Statistical Testing

Question 1

what does it mean if data is related/unrelated?

Answer 1

• unrelated = independent groups
• related = repeated measures/matched pairs

Question 2

what process do you go through if you are looking for a correlation?

Answer 2

• is it interval/ ratio (use Pearson’s r test) or ordinal (use Spearman’s rho)

Question 3

what process do you go through if you are looking for a difference?

Answer 3

what is the level of measurement?
- nominal unrelated (chi square) nominal related (sign test)
- ordinal unrelated (Mann Whitney U) ordinal related (Wilcoxon)
interval/ratio - does it meet other criteria for parametric test?
IF YES THEN
- interval/ratio unrelated (t test - unrelated) interval/ratio related (t test - unrelated)
IF NO - GO TO ORDINAL DATA

Question 4

what is the criteria for a parametric test?

Answer 4

• populations drawn from should be normally distributed
• variances of populations and data should be approximately equal
• has at least interval or ratio data
• no extreme scores

Question 5

what is a one tailed test?

Answer 5

directional

Question 6

what is a two tailed test?

Answer 6

non-directional

Question 7

what is a type 1 error?

Answer 7

• researcher is overly optimistic about findings from research
• researcher wrongly rejected null hypothesis and accepts experimental/alternative hypothesis, assuming findings are significant

Question 8

what is a type 2 error?

Answer 8

• researcher has been too pessimistic about their findings from research
• assume experimental/alternative hypothesis is not significant so wrongly rejected experimental/alternative hypothesis and accepts null hypothesis

Question 9

what values will you be calculating for a chi square test?

Answer 9

df = degrees of freedom
test statistic = x²

Question 10

what values will you be calculating for a sign test?

Answer 10

N = total number of scores (ignoring 0 values)
test statistic = S

Question 11

what values will you be calculating for a Mann Whitney U test?

Answer 11

N1 = no of p’s/values in the smaller sample if 2 groups aren’t same size
N2 = no of p’s/values in the larger sample if 2 groups aren’t same size

Question 12

what values will you be calculating for a Wilcoxon test?

Answer 12

N = no of pairs of scores (excluding 0)
test statistic= T

Question 13

what values will you be calculating for a Spearman’s rho test?

Answer 13

N = number of pairs of scores being correlated
test statistic = rs

Question 14

what values will you be calculating for a Pearson’s r test?

Answer 14

df = degrees of freedom
test statistic = r

Question 15

what values will you be calculating for a t-test?

Answer 15

df = degrees of freedom
test statistic = t

Question 16

when do you use the sign test?

Answer 16

• difference in 2 data sets
• when you have related data (matched pairs/repeated measures)
• if data is nominal

Question 17

how do you do the sign test?

Answer 17

• identify whether hypothesis is directional or not
• record data and work out sign
• find calculated S value (add all plus signs, add all minus signs)
• smaller of the 2 is the S value

Question 18

what do you do after you have calculated your observed/calculated value of S?

Answer 18

• compare observed/calculated value of S to critical value of S in significance table
• identify if result is significant
• check results are in right direction if one tailed test was predicted
• decide whether to accept alternative hypothesis