Null Hypothesis And Significance Testing Flashcards
When we have one sample, how do we work out the standard error of a population?
There are two parts, the first is to calculate the estimated population SD then this value is divided by the square root of the number of participants.
Is standard error effected by the scale?
Yes, because if the results used a small scale of 1-5 and the standard error was 4.1 then this shows a lot of deviation compared to the same error on a scale of 1-100
What is a hypothesis?
A statement of what we expect to find in our research
What is a null hypothesis?
Referred to as H0. This is an assumption no effect will be found.
In research we look at how likely it would be to find those results if the null hypothesis was true.
What is an alternative hypothesis?
Referred to as H1, H2, H3…
States there will be a difference between the groups and may state the direction.
Should be a testable statement
What are one-tailed and two-tailed hypothesis?
One-tailed hypothesis is directional and two-tailed is not.
Which hypothesis does inferential statistics look at?
The null hypothesis.
What are the two key errors in accepting or rejecting the null?
Type 1: rejecting the null when we should accept
Type 2: accepting the null when we should reject
What is meant by statistical significance?
A significant difference in results. The p value tells us the probability that our sample is part of the null hypothesis population.
Referred to as p value.
It is a measure of a probability.
The probability we would get these results by chance.
Helps us decide whether to reject/accept null hypothesis.
What can the significance tell us?
If we have a significance of .99 then it is very likely this result was arrived at by chance and that the null hypothesis is true.
What are the cut off points for significance?
What happens with statistical significance and one/two-tailed tests?
Two-tailed tests have no direction so they are looking at whether the mean is greater or less than the mean of the original population. We look at the extreme 2.5% at each end.
One-tailed tests have a direction so we look for the difference in the specified direction. We look at the extreme 5% at the specified end.
What are the issues with significance with one-tailed tests?
More likely to achieve significance but if it doesn’t show significance we could be ignoring the fact that there might be a negative effect.
E.g. Tablet to help drug addiction should help drug addicts. Results show no significance so therefore we believe they have no effect but it neglects that there might be a negative effect.
When estimating the standard error of a population what do we do?
We have to find the means of each sample then find the standard deviations of these means but this is time consuming as we usually only have one sample and don’t have time to get more.
Are there different significance tables?
Yes, different tables for different tests.
