Ch 8 - Statistical Inference (Vogt, 2007)

Question 1

Statistical inference

Answer 1

make inferences from samples to populations (there are 2 ways: CI & hypothesis testing /stat significance)

Question 2

statistical significance

Answer 2

• the degree to which the data
  contradict the null hypothesis; e.g., .05, or 5%
• that is If a result is likely to be due to chance less than 5% of the time—the familiar p < .05 OR prob that you are wrong

Question 3

alpha level

Answer 3

cutoff point

Question 4

hypothesis testing

Answer 4

cutoff approach

Question 5

exact p-value

Answer 5

• e.g.,
  The p-value is .342, which is bigger than .05. This means that the difference between males
  and females on Exam 1 is not significant at the .05
  level; you could expect a difference of this size in a
  sample of this size 34.2% of the time.

Question 6

confidence intervals

Answer 6

–e.g. How confident am I that I am right? (prob that you are wrong is stat significance);
gives you all the information
the p-value gives you—plus more
- “we can be
95% confident that the true value in the population is
between . . .”—is not technically correct
- instead, one should prob state
“if we were to take an infinite number of random samples
of this size, in 95% of them the mean would be between.
. . .”
- But be forewarned, if you use the first, occasionally you will encounter fastidious scholars, advocates of the second, who will be disappointed in you

Question 7

confidence levels

Answer 7

?

Question 8

effect size (ES)

Answer 8

always report with stat significance;

Question 9

t-tests

Answer 9

• Would you get a mean difference this big if there were no difference in the populations
  from which this sample was drawn?
• how do samples differ
• measures stat sign
• are differences big enough that they are unlikely to be a coincidence?

Question 10

2-direction t-test

Answer 10

• e.g., tests the hypothesis that there is a difference between females and males, but it does not specify which one is bigger
• aka: non-directional t-test
• aka: 2-tailed
• better approach, more conservative
• harder test to pass
• mean difference has to be twice as big for a two-direction test to pass muster as statistically
  significant.

Question 11

statistical power

Answer 11

?

Question 12

false positive

Answer 12

?

Question 13

false negative

Answer 13

?

Question 14

standard error (SE)

Answer 14

• a kind of standard deviation;
  it provides an estimate of how much sampling error one is likely to get in a sample of a particular size
• Remember that the sampling error is the difference between the population value and the sample value
• so the SE answers the question: By how much is the sample value likely to miss the population value?
• The standard error gives you an estimate of the sampling error; it tells you how much error you
  are likely to have if you use the sample to estimate the
  population

Question 15

sampling distribution

Answer 15

?

Question 16

mean square (MS)

Answer 16

?

Question 17

variance

Answer 17

?

Question 18

MS w/in groups

Answer 18

?

Question 19

MS between groups

Answer 19

?

Question 20

F ratio/test

Answer 20

?

Question 21

ANOVA

Answer 21

measures stat sign;

Question 22

SPSS

Answer 22

software program that automatically computes
confidence intervals when computing t-tests
and ANOVAs, it gives everything necessary to compare
the p-value approach to the confidence-interval
approach.

Question 23

t-statistic

Answer 23

• a ratio of an estimate
  to its likely error
• you get the t-value by dividing the actual difference by the average probable error.
• If a difference is statistically significant, it will be bigger
  than the estimate of error (SE), usually at least two times bigger. - The t-statistic is just the sample statistic divided by its probable error