Statistics Flashcards
what is a null hypothesis? what is an alternative hypothesis?
null: independent variable had no effect on dependent variable
alternative: independent variable had an effect on dependent variable
what is p value?
- probability value
- probability that the NULL hypothesis is supported
- low p-valie = indep prob has effect on dep
alpha = 0.05 –> critical value
what is degrees of freedom?
number of samples - 1
what is a continuous variable?
- quantitative value (any value between 0 and infinity)
- event count data (whole numbers)
ex. height, number of leaves
what is a categorical variable?
- limited number of fixed value
ex. assign 3 diff temperatures
if the dependent variable is continuous and the dependent variable is continuous, what statistical test would you perform?
correlation test
- pearson correlation coefficient, n value (sample size), and p -value
how would you present data in a correlation test?
- line/scatter plot
error bars possible around point on statter plot
``
what is the difference between regression and correlation?
- correlation: unsure if variable has an effect on other variable
- regression: expectation that one variable has an effect on the other
what is the multiple R value?
pearson correlation coefficient
if the dependent variable is continuous and the independent variable is categorical woth 2 treatments, what statistical test would you perform?
t-test
- t-test value, degrees of freedom, p value
ex wet vs dry, cold vs hot, two locations
how would you represent data if indep vairable is categorical with 2 treatments?
bar graph of mean values of each treatment with 2 columns and error bars
what are standard error vs deviation? difference?
- measurements of amount of variation in a sample
deviation: how spread out the data are in the dataset
error: estimate of how precise the mean estimate is
calculating standard error?
standard deviation / square root of number of samples
standard error < standard deviation
when to use standard deviation?
describing how variable the data in a dataset are
when to use standard error?
when quantifying the uncertainty around an estimate of the mean
standard error bars overlap = no difference between means
no overlap = treatment is significantly different for each other
what is t-test paired vs t-test two sample?
paired: same individuals under both conditions
two sample: different individuals for each treatment
use + value for t statistic
if the dependent variable is continuous and the independent variable is categorical with more than 2 treatments, what statistical test would you perform?
ANOVA
F-stat, first df, second df, p value
1st df: # groups/treatments -1
2nd df: > 1st df
how would you represent data if indep vairable is categorical with more than 2 treatments?
bar graph of means of each treatment with error bars (standard error)
what is a characteristic of ANOVA/having multiple categorical indep variables?
significant difference among treatments, but we don’t know which treatments have the same effect and which onces are different
to determine this we do a POST-HOC test = multiple t-tests
what does a post-hoc test do? how do you do it?
tells you which treatments are different from each other/have different effects on dep variable
do as many t-tests as the number of comparisons to equal post-hoc test
how do you analyze and report post hoc test data? how is the results represented on the graph?
analyze:
- compare each t test to corrected alpha/critical value; low p value = sig difference between treatments;
- if sig diff, assign diff letters; if no sig diff, assign same letter
- this is seen on graph bc no sig diff = error bars overlap
report:
- say if sig diff and say p value and say which group had higher value
what happens to alpha/critical value in post-hoc test? why do we do this?
alpha value/# comparisons(# tests done between each of the treatments)
more tests = increase likely hood that p-value is less than 0.o5 by chance
test ex. 4 treatments = A+B, A+C, A+D, B+C, B+B, C+D = 6 tests
what is included in fig caption?
indep, dep variables, SE, short interpretation of result (sometimes)
where does legend appear in relation to table?
above table