Analysis of trends Flashcards
what is analysis of trends
seeing if there is a correlation - the effect of one variable on another
what is Pearsons correlation efficient
+1, 0, -1 = r - no units, numerical, range of values between -1 and 1
what are the three things a correlation can be
positive linear, negative linear or 0
what is a bell curve correlation
0 - there is a clear relationship, but its not linear so = 0
what does the Pearson calculation give you
r - the correlation coefficient - has to be between -1 and 1
what do you do to work out if r is significant
r squared - so its always positive
r^2/standard error = t
when is the Pearson calculation not significant
when the t value is below r = conclude there isn’t a significant relationship
what does the regression line do
minimises the sum of the squared deviation of the points from the line
what are the two quantities of the regression line
intercept - c
slope - b
what does the regression line assume - 3 things
random sampling, linear correlation, residuals normally distributed with a constant varience
what are the residuals on a graph
the points scattered around the regression line
if the three assumptions for the regression line are not met what can you do
transform to log - straightens the relationship between the variables - makes residuals normal
what is a ‘linear model’ - common in ecology
where the computer finds the best model for your data - estimates the effect of one variable on another
what is ‘overfitting’
the model can only get better by adding another variable - the more complex model the more fitting opportunity
what is the ‘minimum adequate model’
when you reduce the number of explanatory variables
