly identify a “case” - i.e., to dx a condition
- have a dz and test positive for it
- needs to be high when the dz is lethal (ex: ovarian cancer), rare (ex: PKU), or has a latent period (ex: cervical cancer)
- better to have a few false positives than to miss the dz if it’s present
specificity
the instrument’s ability to correctly identify non-cases, that is, to screen out those w/o the condition
- to give a negative result when a person truly is dz free
- must be 100% to detect all people w/o the dz
- ex: specificity of 85% means it detects 85% of people w/o the dz
- needs to be high when the tx doesn’t markedly alter the outcomes or detection of the dz is important but not life threatening (ex: pregnancy test)
- better to have a few false positives
- tests w/ high specificity have few false positive results
sensitivity and specificity
you’ll see percentages with both
inverse relationship. As one goes up, the other goes down. Can’t have 100% on both of them. Want the highest percentage you can have on both of them
reliability coefficient
a numeric index that quantifies an instrument’s reliability to assess objectively how small the differences are
designed as “r”
range from .00 to 1.00; the higher the value, the more reliable (stable) is the measuring instrument
coefficient alpha
aka Cronbach’s alpha
calculate coefficient alpha to evaluate internal consistency
normal range of values is .00 to +1.00. The higher the coefficient, the more accurate (internally consistent) the measure
interrator reliability
aka interobserver
involve having two or more observers or coders make independent observations
An instrument’s validity is not proved or verified, but rather is supported to a greater or lesser extent by ____.
evidence
basically evidence is needed for validity
cutoff point
a score value to distinguish cases and noncases
T/F: Validity is more difficult to document than reliability.
true
validity coefficient
computed by using a mathematic formula that correlates the 2 scores in criterion-related validity
the coefficients (r) range between .00 and 1.00, with higher values indicating greater criterion-related validity -coefficients of .70 or higher are desirable
give examples of nominal, ordinal, interval, and ratio
Nominal: think “names or labels”
-ex: what is your gender (M/F)? what color is your hair (brown, blonde, black, red)?
Ordinal: think “order;” order of values is significant and important; typically measures non-numeric concepts like satisfaction, discomfort, happiness, etc.
-ex: rate your happiness: 1 = very unhappy, 2 = unhappy, 3 = ok, 4 = happy, 5 = very happy
Interval: think “interval” (“space in between”)
-ex: temperature (there is no such thing as “no temperature” –> interval has no zero meaning)
Ratio: have an absolute zero
-ex: height and weight, time
