ward

Question 1

Significance level or alpha (α) value

Answer 1

Probability value that defines the boundary between rejecting or retaining H0

Question 2

What does it mean when probability is less than alpha?

Answer 2

H0 is rejected

Question 3

Region of rejection

Answer 3

Proportion of area in a sampling distribution that represents the sample means that r improbable if H0 is true

Question 4

Steps 4 finding region of rejection (directional/one-tailed)

Answer 4

1. Go to z-table
2. For p = 0.05, do 1 - 0.05 = 0.9495 (on z-table that corresponds to a z-score of 1.64)
3. This is the critical value, which is the score that u have to either reach or get above to reject H0
4. If observed z (z obs) is greater than or equal to 1.64 → reject H0

Question 5

One-tailed test description

Answer 5

1. Used because we have a directional alternative - used when there is evidence or theory to suggest that treatment will have an effect in one particular direction (decide before experiment)
2. For one-tailed test in other direction, z obs needs to be negative number beyond -1.64 to reject H0 (z obs smaller than or equal to -1.64)

Question 6

When is a two-tailed test used?

Answer 6

Used because we have a nondirectional alternative (e.g. gingko could improve or worsen memory)
Used 2 b more conservative - don’t use one-tailed test unless there’s a theoretical reason to

Question 7

What does carrying out a 2-tailed test entail?

Answer 7

1. Alpha is still 0.05, but now divided by 2 → 0.025 on either end
2. Could be an extremely high score or an extremely low score
3. If u get score on either end of spectrum, say that H1 is real (whether it improves memory or makes it worse)

Question 8

Critical value

Answer 8

Score that u have to either reach or get above to reject H0

Question 9

Characteristics of the population r known as parameters, which r.. ?

Answer 9

μ - mean, σ - standard deviation

Question 10

Characteristics of the sample r known as statistics, which r.. ?

Answer 10

x̄ - mean, s - standard deviation

Question 11

Sampling distributions def

Answer 11

Frequency distribution of sample means obtained from repeated sampling

Question 12

3 steps of sampling distributions + logic:

Answer 12

1. Make a guess abt the population frequency distribution - hypothesise what μ is
2. Take a random sample
3. Decide if sample came from a population like the the one u guessed in step 1 (usually based on how close x̄ is to the hypothesised μ)

Question 13

Central limit theorem (CLT) def

Answer 13

Regardless of if the sample u take or the population has a normal distribution, u can still get normal distribution if u take enough samples + plot the means

Question 14

CLT steps:

Answer 14

1. Assume normal population distribution with a μ and σ
2. Take repeated samples of size n
3. Plot the mean (x̄) of each sample
   Then u end up w/ a sampling distribution that is: normal, has population mean, has standard deviation equal 2 standard error of the pop. mean (σ/[√n])

Question 15

Standard error def

Answer 15

Numerical expression of the degree to which means differ from one sample to another

Question 16

What happens when standard error is large? What happens when it’s small?

Answer 16

1. If standard error is large, then lots of variability (harder to draw conclusions abt pop.)
2. If it’s small, sample mean likely quite close to population mean

Question 17

If u don’t know population standard deviation to calculate standard error, you.. ?

Answer 17

Use ‘s’ instead (standard deviation of a single sample)

Question 18

How to use z-score where CLT applies?

Answer 18

1. Treat x̄ as an individual score + see how it deviates from the population mean, relative to the expected amount of sampling error
2. Go to z-table + calculate amount of area beyond z (1 - prob. of z that we find in the z-table) → probability of finding a score that distant from the mean on the basis of chance alone

Question 19

Z-score formula (4 sampling distribution)

Answer 19

z = (x̄ - μ) / (√ σ^x̄)

Question 20

Single sample t-test formula:

Answer 20

(x̄ - μ) / estimated standard error

Question 21

What percentage of data do upper + lower limits encompass?

Answer 21

96%

Question 22

How would u calculate upper + lower limits 4 sampling distribution of means?

Answer 22

Upper limit: mean + 2SD
Lower limit: mean - 2SD

Question 23

Type I error description

Answer 23

1. We’re willing to reject H0 100% of the time when we have sample mean in region of rejection, despite 5% of the time the sample mean being from the sampling distribution of H0 due to sampling error
2. Therefore, we can only be 95% confident that we have correctly rejected H0 - 5% of time we’ll have made an error (type I error - rejecting H0 even tho it’s true)
3. Alpha level is chance of error ur willing to accept as a researcher, therefore: probability of type I error = alpha (e.g. 0.05)

Question 24

Type I error simplified def

Answer 24

Rejecting H0 when it’s true

Question 25

Type II error description

Answer 25

1. Even if H1 is true, some of the means from H1 could fall close to the μ of the H0 population (i.e. ‘other’ side of critical value), meaning u retain H0 even tho it’s false → type II error
2. Probability is beta (β) - think of this in terms of 2 overlapping distributions –> certain amount of overlap where u can make the wrong decision

Question 26

Type II error simple def

Answer 26

Retain H0 even tho it’s false

Question 27

What r some ways u can minimise type II error?

Answer 27

1. Reducing β + increasing power (1 – β) so that we can reject H0 when false –> can do this by increasing alpha (but this increases chance of type I error)
2. Increasing sample size - less variability –> narrower sampling distribution n less overlap, reduces β

Question 28

Power def

Answer 28

Ability of test to correctly reject H0

Question 29

Convention 4 choosing alpha level

Answer 29

Set very low (e.g. .01) when consequences of type I error r severe
Set higher (e.g. 0.05) when consequences of type I errors r not too serious

Question 30

True negative def

Answer 30

Don’t reject true H0

Question 31

False positive (type I error) def

Answer 31

Reject true H0

Question 32

True positive def

Answer 32

Reject false H0

Question 33

False negative (type II error) def

Answer 33

Retain false H0

Question 34

What happens to power when standard deviation of sample increases + curves become fatter n area of overlap increases?

Answer 34

Power decreases

Question 35

What is the use of a single sample t-test?

Answer 35

1. Gives u an idea of how far from average of population ur mean is + how meaningful that difference is
2. Procedure used to test the null hypothesis for a single-sample experiment when the standard deviation of the population must be estimated
3. Used when u have small sample instead of large one

Question 36

Accuracy of estimated SE for samples where n < 120 can be impaired, so u should.. ?

Answer 36

Use Student’s t-test

Question 37

What determines different t-distributions?

Answer 37

N - 1 (degrees of freedom/ how many scores in sample r free 2 vary), generally all scores except the last one

Question 38

Assumptions of single sample t-test

Answer 38

1. The random sample comprises interval or ratio scores
2. Distribution of individual scores is normal

Question 39

Steps 4 single sample t-test:

Answer 39

1. Calculate the observed t using the estimated standard error of the mean
2. Determine the df
3. Look up the critical t in the t-Table w/ appropriate df (and alpha)
4. If observed t is greater than or equal to the tabled value then reject H0

Question 40

Steps 4 two-sample t-test (r the 2 groups different from each other):

Answer 40

1. Compare difference between 2 sample means, calculate estimated SE of the difference between the sample means
2. Then use Student t-test + distribution ([1st sample mean - 2nd sample mean] / [estimated SE of the difference between means])
3. Estimate of the SE of the means is somewhat different depending on whether the 2 sets of scores r from the same or different groups - assume that any error in measurement will b less in a within-subjects group than between bc it’s the same people