Research Methods - Inferential Statistics

Question 1

inferential statistics

Answer 1

moving beyond describing data to testing hypotheses

Question 2

probability

Answer 2

the likelihood of an event occurring
helps us to determine if results are statistically significant
p ≤ 0.05 - most used but less accurate (more likely to accept alt hypothesis)
p ≤ 0.01 - mostly used in drug tests E.G. depression meds (more likely to accept null)

Question 3

parametric tests

Answer 3

make assumptions about the data’s underlying distribution
assumptions:
data is normally distributed
variances are equal across groups
interval/ratio data

Question 4

strengths of parametric tests

Answer 4

better than non-parametric tests when assumptions are met
allows for more detailed inferences

Question 5

limitations of parametric tests

Answer 5

not suitable when data doesn’t meet assumptions

Question 6

non-parametric tests

Answer 6

makes fewer assumptions about the data’s underlying distribution
assumptions:
few or no assumptions about distribution
can be used for ordinal/ non-normally distributed data

Question 7

strengths of non-parametric tests

Answer 7

more flexible
good for small sample sizes and ordinal data

Question 8

limitation of non-parametric tests

Answer 8

less powerful than parametric
less detailed

Question 9

statistical tests

Answer 9

DIFFERENCE
Independent measures
NOMINAL Chi squared (X2)
ORDINAL Mann Whitney U (U)
INTERVAL/RATIO Unrelated T-test (R)

DIFFERENCE
Repeated measures
NOMINAL Sign test (S)
ORDINAL Wilcoxon test (T)
INTERVAL/RATIO Related T-test (R)

CORRELATION
NOMINAL Chi squared (X2)
ORDINAL Spearman’s Rho (Rs)
INTERVAL/RATIO Pearson’s (R)

Question 10

how to remember the statistical tests

Answer 10

Carrots Should Come Mashed With Swede Under Roast Potatoes

Question 11

calculated value

Answer 11

the value you figure out yourself

Question 12

critical value

Answer 12

the value you look up in a table particular to the test you are using

Question 13

how to find out if the test is statistically significant (null or alt)

Answer 13

the calc and crit values are compared
for some tests calc needs to be equal/less than crit
for others greater than

Question 14

how to remember how to find out if the test is statistically significant

Answer 14

if there is an R in the name of the test it must be gReateR than the crit value to be significant

E.G. chi squaRed

Question 15

how to write significance statements

Answer 15

1 state whether the calc is greater/less than the crit (put values in brackets and put calc symbol B4 number E.G. S(8))
2 state if results are significant or not
3 state if null and alt are accepted/rejected (ALWAYS state null 1st)
4 write out relevant hypothesis (accepted one)
5 report figures in brackets
a: calc b: No of P’s in analysis (N=/N1=) c: whether p was more/les than 0.05 (alt= p < 0.05, null= p > 0.05) d: write whether hypothesis was 1 tailed or 2 tailed

E.G.
S=(3)>0… not significant…null accepted… alt rejected… null hypothesis… (S=3, N1=8, P>0.05, 2 tailed)

Question 16

type 1 error (false pos)

Answer 16

null is mistakenly rejected

Question 17

type 2 error (false neg)

Answer 17

null is wrongly accepted

Question 18

how to remember type1/2 errors

Answer 18

N= neg
P= pos
P has 1 line = type 1 error
N has 2 lines = type 2 error

Question 19

how to do a sign test

Answer 19

1 assign a sign of difference E.G. +/-
2 count total of signs
3 S (calc) = least occuring sign
4 work out crit E.G. p=0.05, n=9
5 compare calc and crit E.G. calc<crit
6 conclude