correlation Flashcards
this is the variable where you do not count but measure
continuous variable
this set the limits to statistical analysis and determines which test are appropriate
type of variables
research design
it involves the same study participants or involves comparison of matched pairs of study participants, a paired test of statistical significance
before-and-after comparison
example:
in drug study
- 1 receives treatment
- other does not
in diet
- new diet
-continues their regular diet
why does some common statistical test might not work properly
u aren’t using random sampling where every1 has an equal chance of being selected
if non-random sampling = misleading results
what type of variables used for
Pearson correlation coefficient/ linear regression = C + C
Spearman correlation coefficient = C + O
student t test = C + DU (example: gender)
paired test = C + DP (example before vs after treatment)
a kind of hypothesis which gives generalization for generating records recording the mean of the primary/ original position
parametric test
- good for random sampling bcs u know the popu and SD
*usually a normal distribution & often focuses on the mean
non parametric test is aka
distribution-free test
what does it mean that non parametric does not require any population
distinct parameters
It is also a kind of hypothesis testing, which is not based on the underlying hypothesis
non parametric test
the test is based on the differences in the median.
non parametric test
what types of variables used in non parametric test and parametric test
non parametric test
- nominal
- ordinal
parametric test
- interval
- ratio
parametric test vs non parametric test
parametric test: (pearson C)
assumes
requires popu knowledge
central tendency value: mean
normal distribution
non parametric test: (spearman C)
does not assume
does not require popu knowledge
central tendency value: median
not normal distribution
Defined as the quantification of the degree to which 2 continuous random variables are related, provided that the relationship is linear.
correlation
linear - r/s or pattern
how do u visualize correlation
joint distribution graph / two-way scatter plot
scatter plot
x axis - independent
y axis - dependent
present r/s btwn 2 variables
represent on a 2 dimensional plane or cartesian system
if points trend upward:
if points trend downward:
random, no pattern:
if points trend upward: +ve C
if points trend downward: -ve C
random, no pattern: no C
when does scatter plot being used
- paired numerical data
- multiple values of dependent values for a unique value of independent variable
- determine r/s btwn 2 variables
(ex: identify potential root causes of problems)
____ variable depend on ____ variable
dependent
independent
r
referred to as r value
- varies from -1 (perfect nega C) to +1 (perfect positive C)
this measures the strength and direction of the relationship between 2 variables
Pearson Correlation Coefficient
* when 1 variable changes, the other variable changes in the same direction
value of r - degree of correlation
0 : no
0.01 - 0.35 : weak/ low
0.36 - 0.70 : average
0.71 - 0.99 : strong/ high
1 : perfect
*consider 0.99 as perfect
seeks to quantify the linear Relationship that may exist between an independent variable x and a dependent variable y
linear regression
it can be changed/ control:
it can be measured/ observed:
it can be changed/ control: independent (x)
it can be measured/ observed: dependent (y)
specifies how much y would be expected to change (and in what direction) for a unit change in x .
regression
true or false:
linear regression is related to correlation
true - produces 2 parameters that can be directly related to the data
what are the 2 parameters that can be directly related to the data
slope
- shows how steep
intercept
-whr the line crosses the y axis
difference btwn correlation and regression
correlation:
measures how strong r/s btwn x and y
regression: (predict)
clearer pic of how changes in x affect y
It’s appropriate when you have a large enough sample size and your data meets certain conditions, like being normally distributed.
parametric test
it’s useful when you have smaller samples, or when your data is ranked or categorized (like survey responses) rather than measured on a continuous scale.
nonparametric test