Statistical Testing

Question 1

Why do we use a sample to answer a q about the pop

Answer 1

Comparing performance between groups (male v female

Has average performance improved

Statistical tests provide evidence for a particular belief, given as a probability or p value

They are not conclusive

The logic underpinning statistical testing is intuitive - we do it anyway

Question 2

Four steps of testing (informal)

Answer 2

1) state a hypothesis
2) draw a sample and calculate something from it (test statistic)
3) assuming the hypothesis is true, consider the probability of obtaining that sample result
4) if the probability is low reject the hypothesis, otherwise retain it

Question 3

More detailed four point structure

Answer 3

1)Identify the null hypothesis H0

Mean weight is 50g(pop mean)

2)Calculate test statistic from the sample
Sample mean of 49.5, from this calculate z value

3)Calculate the probability of getting such an extreme test statistic, assuming H0 is true
Pop= 50 , prob that sample mean is 49.5 is 0.46

4) if probability is small reject null hypothesis
0. 46 is quite likely so retain

Question 4

Probability calc

Answer 4

H0= 50g
Known variance= 100
Sample size = 5

    _
P(x. <49.5) when u=50    _    x ~ N(50,100/5)

49.5 is (49.5-50)/square root of (100/5) = 0.11 SD

Z value = -0.11

Using normal prob p(Z

Question 5

If null hypothesis is rejected

Answer 5

Add an alternative hypothesis
H1

Mean Is more than 50
Mean is less than 50
Or mean is not equal to 50

Question 6

P values

Answer 6

The probability of getting a sample result as extreme or more extreme than the one observed, when the null hypothesis is true is called the p value

P value < 0.05 test is significant, reject H0 adopt H1

P value > 0.05 test is not significant, retain H0

Question 7

SPSS

Answer 7

P values are called sig (2 tailed )
Gives results for two sided test if you want one divide it by 2

Analyse > compare means > one sample T test