inferential statistics Flashcards
what are descriptive statistics?
allows us to describe and draw patterns from the data
what is probability?
is the finding significant, numerical value of how likely an event is to happen
-what level of probability researchers would agree that a finding is not due to chance
- 0 to 1
what are inferential statistics?
infer if your hypothesis was correct and if the data is significant
what does significance mean?
- the probability that the result is down to chance is low
- the iv caused the change in dv
what does distribution mean?
how spread out your data is when plotted on a frequency graph
what is a significance level?
- the probability level that we accept an event is unlikely to have happened
- standard= 5%
using the standard significance level when do we accept the null hypothesis?
p=>0.05= not significant
results by chance
using the standard significance level when do we reject the null hypothesis?
p=<0.05= significant
what are the non parametric tests?
-Mann Whitney U test
- Chi Square
-Binomial Sign Test
-Wilcoxon Signed Ranks Test
-Correlations Spearmans Rho
when would you use a parametric test?
-normal distribution for data
-variances of data population should be equal
-interval data
-small range (no extreme scores)
what phrase helps determine which non parametric test to use?
lions drink cola everyday
-level of data
-difference or correlations
-experimental design
what does observed value mean?
calculated value
-number produced after calculations
what does critical value mean?
table value
-value from statistical test table
-must be reached in order for results to be significant
what does degrees of freedom mean? (df)
for chi squared test
used instead of N
-number of categories of data minus 1
choosing a stats test
-do you have nominal data?
if yes
do you have the same or different participants
choosing stats test
-do you have same or different participants?
same= sign test
-different= chi-squared test
choosing stats test
-do you have nominal data
no
are you testing for a difference or correlation
what test is always used for correlations?
spearmans rank
if testing for a difference and no nominal data what is the next question?
same or different participants
choosing stats test
-do you have same or different participants
not nominal
same= Wilcoxon
different= Mann Whitney U
what is a type 1 error?
-false positive
-accepted alternative when results due to chance
solution for a type 1 error?
more stringent significance level
(p=0.01)
what is a type 2 error?
-false negative
-accepting null hypothesis when alternative correct
solution for a type 2 error?
use less stringent significance level
(p=0.05)
what is the most common significance level and why?
5%
balance between type1 and 2 errors
1 in 20 chance
steps for Mann-Whitney U?
1) rank data across both conditions
2)add ranks for separate conditions
3)use smallest R value in formula
4)use critical table
5) U has to be equal to or below critical value to be significant and alternative hypothesis accepted
steps for Wilcoxon test?
1)find differences between condition 1 and 2
*always subtract the same way
2) rank all differences and ignore signs
3)add up ranks for + and -
4)choose smaller T value
5)compare against table
6)if T<CV result is significant
steps for Chi squared?
1) label cells
2)calculate observed values rxc/t
3) expected values (o-e)squared divided by e
4)total values
5)df (R-1)x(c-1)
6)check for significance
steps for binomial sign?
1)direction of difference
2)least common sign
3)value table
steps for binomial sign?
1)direction of difference
2)least common sign
3)value table
steps for spearman’s rho?
1)rank scored each condition separate
2)find difference between the ranks for each set of data
3)square
4)add all squared
5)put data in formula
6) use table
