Statistics Flashcards
sampling bias
when sample isn’t representative of the population
3 types of correlational research
- naturalistic observation
- survey application
- documentary research
3 categories of descriptive statistics
- distribution
- central tendency (mean, median, mode)
- measures of variability (range, standard deviation, variance)
inferential statistics meaning
drawing conclusions about population’s features based on results from test
one sample t-test meaning
comparing mean sample of population and small sample taken from population
two-sample t-test meaning
comparing mean values of independent sample 1 and independent sample 2
paired sample t-test meaning
comparing value of sample 1 at time 1 to value of sample 1 at time 2
when do we have strong positive correlation?
When Pearson’s r is bigger than 0.5
operational definition meaning
precise description of how variables will be measured
true experiment
scientist has complete control over manipulation of IV + random allocation
randomised control trial
scientific experiment where similar people allocated to 2 groups to test intervention
cross sectional research
benefits
disadvantages
- data collected in one specific point in time to compare subgroups
benefits:
+ compare subgroups
disadvantages:
- no measures over time
systematic sampling definition
every nth person gets chosen
stratified sample definition
type of random sampling
population divided into groups based on characteristics then randomly selected from those
cluster sampling
all population divided into clusters
4 non-probability sampling methods
- non-convenience sampling
- purposive sampling
- quota sampling
- snowball sampling
purposive sampling definition
handpicked participants based on their characteristics
nominal data definition
data without numerical value
*colours of rainbow
*countries
ordinal data definition
categories with specific order that can be ranked
*military ranks
*economy classes
continuous variables
numerical that can assume any number
*weight
*temperature
interval variables definiition
ordered + quantifiable but with no true zero
- temperature in the sense that 0 celsius isn’t absence of temperature
ration intervals
- quantitative but have an actual zero
*height
*income
normal distribution characteristics
- mean = 0
- SD = 1
- kurtosis = 3
- tails touch x axis at infinity (never)
one tailed directional hypothesis
predicts direction of effect
reading will equal HIGHER intelligence levels
two tailed directional hypothesis
predicts that IV will impact DV but doesn’t say what direction
reading will have direct impact on intelligence levels
type 1 error
false positives
- reject null when null is actually true
type 2 error
false negative
- accept null hypothesis when alternative actually true
2 tests for between groups
- independent samples t-test (parametric)
- mann-whitney U test (non-parametric)
parametric meaning
evenly distributed
non-parametiric
not evenly distributed
2 tests within groups design
- related samples t-test (parametric)
- Wilcoxon signed rank test (non-parametric)
what does correlation analysis tell us
- direction
- strength
Pearson’s r
- continuous variables
- perfect correlation at 1 (means data points on straight line)
Spearmann rank order correlation (rho)
- ordinal value
*ranking students’ performance
ANOVA
- comparison of more than 2 groups
correlation what does it show
- degree of association between 2 variables
multiple regression definition
- relationship between multiple IVs on one DV
IV -> predictor variable (continuous or categorical)
DV -> outcome variable (continuous)
hypothesis definition
testable prediction of relationship between two or more variables
standard deviation
spread of data that shows how scores deviate from the mean
square root of variance
variance definition
how scores are distributed around the mean
how to reduce type 1 error
- set lower p value
to reduce type 2 error
- pick bigger sample size
- increase p-value (increases chance of type 1 error)
when to use mann Whitney
for independent samples t-test with non parametric data
when to use Wilcoxon
paired samples t-test with non-parametric data
does correlation have independent variable?
- NO!
we don’t manipulate any variables just look at the relationship
mode what data can it be used for
nominal data
- mode is most common answer
