Ch13: Inferential Stats

Statistics

relationship or value in a sample

Parameters

corresponding values in the population

Sampling Error

random variability in a statistic from sample to sample

Sample statistic

population parameter+ sampling error

What are the goals of psychological research?

it usually measures 1+ variables for a sample, but trying to draw conclusions about the whole population

NHST

formal approach to deciding between 2 interpretations of a statistical relationship in a sample

Null hypothesis

H0 (H naught); no relationship exists in the population, any relationship found in samples is sampling error

Alternative hypothesis

H1; relationship exists in the population

How likely are we to find a relationship as large or larger than the one observed in our sample if null is true?

Unlikely= reject null hypothesis in favour of alternative. Likely (or not extremely unlikely)= fail to reject the null

p value

probability or likelihood of a result as large or larger than the sample result if the null hypothesis were true. Ranges from 0 to 1, can never exactly 0

What does a correlation coefficient of .05, p value of 0.20 mean?

If the population correlation value is 0 (no relationship), and if we conducted this study infinitely many times, we would expect to get a correlation coefficient of 0.05 or larger 20% of the time

p> 0.05

usual threshold for significance. Doesn’t necessarily mean the null is false, but not enough evidence to conclude that its true. We can fail to reject the null, but we can’t retain or accept the null

What are some misconceptions about p values?

1. p value is the likelihood that your results were due to chance 2. p value of 0.05 means that null has 5% chance of being true 3. p value of >.05 means we can accept the null hypothesis 4. p value of >.05 means there is no difference between groups or no relationship between variables 6. p value >.05 means there is a >95% chance of replicating the result

Why is this a misconception of p values (p value is the likelihood that your results were due to chance)?

p values tell us about likelihood of a result assuming the null is true. We don’t know whether the null hypothesis is true (this is why we do studies) so we can’t know the likelihood that our results were due to chance

Why is this a misconception of p values (p value of 0.05 means that null has 5% chance of being true)?

Entire framework for interpreting p values is assuming null is true

Why is this a misconception of p values (p value of >.05 means we can accept the null hypothesis)?

Under NHST we can never accept the null, only fail to reject it. NHST framework starts by assuming the mull is true, so we can’t use this logic to accept or probe the null is true

Why is this a misconception of p values (p value of >.05 means there is no difference between groups or no relationship between variables)?

You will almost never get a sample statistics that is exactly equal to 0. Goal= say whether the difference is large enough to reject the null

Why is this a misconception of p values (p value >.05 means there is a >95% chance of replicating the result)?

p value in study A doesn’t tell us what to expect in study B