Flashcards in Exam 2 Quantitative Measurement Deck (67):
Where is data collection found?
Methods section of the article.
Rule of thumb
"Garbage in, garbage out" If you make poor measurements, you're going to get bad results
Pretest tools. Why?
Performs as expected
Ease of administration
Data is plural, datum singular
Different word, same thing
how to measure something.
Example: A way to measure pain...facial expression, body language
An instrument (or tool) that has been checked for reliability/validity, always will work the same
Take a general topic, break it down
As accurate as possible
Applies to the general population
Tidy flow...visually aesthetic
Specific to topic...
Example: DEPRESSION, not anxiety or sadness
Time appropriate, cost appropriate
High(forces subject to pick a specific answer), middle, low(open ended questions)
Ability of data to be recorded by numbers
To what extend does the subject know they are being observed/assessed
To what extend are the results subject to bias
Observational checklist - unique to this design
Questionnaire - very popular
Biophysical Measurement...why use?
Relevance to nursing (BP, HR, Temp, Lab values...etc)
Impact of nursing actions
Evaluate nursing procedures
Find health related correlations
Biophysical Measurements need to be...
Appropriately collected, recorded, stored, tested, and reported
Interviews, written questionnaires, etc
Checklists, observational rating scales
Ex: strongly agree, agree, undecided, disagree, strongly disagree
Visual Analog Scale (VAS)
100mm line, subject places vertical line on scale based on their answer.
Forced choices, range
Ex: range of sad to happy, five blank lines, place check mark saying how happy you feel.
Sort index cards into piles...agree/disagree, then agree/strongly agree; strongly disagree/disagree
Appropriate for right brained people/children
Ex: Ink blots, pictures, music. Opinions are projected on the image
Criteria when examining tools
Appropriate (to question/design/population)
Acceptable to subject
Reliable and valid
Need to reflect as much truth as possible in our measurements
True value + Error
Error can be:
How can you reduce error?
Use reliable and valid tools.
Examines stability, equivalence and internal consistency of the tool
.90 and above is very reliable
.80-.89 very good
same instrument given more than one time using same conditions
find if there is a correlation (r) between scores
want a strong correlation
two versions of the same instrument
Example: use the same questions, but move them around...put them in a different order
The questionnaire deals with the same conceptual area consistently throughout the tool
ex: all the questions are about "happiness"
see if 1st half and 2nd half of test are highly correlated
item to total
strong correlation in total score
Do two or more people score the observations the same
Gives the researcher a correlation coefficient
If doing the test-retest...
the longer the lag time, the lower the r
The more homogeneous the sample...
they lower the reliability.
The tool actually measures the variable of interest.
Can a tool be valid without being reliable?
NO. If a tool is valid, it must be reliable. However, a tool can be reliable without being valid.
The items (questions) look appropriate to the general population. It "looks right."
Items derived from literature and expert advice.
Items look appropriate to experts in the field
Items with questionable ratings are modified or dropped.
Do the items measure all important aspects of the concept and are those important aspects measured appropriately.
Compares to another measure...logically connected
-Tries to determine how observed score might compare to the true scores
This measurement correlates well with predictions made using this measurement.
Ex. if you have a strong high school GPA, you'll have a strong college GPA
This measurement correlates well with another 'gold standard' measurement given at the same time.
The ability of an instrument to correctly identify a "case" or correctly screen for or diagnose a condition.
Ex. This IS depression
The ability of an instrument to correctly identify 'non-cases' or to rule out those without the condition.
Ex...this shows it is NOT depression
Very low... 5%, 10% with a reminder
may limit responses...age, physical limitations
Needs to be low in complexity in order to be comprehended
specifies all the possible values a given measurement may have
All scales have:
-At least two values
-Mutually exclusive categories
1 for boy, 2 for girl
Using an arbitrary number represent your variable
ex. 1 for boy, 2 for girl
Using a systematic ordering of numbers to represent ordered responses.
Not measured...no 'distance' between numbers.
Ex. Strongly agree=5....strongly disagree=1
Using a systematic ordering of numbers to represent ordered responses
there IS a true distance between the numbers
NO true zero point
ex. temperature (above and below zero)
There is a true distance between the numbers
IS a true zero point