choosing a statistical test Flashcards
what is a statistical test
statistical testing is used to determine whether a difference or an association found in a particular investigation is statistically significant -did not occur by chance
The outcome of this has implications (influences) whether or not we accept or reject the null hypothesis
what are the three factors used to decide which statistical test to use
- whether a researcher is looking a DIFFERENCE or CORRELATION
- in the case of a difference, what EXPERIMENTAL DESIGN is being used
- the LEVEL OF MEASUREMENT
DIFFERENCE OR CORRELATION
the fist thing to consider when deciding a statistical test is to use relates to the AIM or PURPOSE of the investigation - specifically if the professor is looking for a DIFFERENCE or CORRELATION
This should be obvious from the wording of the hypotheses
e.g. correlation analyses as well as investigation that are looking for an association
what are the different experimental designs
- independent groups
- repeated measures
- matched pairs
what are the experimental designs that are related
repeated measures and matched pairs
what is the experimental designs that are unrelated
independent groups
when does the experimental design not apply to choosing the statistical test
if an investigation is looking fora correlation, rather than a difference, the experimental design does not matter
what are the different levels of measurement
Quantitative data can be divided into different levels of measurement and this is the third factor influencing the choice of statistical test
These are the levels of measurement:
- nominal data
- ordinal data
- interval data
what is nominal data
it is data that is represented in the form of categories hence nominal data is sometimes referred to as categorical data
( it shows difference in groups rather than in individuals)
e.g. you can count how many boys and girls in your year group.
The categories are male and female
nominal data is discrete in that one intention only appear in once of the categories e.g. if you asked people what their favourtie football team was, their vote only appears in one category
what is interval data
interval data is different to ordinal data
interval data is based on numerical scales that include units of equal, precisely defined size
e.g. like the measurements in maths or other sciences:
time,temperature, weight
why is interval data better than ordinal data
it is “better” than ordinal data because more detail is preserved ( and ordinal is “better” than nominal level)
this is because it is the most precise and sophisticated form of data in psychology and is a necessary criterion for the use of parametric tests
what is ordinal data
ordinal data is ordered in some way
e.g. asking someone in your class to rate how much they like psychology from a scale of 1 to 10 where “do not like psychology at all is a 1” and 10 is “absolutely love psychology”
its does not have equal intervals between unit (unlike interval data)
e.g. it would not make sense to say that someone who rated psychology an 8 enjoys it twice as much as someone who gave it a 4
what are the disadvantages of using ordinal data
it lacks precision because it is based on subjective opinion rather than objective measures
e.g. in the case of an IQ test the questions are derived from a view of what constituted intelligence rater than any universal measurement
For these reasons, ordinal data is sometimes referred to as “unsafe” data because it lacks precision
why is ordinal data not part of statistical testing
due to its unsafe nature, ordinal data is noticed as a part of statistical testing
Instead, raw scores are converted to ranks (i.e. 1st, 2nd, 3r) and it is the ranks not the scores that are used in the calculation
what are tests of differences - unrelated design
chi square - NOMINAL DATA
mann - whitney - ORDINAL DATA
unrelated t - test - INTERVAL DATA
