3. Data And Sources And Ch.1 of E-book Flashcards
what is a population
every member of a defined interest group
census
everyone in the population is measured and counted
sample
selection of smaller number of people that are representative of the whole population
variable meaning and examples
the collection of data, e.g., age, height etc.
-classed into 4 main types
what is a categorical variable
can only be assigned to a number of distinct categories e.g., blood type has to be either A, B, AB, O
what two types can categorical variables be divided into
nominal
ordinal
what is a nominal variable and example
category has no natural ordering eg, sex is either male or female
what is an ordinal variable
have categories that are ordered e,g pain can be absent, mild, severe
numerical variable meaning and example
take a numerical value
e.g. age, number of siblings
discrete variable meaning and example
only takes whole numerical values: 0, 1, 2 etc.
e.g. the number of hospital episodes a patient has
what two types can numerical data be divided into
discrete and continuous
continuous variable meaning and example
no limitations on values
e.g. weight can be 8.75426564… kg (still continuous variable even if it is recorded as a whole number)
what is the frequency distribution
description of the manner in which values of a variable are scattered
What is Anscombe’s quartet
Four sets of data with the same statistical properties but different graphical representations
In a bar chart are all the bars the same width
Yes
What is the area of a bar in a histogram proportional to
The frequency
What is the height of a bar in a histogram proportional to
Frequency/ class width
What does mean tell you and how is it calculated
Where data is centred
Add up all values/ number of values
What is standard deviation
Measure of the average distance of all the data values from the mean
What does a larger standard deviation value indicate
The values are collectively more further away from the mean and so the data values are more spread out
How to calculate standard deviation
- Calculate the mean
- Subtract the mean from every value.
- Square each of the values obtained in Step 2. Add these squared values together to give the sum of the squares value
- Divide the final result obtained from Step 3 by the total number of values in the sample minus 1, i.e. divide the sum of squares by (n - 1). This is known as the variance.
- Take the square root of the result obtained in Step 4. This is the sd.
What is the central line in the box of a box plot
The median
What is a health outcome
The impact healthcare activities have on people
- course of symptoms
- live or die
- care costs
- treatment satisfaction
What are the three types of health outcomes
Biological/ clinical eg, BMI/ blood pressure
Clinician/ patient reported outcomes eg, symptom scores, health related QoL
Record based outcomes eg, mortality, disease incidence
Examples of health outcomes that may be objective
Mortality
Disease incidence
BMI
Blood pressure
Examples of health outcomes that may be subjective
Pain
Mental health
Fatigue
What is validity and what are the 3 main types
- outcome measure what it is supposed to measure
1. Construct validity
2. Content validity
3. Face validity
What can construct validity be divided into
Convergent = are constructs that should be related, actually related Discriminant = it doesn’t measure what it shouldn’t
What is content validity
The outcome measures all facets of a given outcome
E.g. look at depression: includes both affective and behavioural symptoms such as energy loss, depressed mood, loss of interest etc
What is face validity
The outcome appears to measure what it should measure
E.g. is obesity measured as weight, BMI, body fat%
What is test- retest reliability
Are measurements consistent over time if nothing else has changed
Inter - rater reliability
Do different assessors give same results
What is responsiveness
The outcome should be able to detect real changes when they occur
What is effect size
Magnitude of change in outcomes
What is selection bias
Bias when choosing participants
E.g. not using older people as other factors would influence data but this is still biased in a way
