Statistical Testing Flashcards
What is a null hypothesis?
A statement of no difference, correlation, or association between variables being studied.
What is an alternative hypothesis?
A testable statement about the relationship (difference or association) between variables.
What is sampling?
The process of obtaining a sample.
What is chance?
The extent to which something occurs randomly.
What is probability?
A numerical measure of the likelihood or chance that certain events will occur.
What 3 factors affect the choice of statistical test?
1.) Whether or not the study is looking for a difference or correlation. Obvious in wording of the aim/hypotheses, or the scenarios based question that you are presented with.
2.) What experimental design is being used.
- Independent groups.
- Repeated measures.
- Matched pairs.
3.) Level of measurement.
- Nominal.
- Ordinal.
- Interval.
Outline nominal data.
Represented in categorical form - referred to as categorical data.
E.g. male and female; number of monolingual, bilingual, and multilingual students in the class.
Discrete data -> item can only appear in one category.
Outline ordinal data.
Ordered in some way (e.g. rating scale) - unit not at equal intervals.
E.g. first, second, or third in a race.
Lacks precision as ordering is subjective.
Often measures things that aren’t ‘real’ (not observable, physical entities).
Non-standardised tests are ordinal level.
Unsafe data - the raw scores need converting into ranks to use in the test.
Outline interval data.
Based on numerical scales with equal units (precisely defined) - this preserves more about the data.
Uses public scales of measurement - e.g. a stopwatch, thermometer - producing data based on accepted units of measurement.
Most precise and sophisticated level of data.
Give an example of how nominal, ordinal, and interval data can be changed.
Person x, y, and z complete a running race.
Here are their times displayed as interval data:
- Person x: 28:43
- Person y: 19:32
- Person z: 17:16
How could this interval data be displayed as ordinal data?
- Person z, 1st place.
- Person y, 2nd place.
- Person x, 3rd place.
How could this ordinal data be displayed as nominal data?
- Person y and z both completed the race under 20mins.
- Person x took longer than 20 mins to complete the race.
What is the way to remember the stat test table?
Carrots Should Come Mashed With Spinach Under Roast Potatoes
Outline probability within inferential testing.
Inferential tests (try to infer from the sample data what the population might think) allow psychologists to draw conclusions from their findings.
These conclusions are based on the probability that a particular pattern of results could have arisen by chance or not.
Why is chance important to consider in psychology?
Psychologists cannot get away with guessing probabilities in psychological research.
In research you must be more precise.
In order to work out whether a difference is or is not significant we use inferential tests.
These tests permit you to work out, at a given probability, whether a pattern in the data could of arisen by chance or because there was a real difference/ correlation.
What does chance mean?
Chance means that the psychologist decides on the probability that will be ‘risked’.
You cannot be 100% certain that an observed effect was not due to chance but we can state how certain we are.
What probabilities do psychologists normally use?
Generally, psychologists use a probability of 95%.
This expresses the degree of uncertainty.
It means that there is a 5% chance of the results occurring if the null hypothesis is true.
A probability of 5% is recorded at….
A probability of 5% is recorded at p ≤ 0.05. (normally used as is more certain).
A probability of 10% is recorded at….
A probability of 10% is recorded at p ≤ 0.1. (normally not used as too risky).
A probability of 1% is recorded at….
A probability of 1% is recorded at p ≤ 0.01.
When are more stringent probabilities chosen?
More stringent probabilities (1%) are chosen when working with human biological data, e.g. blood tests.
What is significance?
A statistical term that tells us whether a difference or correlation exists.
The level of significance in psychology is 0.05 (5%).
When you have significant results, you reject the null hypothesis and accept the (alternative) hypothesis.
Level of significance should always be what?
0.05.
What are critical value tables?
The numerical boundary between acceptance and rejection of the null hypothesis.
Checks significance: compare with the calculate value.
Tells us whether to accept or reject the null hypothesis.
Each statistical test has its own critical value table.
How should critical value tables be answered?
Make reference to the obtained value and the critical value.
Reference the significance level and whether the hypothesis is one-tailed or two-tailed.
Reference the number of participants.
Is it significant or not?
Example:
The obtained value of +0.42 exceeds the critical value for a two-tailed hypothesis (0.362) for N = 30. The results are therefore statistically significant (p less than or equal to 0.05).
Outline type 1 errors?
False positives.
Incorrect rejection of the null hypothesis - (need to accept instead).
It is an optimistic error - claiming to have found significant results when they are not.
Most likely if the significance level is too lenient - e.g. 0.10, which is why 0.05 is accepted in most cases.
For example, finding that a man is pregnant (the null stating that he cannot be pregnant would need to be instead accepted).
