Lecture 12_statistical errors and effect size

Lecture 12_statistical errors and effect size - 10/6/2010...

Info iconThis preview shows pages 1–4. Sign up to view the full content.

View Full Document Right Arrow Icon
10/6/2010 1 1 Review • Hypothesis testing procedures provide us with an objective means to evaluate our research findings • In every case, we wanted to know if the observed relationships are greater than what would be expected by sampling fluctuation, or by chance alone Fallibility of Significance Tests • Despite our best efforts, statistically significant results can still occur simply due to sampling fluctuation • Statistical significance does not guarantee that the conclusions drawn from a study are correct
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
10/6/2010 2 Fallibility of Significance Tests • Statistical significance only guarantees that there is a low probability of obtaining our results due to chance • With any study, we can never be completely certain that our final conclusions are correct • We can never rule out chance, only minimize it Recall… • Using the central limit theorem, we constructed a sampling distribution of the population associated with the null hypothesis null Recall… • By convention, we decided that if our sample fell in the most extreme 5% of the curve ( =.05), we would reject the null
Background image of page 2
10/6/2010 3 Chance Results • It is possible for a sample to fall in the critical region simply due to chance alone • Suppose we did not actually conduct a study, but instead just took 100 random samples from the null population • Five of those samples (5%) would fall in the critical region--- and be “significant”--- even if we did nothing at all Alpha Level and Chance • Our alpha level defines the degree to which we are willing to tolerate chance results in our experiment/research study – When = .10, we will reject H 0 10% of the time simply due to chance alone – When = .01, we will reject H 0 only 1% of the time simply due to chance alone • More stringent (i.e., smaller) levels mean less tolerance of chance results
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 4
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 02/17/2011 for the course PYSC 227 taught by Professor Fairchild during the Spring '10 term at South Carolina.

Page1 / 10

Lecture 12_statistical errors and effect size - 10/6/2010...

This preview shows document pages 1 - 4. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online