Topic 5 Flashcards
How is a standardised variable calculated?
Give the formula for estimated population variance
Show the formula for a beta1/beta2 hat confidense internal
Show the confidense interval for the variation estimate
What are confidense intervals?
Random intervals, which once calculated can’t be said to have a probability of containing their parameter, as they either do or don’t. The formula however, has 1-alpha chance of containing the parameter.
How is a two tail test conducted?
Null: beta-x = a
Alt: beta-x = b
Use t-alpha/2
Show an ANOVA table for CNLRM
What influences our prediction of Yi?
- Distance from mean X increases variation
- large n decreases variation
- low sample variance decreases variation
How is a one tail test conducted?
Null: beta-x < a
Alt: beta-x > a
Using t-(1 - alpha)
How is the t & F test related?
t^2 = F,
where the null is beta2 = 0 and alt beta2 != 0
How can we test to confirm normality
- Histogram of residuals
- Normal probability plot
- Jarque - Bern test of normality
How do we examine the normality assumption using a histogram?
We make a histogram of eresiduals (ie, residuals in frequency bins) and examine it to see if it looks normal). Can’t say much at all with low sampl numbers.
How do we use normal probability plots to cooberate the normality assumption?
Use special graph paper to graph the error terms on the x-axis and on the y-axis, the expected value if they were normal is shown. Hence, a roughly straight line indicates the assumption is pretty dec. Ps. Excel is a cheating liar
How do we use the Jarque - Bern test of normality?
- Only with large samples
- Measures skewness and kurtosis (concentration around mean)
- Compares these values with the normal distibution (S = 1 K=2) *maybe* **totally bad memory**
- Gives a test statistic that is totally chi square
