Stats - 2 Flashcards
What is the definition of standard error of mean? (SEM)
SD/ root of number of samples
How do SD and the number of samples influence SEM?
SD increases, so does SEM
N increases, SEM decreases
How do you use SEM to estimate the confidence interval?
confidence interval 95% = mean +/- 2SEM
What does ANOVA stand for?
analysis of variance
When is ANOVA used?
used for normal distribution, comparison of multiple groups with different sample sizes, numerical data
If p<0.05 is found from ANOVA, does that mean all results are significantly different from each other?
not sure - need to do a post-hoc test
When estimating the mean in a range of 95% confidence interval, what are the upper and lower boundaries of the mean?
+/- 2SEM
What is a post-hoc test?
when ANOVA = p<0.05 - all pairs are checked one by one to find which pairs are significantly different and which aren’t
If data is ordinal, what main methods are used to test the hypothesis?
Chi-square
In Chi-square analysis, what are the theoretically expected values used to compare with practical data?
Equal percentage or equal sample size from all groups are theoretically same value. Practical value is the value you get from the data
In a chi-square, assuming that multiple groups of data are significantly different, does this mean any two of them are significantly different?
No
What statistical methods are used with ordinal or nominal data?
Chi-square
non-parametric test
In what situations should a non-parametric test be applied with two or more groups of data?
non-normal distribution
use sign or rank as info to compare the groups
What are the main differences between parametric and non-parametric tests?
non-parametric - non normal distribution
parametric - normal distribution
In non-parametric data, what information or values are used to assess differences between groups of data?
rank
sign
When a scatter plot show that two variables have a certain association can we say they are linearly correlated?
no - need to calculate the linear correlation and p value
If two variables are linearly correlated, how much significance interval would be expected after a hypothesis test?
p <0.05
What range should the correlation coefficient be kept in?
-1
IS it possible that the correlation coefficient is negative? what would this mean?
Yes
that an increase in X means a decrease in Y
what coefficients are calculated/estimated to produce a regression equation?
b1 and b2
(y = mx+c)
(Y = b2x + b1)
What is a residual in linear regression?
the errors between the predicted value and the measured value
What does a high or low residual mean?
high - poor quality. the lower value the better
Can linear regression methods be extended to non-linear situations? how do we process the original data in such a situation?
Yes - transform non-linear variables into linear variables then use these to construct a linear relation
Why does sampling require randomisation?
If not then the samples will not be representative of the bigger population
What is the idea behind meta-analysis?
using multi-sourced data to see if it favours one direction - studies would be in the same format and be of high quality
What is censored case? what main point shoud be noted when doing this?
survival analysis - requires larger sample size
What is statistics?
using a small sample size to find information about a bigger population.