Basic and the normal distribution Flashcards
H0 (null hypothesis) ?
no difference between the means
we can reject the null hypothesis which will mean that there is a significant difference, but never prove the alternative hypothesis
H1 (alternative hypothesis) ?
difference between the means
we can reject the null hypothesis which will mean that there is a significant difference, but never prove the alternative hypothesis
Systematic variance:
introduced by the experimental manipulation (good)
Unsystematic variance (aka error variance, residual variance, unexplained variance):
inherent to the system and/or the measurement (bad)
Describe the different kinds of variables
Binary, nominal, ordinal, interval, ratio
Categorical
Binary/Logical (frequency): two categories
Nominal (frequency): >2 categories
Ordinal (frequency + order): nominal but with logical order
(in R: all factors either logical or character )
Continuous
Interval (full arithmetic): infinite amount of meaningful values between two score (zero is valued, like thermometer)
Ratio (full arithmetic): Interval except zero means absence of score/value (like ruler)
(in R: numeric)
How does distribution that are skewed to left or right look like ?
Negatively skewed = LEFT
Positively skewed = RIGHT
What is kurtosis ?
How light- vs. heavy-tailed a distribution is
Leptokurtic (k > 0) or platykurtic (k < 0)
Bimodality
Distributions with 2 modes!
Often means two underlying distribution like height of women and men
What is The Central Limit Theorem ?
No matter the sample distribution (could be uniform, bimodal, fucked up in general), if you make a new distribution based on sample means “the sample distribution of sample means” this will approximate the normal distribution!
What does this distribution actually tell us?
What is the normal distribution ?
A perfect distribution has no skewness a kurtosis of 3 and a standard deviation of 1
This means we have a symmetrical gravitation toward the mean
A lot of cognitive and behaviour processes are normally distributed
How can you transform your data so it becomes normal?
log()
sqrt()
1/x
How do you calculate z-score?
And why is it useful?
z = (xi-x) / s : divide the deviance with the standard deviation
We use this to find the probability of a certain score within a normal distribution.
This also allow us to compare two distributions that uses different scales to begin with
How do you calculate z-score?
And why is it useful?
z = (xi-x) / s : divide the deviance with the standard deviation
We use this to find the probability of a certain score within a normal distribution.
This also allow us to compare two distributions that uses different scales to begin with
Population, sample, observation ?
Population: all subjects we are interested in
Sample: A subset of the population
Observation: A specific subject from the sample
What is parameters?
Numbers that tell us something about the population (e.g. u, m, etc. (greek letters)) whereas statistics tells us something about the sample (x, s, etc.)
