Professional Documents
Culture Documents
Descriptive Statistics
Parametric Statistics
Non-Parametric Statistics
Null Hypothesis
Alternative Hypothesis
Mean
Standard Deviation
Correlation
Confidence Interval
Fit the statistics to the research question, not the
other way around!
First, ask yourself, “Am I interested in….
Describing a sample or outcome?”
Looking at how groups differ?”
Looking at how outcomes are related?”
Looking at changes over time?”
Creating a new scale orinstrument?
Assessing reliability and/or validity of an instrument?
Second, “How am I measuring my outcomes?”
Descriptive Statistics
Parametric Statistics
Common tests of relationships
Pearson r
Linear/multiple regression
Common tests of group differences
Independent t-test
Between subjects analysis of variance (ANOVA)
Common tests of repeatedmeasures
Dependent t-test
Within subjects ANOVA
Tests of categorical data
Odds Ratio / Chi Square
Logistic Regression
Common Psychometric tests
Cronbach’s Alpha
Principal Components and Factor Analysis
Numbers used to describe the sample
They do not actually test any hypotheses (or yield any
p-values)
Types:
Measures of Center -
Mean
Median
Mode
Measures of Spread -
Quartiles
Standard Deviation
Range
Variance
Frequencies
Most powerful type of statistics we use
Researchers must make sure their data meets a
number of assumptions (or parameters) before
these tests can be used properly.
Some keyassumptions
Normality
Independence of observations
In research, you always want to use parametric
statistics if possible.
Pearson r correlation
Linear/Multiple Regression
What is it?
A statistical analysis that tests the relationship
between two continuous variables.
Commonly Associated Terms:
Bivariate correlation, relationship, r-value, scatterplot,
association, direction, magnitude.
StrWoNnegoakRMeReleaoaltdtaieotorinnoasnthhesihppi:pr:>
.50 Rre≈≈lat|i..o01n0s|hip:
r ≈ |.30|
10
Each has a Pearson
Correlation of r=.82, is & is
statistically
significant
11
Anscombe, F.J., Graphs in Statistical Analysis, American Statistican, 27,17-21
What you read:
Study found a relationship between GPA and sense of
belonging, r=.35, p = .03.
What to interpret:
Results show r = .35, p = .03, R2=.12
How to interpret:
There is a weak, significant positive relationship
between college GPA and students’ sense of
belonging to the university. As sense of belonging
increases, GPA also increases.
What is it?
A statistical analysis that tests the relationship
between multiple predictor variables and one
continuous outcome variable.
Predictors: Any number of continuous or
dichotomous variables, e.g. age, anxiety, SES
Outcome: 1 Continuous variable, e.g. ER visits per
Month
Commonly Associated Terms:
Multivariate, beta weight, r2-value, model,
forward/backward regression,
sequential/hierarchical regression,
standard/simultaneous regression,
statistical/stepwise regression.
13
Independent t-test
Between Subjects Analysis of Variance (ANOVA)
What is it?
Tests the difference between two groups on a single,
continuous dependent variable.
Commonly associated terms:
Two sample t-test, student’s t-test, means, group
means, standard deviations, mean differences, group
difference, confidence interval, group comparison.
What to interpret?
p-values (<.05)
Mean differences and standarddeviations
Confidence intervals
How to interpret?
There is a significant difference between the two
groups where one group has a significantly
higher/lower score on the dependent variable than the
other.
What you read:
Students who were put on academic probation (M=1.50,
SD=.40) had lower sense of belonging than students who
were not put on academic probation (M=3.50, SD=.75), p =
.02.
What to interpret:
p-value: .02
Mean sense of belonging for both groups: academic
probation = 1.50 & non-academic=3.50.
Standard deviations for both groups: on academic probation
=.40 & not on academic probation=.75.
How to interpret:
Participants on academic probation had significantly lower
sense of belonging than students who were not put on
academic probation.
What is it?
Tests the difference among more than two groups on a
single, continuous variable.
Post-Hoc tests are required to examine where the differences
are.
Commonly associated terms:
F-test, interactions, post-hoc tests (tukey HSD,
bonferroni, scheffe, dunnett).
What to interpret?
p-values (<.05)
Main effect: Shows overall significance
Post-hoc tests: shows specific group differences
Mean differences, standard deviations
How to interpret?
Main Effect: There was an overall significant
difference among the groups of the independent
variable on the dependent variable.
Post-Hoc: Same interpretation as an independent t-
test
What you
read:A researcher looks at differences in average satisfaction on
three different reading interventions (A, B, and C).
Main effect: Overall F=20.10, p=.01
Post-hoc: Comparison of Intervention “A”to Intervention “B” shows
average satisfaction to be 4.32 (SD=.50) and 3.56 (SD=1.2),
respectively, p=.04.
What to interpret:
Main effect: p-value=.01
Post-hoc: p-value=.04, group means show Intervention “A”has
higher satisfaction ratings than Intervention “B”.
How to interpret:
Main effect: There is a significant overall difference among the
three interventions on satisfaction.
Post-hoc: Students who received Intervention “A”have
significantly higher satisfaction than those who received
Intervention “B”
Dependent t-test
Within Subjects Analysis of Variance (ANOVA)
What is it?
Tests the differences for one group between two time-points
or matched pairs
Commonly Associated Terms:
Pre and posttest, matched pairs, paired samples, time.
What to interpret?
p-values (<.05)
Mean change between measurements (i.e. over time or
between pairs)
How to interpret:?
There is a significant difference between the pretest and
posttest where the score on the posttest was significantly
higher/lower on the dependent variable than the pretest.
What you read:
An article shows a difference in average test score
before (M=79.50, SD=8.00) and after (M=85.25,
SD=7.90) an educational intervention, p=.08.
What to interpret:
p-value=.08
Mean change=7.75 more points after the educational
intervention.
How to interpret:
Average test score did not significantly change from
before the intervention to after the intervention;
however, there may be a practically relevant difference.
What is it?
A statistical analysis that tests differences of one group
between two or more time-points or matched pairs (e.g.
pretest, posttest, & follow-up or treatment “A”patient,
treatment “B” matched patient, & placebo matched patient).
Commonly Associated Terms:
Multiple time-points/matched pairs, repeated measures, post-
hoc.
What to interpret?
Main effect: p-values
Post-hoc: p-values, mean change, direction of change.
How to interpret:
Main Effect – There was an overall significant difference
among the time points/matched pairs on the dependent
variable.
Post-Hoc: Same as a dependentt-test.
What you read:
An article shows a difference in average classroom comfort
before (M=1.5, SD=2.0), after (M=3.30, SD=.90), and six
months following a cohort-building intervention (M=4.20,
SD=3.0).
Main effect: Overall F=3.59, p=.02.
What to interpret:
p-value=.02, statistically significant
Mean change=1.8 higher classroom comfort at post-
intervention
How to interpret:
Classroom comfort significantly increased from baseline to
six-months following a cohort-building intervention;
however, post-hoc tests will be needed to show where that
differences lies.
Mixed ANOVA: Used when comparing more than one group over
more than one time-point on a measure
Example – Males vs. female students, before and after a foreign
language course – Average score on an assessment