The SPINE of Stats

Question 1

What does the SPINE of stats refer too?

Answer 1

Standard error
Parameters
Interval estimates
Null hypothesis testing
Estimation

Question 2

What does estimation refer too?

Answer 2

Sampling error
sampling variation - when you take a sample, you get an estimation, if take another, get a different estimation - always slightly different answers
sampling distribution - if you take lots of samples, there will be distributions among different estimations

Question 3

What is the standard error?

Answer 3

The width of the sampling distribution
Big = wide distribution, 2 samples could produce different results
Small = not much variation, samples close to the population
Refers to variability across samples - how a parameter differs from sample to sample

Question 4

What happens if a sample is big enough?

Answer 4

The sampling distribution will be normally distributed

Question 5

Where do 95% of the scores lie?

Answer 5

Between +/- 1.96 SD

Question 6

How do you create a confidence interval?

Answer 6

Take the mean, add/minus 1.96 times standard error

use the SE to create them

Question 7

How do you know if parameters come from the same population?

Answer 7

If they overlap = same population

If they don’t = different

Question 8

What are confidence intervals?

Answer 8

Intervals which contain the true value of the parameter in 95% of samples - don’t know if it is one of the 95%

Question 9

What does a narrow confidence interval mean?

Answer 9

All samples would get estimates of B close to the population value

Question 10

What does a wide confidence interval mean?

Answer 10

Lots of uncertainty about the estimation of B

Question 11

What does it mean if confidence intervals straddle 0?

Answer 11

Population value of B could be 0

Don’t know which direction the relationship goes, could be positive or negative

Question 12

What are the steps of null hypothesis significance testing?

Answer 12

Generate a hypothesis - null or alternate
Specify a P value
Pick sampling distribution and choose sample size
Sample, compete the statistic
Compare long run probability p 
Compare p to a
if less than a - reject null
if more than a - accept null

Question 13

What are parameters doing?

Answer 13

Testing hypotheses for us - work out probability of getting value we have if null was true

Question 14

What is null hypothesis significance testing related too?

Answer 14

Sample size
the bigger the same size - the more likely to show a significant effect, has the power to detect small differences
small sample size - big effects won’t be significant

need to interpret p in terms of sample size

Question 15

What are the ways of testing effect sizes?

Answer 15

Parameters
Standardised B
Pearsons correlation = .1 is small .5 is big
Cohens D

Question 16

How to calculate cohens D

Answer 16

Difference between two means divided by some estimate of the standard deviation

Question 17

What does cohens d show?

Answer 17

D = 1 - means are 1 SD apart
D = 0 - means are same
0.2 = small effect
0.8 = large effect

Question 18

Which standard deviation do you use when calculating cohens d?

Answer 18

Control group
Either group (if assume HOV)
Pooled estimate - weighted average by sample size

Question 19

How do you calculate pooled estimate?

Answer 19

(N1 - 1) SD1 squared + (N2 - 1) SD2 squared

divided by N1 + N2 - 2

Question 20

What do P values show in comparison to D values?

Answer 20

P show inconsisties, making it look like there are lots of differences
D show consistencies