Inferential statistics Flashcards
What are inferential statistics?
Tests applied to data to work out whether the null hypothesis or alternative hypothesis was supported
Allows us to draw conclusions
Accepting the alternative hypothesis
The data found that the IV had a significant effect on the DV
because there was a significant difference between data scores of conditions with IV, and without the IV
Accepting the null hypothesis
The data found the IV had no significant effect on DV and any differences were due to chance
No significant difference between scores of condition with IV and scores of conditions without the IV
2 types of Inferential statistic test
Parametric (powerful)
Non parametric (less powerful)
What does the non parametric test that we use on a data set depend on?
Experimental design
Level of data
what test do we use for Nominal level data from an independent measures experiment?
Chi square test
what test do we use for Nominal level data from a repeated measures/ matched participant experiment?
Binomial sign test
what test do we use for ordinal level data from an independent measures experiment?
Mann-Whitney U test
Why is there no test for nominal level data from a correlation study?
Nominal data is headcounts so data can’t be put on a scale thus cannot be used for a correlation study
what test do we use for ordinal level data from a repeated measures/ matched participant experiment?
Wilcoxon signed ranks test
What test do we use when we have ordinal level data from a correlation study?
Spearman’s Rho correlation coefficient
3 criteria for when we use a parametric test
Data is interval or ratio
It has a normal distribution curve
The variances of each conditions’ data are similar (similar distribution around mean)
Level of significance
The level at which the difference between 2 values are found to be statistically significant
SO there is a low probability that the differences are due to chance
p value
The probability found that the results are due to chance
Type 1 error
false positive error = falsely claimed to have found significant difference
They found the probability results were due to chance was lower than reality: lower than level of significance allowing them to make false claim
Type 1 error in terms of accepting/rejecting hypothesis
falsely accepted alternative hypothesis when they should have rejected it
Should have accepted null hypothesis
Type 2 error
false negative error = falsely claimed to have found no significant difference when there was one
Found probability the results were due to chance to be larger than reality: higher than level of significance allowing them to make false negative claim
Type 2 error in terms of accepting/rejecting hypothesis
falsely accepted null hypothesis when they should have rejected it
falsely rejected alternative hypothesis when they should have accepted it
How to remember difference between type 1 and 2 errors
Type 1 error = person boastful about getting number 1 when they shouldn’t
Type 2 error = person who always puts themselves second wrongly when they shouldn’t
Distribution curves
A type of graph with a (distorted or bell shaped) curve that data about a behaviour from a sample will fall around: shows the data’s distribution and range
Peak of all distribution curves
The mode = scores the most number of participants achieved
Types of distribution curves
Normal distribution
Positively skewed distribution
Negatively skewed distribution
As you move away from the mode on a distribution curve what happens?
Fewer and fewer participants scored these values
Normal distribution curve
All measures of central tendency fall around the centre of the curve as most people achieved this: the median nor mode was not skewed to be higher/lower because there were no outlier scores
Normal distribution curves usually represent the…
Average sample of the whole population
Negatively skewed distribution curve
The mean is smaller than the mode (mode is the peak) (mean is to the left of the mode)
Curve slopes into the y axis
What do negatively skewed distribution curves show?
More people scored towards the higher end of data set and achieved a high score (mode is high)
But data and mean is skewed by few people who scored lower (negative skew)
How to remember negatively skewed distribution?
Slide that hits into the y axis (slopes into y axis and steep climb as the ladder on the right side)
A NEGATIVE EXPERIENCE
Positively skewed distribution curve
The mean is larger than the mode (mode is peak) mode is further to the right than the mode peak
Curve slopes away from the y axis
What do positively skewed distribution curves show?
More people achieved a lower score thus mode decreases but few people scored very high thus giving a positive skew
So data and mean is skewed by those who achieved toward higher end thus positive skew
How to remember positively skewed distribution?
Slide that you climb the steep end near the y axis and slide away from it
A POSITIVE EXPIRIENCE
When can we use standard deviation on a distribution curve?
Only for normal distribution curves not one with positive or negative skew
Standard deviation on a distribution curve: 1 s.d
34% of results will score 1 s.d above the mean
34% of results will score 1 s.d below the mean
68% of results will score 1 s.d above and below the mean
Standard deviation on a distribution curve: 2 s.d
95% will score 2 s.d above and below the mean
