Choosing a statistical test

Question 1

what is a statistical test used for?

Answer 1

to determine whether a difference or an association/ correlation found in a particular investigation is statistically significant - more than could have occurred by chance

Question 2

what are the three factors used to determine what statistical test to use?

Answer 2

whether the researcher is looking for a difference or correlation

what experimental design

the level of measurement

Question 3

difference or correlation?

Answer 3

the purpose of the investigation - should be clear from the hypothesis - correlation include investigations looking for an ‘association’

Question 4

experimental design?

Answer 4

three types of experimental design: independent groups, repeated measures, matched pairs.

repeated measures and matched pairs are related designs

RM- same participant in both condition

MP- participants in each condition are not the same but have been ‘matched’ on a variable that is importnant for the investigation

because participants are different in independent groups - it is unrelated

Question 5

what are the different levels of measurement that quantitative data can be divided into?

Answer 5

nominal, ordinal and interval

Question 6

what is nominal data?

Answer 6

data represented in the form of categories (categorical data)

e.g. people who say yes and people who say no are put into two categories

nominal data is discrete - only one item can appear in one of the categories

Question 7

what is ordinal data?

Answer 7

data which is ordered in some way

e.g. asking people to rate something on a scale of 1-10

• it does not have equal intervals between each unit
• it lacks precision as it is based on subjective opinion rather than objective methods

ordinal data is often referred to as ‘unsafe’ data as it lacks precision - due to this it is not used as part of statistical testing

instead, ordinal data is converted to ranks - which are used in the calculation

Question 8

what is interval data?

Answer 8

based on numerical scales - which include equal units which are precisely defined in size

• the most precise and sophisticated form of data collection

it is a necessary criterion for the use of parametric tests.