research lecture 5 Flashcards
what are parametric statistics used for
to analyze QUANTITATIVE DATA
what are examples of parametric Statistics that are used to analyze quantitative data
t-test, ANOVA, Pearson correlation, linear regression
parametric stats are based on which distrcutions so the data needs to be normalized
t-distribution, F-distribution, chi-
square distribution
non parametric statistics are used to analyze what kind of data
qualitative
t Spearman rho, Mann-Whitney U, Friedman’s ANOVA, Wilcoxon-signed ranks .. these are examples of what
non parametric statistics that are used to analyze qualitative data
what is our options when we have =violated assumptions or have nominal or ordinal data
non parametric statistics
what is a statistical way to look at relationship among variables and it is based on a linear models
regression
when is a linear regression significant
when the slop is not equal to 0
what is a T test
see the difference between 2 means and it is significant if the slop does not equal 0
what are teh parametric assumptions for t-test or one way ANOVA
- I/R data
- Normality
- Homogeneity of Variance
- Free of Extreme outliers
- Independence of observations
what is normality
real concern in smaller studies with a n < 30 (bc of the central limit theory)
what are 3 way to assess normality
-check histogram
-skewness/kurtosis (if >2 or < -2 there is a problem)
- shapiro wilk test (want this to be greater then .05 to be significant)
what happens to skewness as the sample size increases
it improves
what is homogeneity of variance looking for and what should be the same
looking for difference , the variances of the outcome variable should be able the same in each group
how to assess for homogeneity of variance
Levene’s test
what do u want the levenes test to show
want it not to be significant , u want them to show “no difference”
set alpha at .05
data must be ___ of observations .. scored must not follow a pattern over time also scores from one participants can’t influence another participants score
independent
when u are looking at a graph how do u know if there is a problem with independence
if some of the scores are the same or follow a pattern there is a problem , it should all be reandiozned
what are 3 regression assumptions
- linearity
- homoscedasticity
- outlier testing in regression
what is the difference between homogeneity of variance and homoscedasticity
HOV is the difference in stats and homoscedasticity is the relationship between stats
(T/F) In correlational/relationship analyses (ex regression), the variance of the outcome variable must be about the same at all levels of the predictor variable
T
if the variance is no evenly distributed what is it called
having hereoscedasticity
when looking at a graph how can u tell the difference between homoscedasticity and heterosscedasticity
Homoscedasticity the dots are along the line of regression and for the other one the dots kinda look like a funnel it does from skinny to wide
how are the data points arranged in linearity
in a linear pattern
what is the easiest way to check for linearity
creat a scatterplot
what kind of residual does the outlier have
a large one
how do u find residual on a linear graph
its the distance from the dot to the actual line
what are scatterplots also used for
to check homoscedasticity and outliers
Scatterplots are also used to check for homoscedasticity and outliers. To do this…we must do what
figure out every participants residual score
to have homoscedasticity all of the ___ should be about the same
residuals
when you violate an assumption what are 2 ways that is good to do to fix it
- analyze with bootstrapping in SPSS
- use non parametric statistics
what are 3 bad things to do if u violate the assumption and u want to fix it
- trim the data
- windsorizing
- transform the data
what is a hypothesis test
a statical method that uses the sample data to evaluate a hypothesis about a population
what is the general goal of a hypothesis test
rul out chance (sampling error) as a plausible explanation for the results from a research study
what are teh steps in a hypothesis testing
- bathing evidence . make a case for ur study
- state hypothesis
- select test statistic
- calculate test statistic
- statistical descion
- reject the HO or do no reject the HO , if reject then alternative HO is true if do not reject then null is true
the purpose of the hypothesis test is to decide between 2 exmplnations
- the difference between the pre/post samples CNA BE EXPLAINED BY THE SAMPLING ERROR (fail o reject the null
- he difference between the pre/post samples is too large to be explained by sampling error. (there does appear to be a treatment effect).
- reject the null
what hypothesis states there is no change int he population before and after and intervention
null
in the context of an experiment, H0 predicts that the independent variable had no effect on the dependent variable…. what states this
the null
what hypothesis states there is a change in the population following an intervention
alternative hypothesis
what type of error fixes the alpha level determine
the risk for type 1 error
what is it called on a graph where it consists of outcomes that are very unlikely to occur if the null hypothesis is true
critical region
what is calculated as the difference between he pre and post means / the amount of diffference one would expect without any treatment effect
test statistic
if the test statistic is int he critical region not he graph what can we conclude and what do we reject
the difference is significant or that the treatment has a significant effect and we reject the null
If the test statistic is not in the critical region, what can we conclude
that the evidence from the sample is not sufficient and the descision is fail to reject the null
what is the actual probability that the results occurred just because of sampling error.
the p value
when is the p value significant
when it is smaller (or equal) to the alpha level that is set in advance
if the p value is within the small end of the tails then it is ___
significant
The hypothesis test is influenced not only by the size of the treatment effect and the variability of the sample but also the size of the ____ .
sample
what do we use as a standaridized measure of effect size
cohens D
what does cohens d measure
the size of the mean difference in terms of the standard deviation
whar are 2 ways to increase the ES
- increase the mean difference
- decrease the SD
if r (correlation analyses) =.1 , d=.2 what kind of effect size does it have
small
if r (correlation analyses) =.3 , d=.5 what kind of effect size does it have and what does r^2 mean
medium
the effect explains 9% of the total variance between the 2 samples
if r (correlation analyses) =.5 , d=.8 what kind of effect size does it have and what does r^2 mean
large
the effect explains 25% of the total variance between the two
what does the larger the cohens D mean
the bigger the effect size and power increases
what is the power of the hypothesis test
the probability that the statistical test will reject the null hypothesis when the treatment does not have an effect
how do we increase our power of a hypothesis test
1- increase the ES (increase the difference between the groups or decrease the variability)
2. increase the a sample size
3. increase the alpha (.01 to .05)
4. use a 1-tail test
what has more power
More difference between groups or Less difference
More variability within groups or Less variability Larger sample size or Smaller sample size
Alpha = .01 or Alpha = .05
1 tail test or 2 tail test
more
less
larger
.05
1
what is used to do power before and after research
g power
if u find a significant result then u had enough ___
power
what are the 2 types of t est
independent test and dependent test
what T test Compares two means based on independent data
Ex/ Data from different groups of people
independent
what t-test Compares two means based on related data
Ex/ Data from the same people measured
at different times (pre-post testing)
Ex/ Data from ‘matched’ samples or twins
dependent
independent t-tests have one ____ variable and one —— variable
idepdnet and dependent
skewness and kurtosis is okay is it is between what
+2 and -2
do we want the shapiro wilk for testing normality to be above or below .05
greater bc we do not want it to be significant
if our critical t value is 2.191 what does the t value need to be in order to be significant
greater then 2.191 ir smaller then -2.191
what do we want th number for levenes test for equality of variances to be
larger then .05 for assumptions to be met
what is smaller if the HOV is not met
DF is smaller.. less likely to reject the null
when do we use repeated measure TTest
- comparing twins or matched pairs
- studies that measure the same participants twice (ex. pre/post)
when will repeated measures T-test be more powerful
if the degress of freedom (number of participants) for the 2 t-test are about the same
what is a independent T-test for a non parametric
mann - whitney U test
what is a repeated measures T-test for a non parametric
wilcoxon signed ranks test
the indepdent t test and dependent t test for non parametric are done on the ___ ___
mean ranking
if you are repeatedly testing the same dataset , you are more likely to commit what type of error
type 1
what is the bonferroni corrections
divides the alpha by # of tests u plan to run